search

Mozilla-Ocho / llamafile

Distribute and run LLMs with a single file.
https://llamafile.ai
Other
19.88k stars 999 forks source link

Bug: Unable to load Mixtral-8x7B-Instruct-v0.1-GGUF on Amazon Linux with AMD EPYC 7R13 #512

Open rpchastain opened 2 months ago

rpchastain commented 2 months ago

Contact Details

rpchastain@proton.me

What happened?

I'm attempting to use mixtral-8x7b-instruct-v0.1.Q5_K_M.gguf weights on an AWS ec2 instance with an AMD EPYC 7R13 and 4 NVidia L4 gpus. Llamafile fails to load with the attached log output.

Version

llamafile v0.8.12

What operating system are you seeing the problem on?

No response

Relevant log output

./llamafile -m mixtral-large.gguf --server --nobrowser -ngl 999
import_cuda_impl: initializing gpu module...
get_rocm_bin_path: note: hipcc not found on $PATH
get_rocm_bin_path: note: $HIP_PATH/bin/hipcc does not exist
get_rocm_bin_path: note: /opt/rocm/bin/hipcc does not exist
link_cuda_dso: note: dynamically linking /home/ec2-user/.llamafile/v/0.8.12/ggml-rocm.so
link_cuda_dso: warning: libamdhip64.so.6: cannot open shared object file: No such file or directory: failed to load library
link_cuda_dso: note: dynamically linking /home/ec2-user/.llamafile/v/0.8.12/ggml-cuda.so
ggml_cuda_link: welcome to CUDA SDK with cuBLAS
link_cuda_dso: GPU support loaded
{"build":1500,"commit":"a30b324","function":"server_cli","level":"INFO","line":2873,"msg":"build info","tid":"10497024","timestamp":1722175351}
{"function":"server_cli","level":"INFO","line":2876,"msg":"system info","n_threads":24,"n_threads_batch":-1,"system_info":"AVX = 1 | AVX_VNNI = 0 | AVX2 = 1 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | AVX512_BF16 = 0 | FMA = 1 | NEON = 0 | ARM_FMA = 0 | F16C = 1 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 0 | SSE3 = 1 | SSSE3 = 1 | VSX = 0 | MATMUL_INT8 = 0 | LLAMAFILE = 1 | ","tid":"10497024","timestamp":1722175351,"total_threads":48}
llama_model_loader: loaded meta data with 26 key-value pairs and 995 tensors from mixtral-large.gguf (version GGUF V3 (latest))
llama_model_loader: Dumping metadata keys/values. Note: KV overrides do not apply in this output.
llama_model_loader: - kv   0:                       general.architecture str              = llama
llama_model_loader: - kv   1:                               general.name str              = mistralai_mixtral-8x7b-instruct-v0.1
llama_model_loader: - kv   2:                       llama.context_length u32              = 32768
llama_model_loader: - kv   3:                     llama.embedding_length u32              = 4096
llama_model_loader: - kv   4:                          llama.block_count u32              = 32
llama_model_loader: - kv   5:                  llama.feed_forward_length u32              = 14336
llama_model_loader: - kv   6:                 llama.rope.dimension_count u32              = 128
llama_model_loader: - kv   7:                 llama.attention.head_count u32              = 32
llama_model_loader: - kv   8:              llama.attention.head_count_kv u32              = 8
llama_model_loader: - kv   9:                         llama.expert_count u32              = 8
llama_model_loader: - kv  10:                    llama.expert_used_count u32              = 2
llama_model_loader: - kv  11:     llama.attention.layer_norm_rms_epsilon f32              = 0.000010
llama_model_loader: - kv  12:                       llama.rope.freq_base f32              = 1000000.000000
llama_model_loader: - kv  13:                          general.file_type u32              = 17
llama_model_loader: - kv  14:                       tokenizer.ggml.model str              = llama
llama_model_loader: - kv  15:                      tokenizer.ggml.tokens arr[str,32000]   = ["<unk>", "<s>", "</s>", "<0x00>", "<...
llama_model_loader: - kv  16:                      tokenizer.ggml.scores arr[f32,32000]   = [0.000000, 0.000000, 0.000000, 0.0000...
llama_model_loader: - kv  17:                  tokenizer.ggml.token_type arr[i32,32000]   = [2, 3, 3, 6, 6, 6, 6, 6, 6, 6, 6, 6, ...
llama_model_loader: - kv  18:                tokenizer.ggml.bos_token_id u32              = 1
llama_model_loader: - kv  19:                tokenizer.ggml.eos_token_id u32              = 2
llama_model_loader: - kv  20:            tokenizer.ggml.unknown_token_id u32              = 0
llama_model_loader: - kv  21:            tokenizer.ggml.padding_token_id u32              = 0
llama_model_loader: - kv  22:               tokenizer.ggml.add_bos_token bool             = true
llama_model_loader: - kv  23:               tokenizer.ggml.add_eos_token bool             = false
llama_model_loader: - kv  24:                    tokenizer.chat_template str              = {{ bos_token }}{% for message in mess...
llama_model_loader: - kv  25:               general.quantization_version u32              = 2
llama_model_loader: - type  f32:   65 tensors
llama_model_loader: - type  f16:   32 tensors
llama_model_loader: - type q8_0:   64 tensors
llama_model_loader: - type q5_K:  833 tensors
llama_model_loader: - type q6_K:    1 tensors
llm_load_vocab: special tokens definition check successful ( 259/32000 ).
llm_load_print_meta: format           = GGUF V3 (latest)
llm_load_print_meta: arch             = llama
llm_load_print_meta: vocab type       = SPM
llm_load_print_meta: n_vocab          = 32000
llm_load_print_meta: n_merges         = 0
llm_load_print_meta: n_ctx_train      = 32768
llm_load_print_meta: n_embd           = 4096
llm_load_print_meta: n_head           = 32
llm_load_print_meta: n_head_kv        = 8
llm_load_print_meta: n_layer          = 32
llm_load_print_meta: n_rot            = 128
llm_load_print_meta: n_swa            = 0
llm_load_print_meta: n_embd_head_k    = 128
llm_load_print_meta: n_embd_head_v    = 128
llm_load_print_meta: n_gqa            = 4
llm_load_print_meta: n_embd_k_gqa     = 1024
llm_load_print_meta: n_embd_v_gqa     = 1024
llm_load_print_meta: f_norm_eps       = 0.0e+00
llm_load_print_meta: f_norm_rms_eps   = 1.0e-05
llm_load_print_meta: f_clamp_kqv      = 0.0e+00
llm_load_print_meta: f_max_alibi_bias = 0.0e+00
llm_load_print_meta: f_logit_scale    = 0.0e+00
llm_load_print_meta: n_ff             = 14336
llm_load_print_meta: n_expert         = 8
llm_load_print_meta: n_expert_used    = 2
llm_load_print_meta: causal attn      = 1
llm_load_print_meta: pooling type     = 0
llm_load_print_meta: rope type        = 0
llm_load_print_meta: rope scaling     = linear
llm_load_print_meta: freq_base_train  = 1000000.0
llm_load_print_meta: freq_scale_train = 1
llm_load_print_meta: n_yarn_orig_ctx  = 32768
llm_load_print_meta: rope_finetuned   = unknown
llm_load_print_meta: ssm_d_conv       = 0
llm_load_print_meta: ssm_d_inner      = 0
llm_load_print_meta: ssm_d_state      = 0
llm_load_print_meta: ssm_dt_rank      = 0
llm_load_print_meta: model type       = 8x7B
llm_load_print_meta: model ftype      = Q5_K - Medium
llm_load_print_meta: model params     = 46.70 B
llm_load_print_meta: model size       = 30.02 GiB (5.52 BPW) 
llm_load_print_meta: general.name     = mistralai_mixtral-8x7b-instruct-v0.1
llm_load_print_meta: BOS token        = 1 '<s>'
llm_load_print_meta: EOS token        = 2 '</s>'
llm_load_print_meta: UNK token        = 0 '<unk>'
llm_load_print_meta: PAD token        = 0 '<unk>'
llm_load_print_meta: LF token         = 13 '<0x0A>'
ggml_cuda_init: GGML_CUDA_FORCE_MMQ:   no
ggml_cuda_init: CUDA_USE_TENSOR_CORES: yes
ggml_cuda_init: found 4 CUDA devices:
  Device 0: NVIDIA L4, compute capability 8.9, VMM: yes
  Device 1: NVIDIA L4, compute capability 8.9, VMM: yes
  Device 2: NVIDIA L4, compute capability 8.9, VMM: yes
  Device 3: NVIDIA L4, compute capability 8.9, VMM: yes
llm_load_tensors: ggml ctx size =    2.42 MiB
llama_model_load: error loading model: check_tensor_dims: tensor 'blk.0.ffn_gate_exps.weight' not found
llama_load_model_from_file: failed to load model
llama_init_from_gpt_params: error: failed to load model 'mixtral-large.gguf'
{"function":"load_model","level":"ERR","line":450,"model":"mixtral-large.gguf","msg":"unable to load model","tid":"10497024","timestamp":1722175352}
rpchastain commented 2 months ago

Here's the output of strace:

SYS  13477  13477        775'996'293 pread(3, [u"_restrict__ vx, int * __restrict__ x_ql,"...], 512, 284'160) → 512 ENOENT
SYS  13477  13477        775'998'078 pread(4, [u"ock_q8_0 * bx0 = (const block_q8_0 *) vx"...], 512, 284'672) → 512 ENOENT
SYS  13477  13477        776'000'259 pread(3, [u"ock_q8_0 * bx0 = (const block_q8_0 *) vx"...], 512, 284'672) → 512 ENOENT
SYS  13477  13477        776'005'259 pread(4, [u"nwarps * QI8_0) {◙        int i = i0 + i"...], 512, 285'184) → 512 ENOENT
SYS  13477  13477        776'007'671 pread(3, [u"nwarps * QI8_0) {◙        int i = i0 + i"...], 512, 285'184) → 512 ENOENT
SYS  13477  13477        776'009'437 pread(4, [u"trict__ y_qs, const half2 * __restrict__"...], 512, 285'696) → 512 ENOENT
SYS  13477  13477        776'015'082 pread(3, [u"trict__ y_qs, const half2 * __restrict__"...], 512, 285'696) → 512 ENOENT
SYS  13477  13477        776'016'848 pread(4, [u"oid allocate_tiles_q2_K(int ** x_ql, hal"...], 512, 286'208) → 512 ENOENT
SYS  13477  13477        776'018'986 pread(3, [u"oid allocate_tiles_q2_K(int ** x_ql, hal"...], 512, 286'208) → 512 ENOENT
SYS  13477  13477        776'024'260 pread(4, [u"estrict__ vx, int * __restrict__ x_ql, h"...], 512, 286'720) → 512 ENOENT
SYS  13477  13477        776'026'706 pread(3, [u"estrict__ vx, int * __restrict__ x_ql, h"...], 512, 286'720) → 512 ENOENT
SYS  13477  13477        776'028'323 pread(4, [u"◙    for (int i0 = 0; i0 < mmq_y; i0 += "...], 512, 287'232) → 512 ENOENT
SYS  13477  13477        776'033'314 pread(3, [u"◙    for (int i0 = 0; i0 < mmq_y; i0 += "...], 512, 287'232) → 512 ENOENT
SYS  13477  13477        776'035'204 pread(4, [u" * QI2_K + k / blocks_per_tile_x_row) % "...], 512, 287'744) → 512 ENOENT
SYS  13477  13477        776'040'963 pread(3, [u" * QI2_K + k / blocks_per_tile_x_row) % "...], 512, 287'744) → 512 ENOENT
SYS  13477  13477        776'042'686 pread(4, [u"RP_SIZE/4)) / (QI2_K/4);◙◙        x_sc[i"...], 512, 288'256) → 512 ENOENT
SYS  13477  13477        776'045'097 pread(3, [u"RP_SIZE/4)) / (QI2_K/4);◙◙        x_sc[i"...], 512, 288'256) → 512 ENOENT
SYS  13477  13477        776'050'150 pread(4, [u" QI2_K;◙    const int ky  = (k % QI2_K) "...], 512, 288'768) → 512 ENOENT
SYS  13477  13477        776'052'341 pread(3, [u" QI2_K;◙    const int ky  = (k % QI2_K) "...], 512, 288'768) → 512 ENOENT
SYS  13477  13477        776'054'090 pread(4, [u") + ky/4;◙◙    const int index_y = j * W"...], 512, 289'280) → 512 ENOENT
SYS  13477  13477        776'059'063 pread(3, [u") + ky/4;◙◙    const int index_y = j * W"...], 512, 289'280) → 512 ENOENT
SYS  13477  13477        776'060'794 pread(4, [u"x_qh[mmq_y * (WARP_SIZE/2)     + mmq_y/2"...], 512, 289'792) → 512 ENOENT
SYS  13477  13477        776'066'536 pread(3, [u"x_qh[mmq_y * (WARP_SIZE/2)     + mmq_y/2"...], 512, 289'792) → 512 ENOENT
SYS  13477  13477        776'068'197 pread(4, [u"nst int & blocks_per_row) {◙◙    GGML_CU"...], 512, 290'304) → 512 ENOENT
SYS  13477  13477        776'070'299 pread(3, [u"nst int & blocks_per_row) {◙◙    GGML_CU"...], 512, 290'304) → 512 ENOENT
SYS  13477  13477        776'075'670 pread(4, [u"_row + kbx;◙◙        x_ql[i * (WARP_SIZE"...], 512, 290'816) → 512 ENOENT
SYS  13477  13477        776'077'878 pread(3, [u"_row + kbx;◙◙        x_ql[i * (WARP_SIZE"...], 512, 290'816) → 512 ENOENT
SYS  13477  13477        776'079'539 pread(4, [u" kbxd;◙◙        x_dmf[i * (WARP_SIZE/QI3"...], 512, 291'328) → 512 ENOENT
SYS  13477  13477        776'085'519 pread(3, [u" kbxd;◙◙        x_dmf[i * (WARP_SIZE/QI3"...], 512, 291'328) → 512 ENOENT
SYS  13477  13477        776'087'197 pread(4, [u"int_from_uint8(bxi->hmask, k % (QI3_K/2)"...], 512, 291'840) → 512 ENOENT
SYS  13477  13477        776'089'291 pread(3, [u"int_from_uint8(bxi->hmask, k % (QI3_K/2)"...], 512, 291'840) → 512 ENOENT
SYS  13477  13477        776'094'608 pread(4, [u"_uint8(bxi->scales, ksc_low) >> shift_lo"...], 512, 292'352) → 512 ENOENT
SYS  13477  13477        776'096'781 pread(3, [u"_uint8(bxi->scales, ksc_low) >> shift_lo"...], 512, 292'352) → 512 ENOENT
SYS  13477  13477        776'098'415 pread(4, [u" x_dm, const int * __restrict__ x_qh, co"...], 512, 292'864) → 512 ENOENT
SYS  13477  13477        776'103'424 pread(3, [u" x_dm, const int * __restrict__ x_qh, co"...], 512, 292'864) → 512 ENOENT
SYS  13477  13477        776'105'076 pread(4, [u"(int l = 0; l < QR3_K*VDR_Q3_K_Q8_1_MMQ;"...], 512, 293'376) → 512 ENOENT
SYS  13477  13477        776'110'482 pread(3, [u"(int l = 0; l < QR3_K*VDR_Q3_K_Q8_1_MMQ;"...], 512, 293'376) → 512 ENOENT
SYS  13477  13477        776'112'425 pread(4, [u"SIZE;◙    return vec_dot_q3_K_q8_1_impl_"...], 512, 293'888) → 512 ENOENT
SYS  13477  13477        776'117'999 pread(3, [u"SIZE;◙    return vec_dot_q3_K_q8_1_impl_"...], 512, 293'888) → 512 ENOENT
SYS  13477  13477        776'119'668 pread(4, [u"    *x_ql = tile_x_ql;◙    *x_dm = tile_"...], 512, 294'400) → 512 ENOENT
SYS  13477  13477        776'121'965 pread(3, [u"    *x_ql = tile_x_ql;◙    *x_dm = tile_"...], 512, 294'400) → 512 ENOENT
SYS  13477  13477        776'125'313 pread(4, [u"GML_CUDA_ASSUME(k >= 0);◙    GGML_CUDA_A"...], 512, 294'912) → 512 ENOENT
SYS  13477  13477        776'127'716 pread(3, [u"GML_CUDA_ASSUME(k >= 0);◙    GGML_CUDA_A"...], 512, 294'912) → 512 ENOENT
SYS  13477  13477        776'132'459 pread(4, [u"int8_aligned(bxi->qs, kqsx);◙    }◙◙    "...], 512, 295'424) → 512 ENOENT
SYS  13477  13477        776'134'667 pread(3, [u"int8_aligned(bxi->qs, kqsx);◙    }◙◙    "...], 512, 295'424) → 512 ENOENT
SYS  13477  13477        776'136'319 pread(4, [u"K) + i / QI4_K + kbxd] = bxi->dm;◙    }◙"...], 512, 295'936) → 512 ENOENT
SYS  13477  13477        776'142'070 pread(3, [u"K) + i / QI4_K + kbxd] = bxi->dm;◙    }◙"...], 512, 295'936) → 512 ENOENT
SYS  13477  13477        776'143'748 pread(4, [u",...,sc7, m0,...,m3, m4,...,m8◙        i"...], 512, 296'448) → 512 ENOENT
SYS  13477  13477        776'145'948 pread(3, [u",...,sc7, m0,...,m3, m4,...,m8◙        i"...], 512, 296'448) → 512 ENOENT
SYS  13477  13477        776'151'106 pread(4, [u"onst int * __restrict__ y_qs, const half"...], 512, 296'960) → 512 ENOENT
SYS  13477  13477        776'153'368 pread(3, [u"onst int * __restrict__ y_qs, const half"...], 512, 296'960) → 512 ENOENT
SYS  13477  13477        776'154'993 pread(4, [u"t mmq_y> static __device__ __forceinline"...], 512, 297'472) → 512 ENOENT
SYS  13477  13477        776'159'940 pread(3, [u"t mmq_y> static __device__ __forceinline"...], 512, 297'472) → 512 ENOENT
SYS  13477  13477        776'161'768 pread(4, [u"_ void load_tiles_q5_K(◙    const void *"...], 512, 297'984) → 512 ENOENT
SYS  13477  13477        776'167'527 pread(3, [u"_ void load_tiles_q5_K(◙    const void *"...], 512, 297'984) → 512 ENOENT
SYS  13477  13477        776'169'321 pread(4, [u"== k if QK_K == 256◙◙    const block_q5_"...], 512, 298'496) → 512 ENOENT
SYS  13477  13477        776'171'414 pread(3, [u"== k if QK_K == 256◙◙    const block_q5_"...], 512, 298'496) → 512 ENOENT
SYS  13477  13477        776'176'546 pread(4, [u"nst int qh = get_int_from_uint8_aligned("...], 512, 299'008) → 512 ENOENT
SYS  13477  13477        776'178'861 pread(3, [u"nst int qh = get_int_from_uint8_aligned("...], 512, 299'008) → 512 ENOENT
SYS  13477  13477        776'180'486 pread(4, [u"r_tile_x_row = WARP_SIZE / QI5_K; // == "...], 512, 299'520) → 512 ENOENT
SYS  13477  13477        776'184'116 pread(3, [u"r_tile_x_row = WARP_SIZE / QI5_K; // == "...], 512, 299'520) → 512 ENOENT
SYS  13477  13477        776'188'939 pread(4, [u" for (int i0 = 0; i0 < mmq_y; i0 += nwar"...], 512, 300'032) → 512 ENOENT
SYS  13477  13477        776'191'210 pread(3, [u" for (int i0 = 0; i0 < mmq_y; i0 += nwar"...], 512, 300'032) → 512 ENOENT
SYS  13477  13477        776'195'414 pread(4, [u"s[(ksc%2) + (ksc!=0)] >> (4 * (ksc & (ks"...], 512, 300'544) → 512 ENOENT
SYS  13477  13477        776'197'702 pread(3, [u"s[(ksc%2) + (ksc!=0)] >> (4 * (ksc & (ks"...], 512, 300'544) → 512 ENOENT
SYS  13477  13477        776'202'790 pread(4, [u"ds, const int & i, const int & j, const "...], 512, 301'056) → 512 ENOENT
SYS  13477  13477        776'204'963 pread(3, [u"ds, const int & i, const int & j, const "...], 512, 301'056) → 512 ENOENT
SYS  13477  13477        776'206'774 pread(4, [u"nt mmq_y> static __device__ __forceinlin"...], 512, 301'568) → 512 ENOENT
SYS  13477  13477        776'217'498 pread(3, [u"nt mmq_y> static __device__ __forceinlin"...], 512, 301'568) → 512 ENOENT
SYS  13477  13477        776'219'194 pread(4, [u"__ void load_tiles_q6_K(◙    const void "...], 512, 302'080) → 512 ENOENT
SYS  13477  13477        776'224'803 pread(3, [u"__ void load_tiles_q6_K(◙    const void "...], 512, 302'080) → 512 ENOENT
SYS  13477  13477        776'230'218 pread(4, [u" == k if QK_K == 256◙◙    const block_q6"...], 512, 302'592) → 512 ENOENT
SYS  13477  13477        776'235'677 pread(3, [u" == k if QK_K == 256◙◙    const block_q6"...], 512, 302'592) → 512 ENOENT
SYS  13477  13477        776'336'500 pread(4, [u" qh = get_int_from_uint8(bxi->qh, (QI6_K"...], 512, 303'104) → 512 ENOENT
SYS  13477  13477        776'339'539 pread(3, [u" qh = get_int_from_uint8(bxi->qh, (QI6_K"...], 512, 303'104) → 512 ENOENT
SYS  13477  13477        776'344'335 pread(4, [u"1) + kq1] = __vsubss4(ql1 | qh1, 0x20202"...], 512, 303'616) → 512 ENOENT
SYS  13477  13477        776'349'370 pread(3, [u"1) + kq1] = __vsubss4(ql1 | qh1, 0x20202"...], 512, 303'616) → 512 ENOENT
SYS  13477  13477        776'353'708 pread(4, [u"cks_per_row + kbxd;◙◙        x_dmf[i * ("...], 512, 304'128) → 512 ENOENT
SYS  13477  13477        776'355'916 pread(3, [u"cks_per_row + kbxd;◙◙        x_dmf[i * ("...], 512, 304'128) → 512 ENOENT
SYS  13477  13477        776'357'895 pread(4, [u"e__ __forceinline__ float vec_dot_q6_K_q"...], 512, 304'640) → 512 ENOENT
SYS  13477  13477        776'360'076 pread(3, [u"e__ __forceinline__ float vec_dot_q6_K_q"...], 512, 304'640) → 512 ENOENT
SYS  13477  13477        776'361'684 pread(4, [u"nt index_x = i * (QR6_K*WARP_SIZE + 1) +"...], 512, 305'152) → 512 ENOENT
SYS  13477  13477        776'363'972 pread(3, [u"nt index_x = i * (QR6_K*WARP_SIZE + 1) +"...], 512, 305'152) → 512 ENOENT
SYS  13477  13477        776'365'580 pread(4, [u"__forceinline__ void mul_mat_q(◙    cons"...], 512, 305'664) → 512 ENOENT
SYS  13477  13477        776'369'555 pread(3, [u"__forceinline__ void mul_mat_q(◙    cons"...], 512, 305'664) → 512 ENOENT
SYS  13477  13477        776'371'277 pread(4, [u";◙◙    const int row_dst_0 = blockIdx.x*"...], 512, 306'176) → 512 ENOENT
SYS  13477  13477        776'373'424 pread(3, [u";◙◙    const int row_dst_0 = blockIdx.x*"...], 512, 306'176) → 512 ENOENT
SYS  13477  13477        776'375'641 pread(4, [u"mq_x/nwarps] = {{0.0f}};◙◙    for (int i"...], 512, 306'688) → 512 ENOENT
SYS  13477  13477        776'379'686 pread(3, [u"mq_x/nwarps] = {{0.0f}};◙◙    for (int i"...], 512, 306'688) → 512 ENOENT
SYS  13477  13477        776'381'753 pread(4, [u"           const int col_y_eff = min(col"...], 512, 307'200) → 512 ENOENT
SYS  13477  13477        776'383'926 pread(3, [u"           const int col_y_eff = min(col"...], 512, 307'200) → 512 ENOENT
SYS  13477  13477        776'389'324 pread(4, [u"     const int ids = (ids0 + threadIdx.y"...], 512, 307'712) → 512 ENOENT
SYS  13477  13477        776'391'523 pread(3, [u"     const int ids = (ids0 + threadIdx.y"...], 512, 307'712) → 512 ENOENT
SYS  13477  13477        776'393'705 pread(4, [u"P_SIZE/QI8_1) + kby];◙                if"...], 512, 308'224) → 512 ENOENT
SYS  13477  13477        776'397'768 pread(3, [u"P_SIZE/QI8_1) + kby];◙                if"...], 512, 308'224) → 512 ENOENT
SYS  13477  13477        776'399'517 pread(4, [u"warps) {◙#pragma unroll◙                "...], 512, 308'736) → 512 ENOENT
SYS  13477  13477        776'403'422 pread(3, [u"warps) {◙#pragma unroll◙                "...], 512, 308'736) → 512 ENOENT
SYS  13477  13477        776'405'656 pread(4, [u"+ j + threadIdx.y;◙◙        if (col_dst "...], 512, 309'248) → 512 ENOENT
SYS  13477  13477        776'408'757 pread(3, [u"+ j + threadIdx.y;◙◙        if (col_dst "...], 512, 309'248) → 512 ENOENT
SYS  13477  13477        776'410'859 pread(4, [u"PBLAS) && defined(__HIP_PLATFORM_AMD__)◙"...], 512, 309'760) → 512 ENOENT
SYS  13477  13477        776'414'790 pread(3, [u"PBLAS) && defined(__HIP_PLATFORM_AMD__)◙"...], 512, 309'760) → 512 ENOENT
SYS  13477  13477        776'416'460 pread(4, [u"if defined(RDNA3) || defined(RDNA2)◙    "...], 512, 310'272) → 512 ENOENT
SYS  13477  13477        776'418'553 pread(3, [u"if defined(RDNA3) || defined(RDNA2)◙    "...], 512, 310'272) → 512 ENOENT
SYS  13477  13477        776'421'804 pread(4, [u"_config_device(MMQ_CONFIG_Q4_0);◙◙    mu"...], 512, 310'784) → 512 ENOENT
SYS  13477  13477        776'426'795 pread(3, [u"_config_device(MMQ_CONFIG_Q4_0);◙◙    mu"...], 512, 310'784) → 512 ENOENT
SYS  13477  13477        776'430'505 pread(4, [u"◙◙template <bool need_check> static __gl"...], 512, 311'296) → 512 ENOENT
SYS  13477  13477        776'435'911 pread(3, [u"◙◙template <bool need_check> static __gl"...], 512, 311'296) → 512 ENOENT
SYS  13477  13477        776'440'318 pread(4, [u"t ncols_x, const int nrows_x, const int "...], 512, 311'808) → 512 ENOENT
SYS  13477  13477        776'442'465 pread(3, [u"t ncols_x, const int nrows_x, const int "...], 512, 311'808) → 512 ENOENT
SYS  13477  13477        776'446'811 pread(4, [u"ols_y, nrows_y, nrows_dst);◙#else◙    GG"...], 512, 312'320) → 512 ENOENT
SYS  13477  13477        776'451'872 pread(3, [u"ols_y, nrows_y, nrows_dst);◙#else◙    GG"...], 512, 312'320) → 512 ENOENT
SYS  13477  13477        776'456'210 pread(4, [u"mul_mat_q5_0(◙    const void * __restric"...], 512, 312'832) → 512 ENOENT
SYS  13477  13477        776'458'550 pread(3, [u"mul_mat_q5_0(◙    const void * __restric"...], 512, 312'832) → 512 ENOENT
SYS  13477  13477        776'460'193 pread(4, [u"_config.nwarps, need_check>, VDR_Q5_0_Q8"...], 512, 313'344) → 512 ENOENT
SYS  13477  13477        776'467'039 pread(3, [u"_config.nwarps, need_check>, VDR_Q5_0_Q8"...], 512, 313'344) → 512 ENOENT
SYS  13477  13477        776'472'065 pread(4, [u"◙#endif // defined(RDNA3) || defined(RDN"...], 512, 313'856) → 512 ENOENT
SYS  13477  13477        776'476'226 pread(3, [u"◙#endif // defined(RDNA3) || defined(RDN"...], 512, 313'856) → 512 ENOENT
SYS  13477  13477        776'477'913 pread(4, [u"ch_config.y, arch_config.nwarps, allocat"...], 512, 314'368) → 512 ENOENT
SYS  13477  13477        776'481'826 pread(3, [u"ch_config.y, arch_config.nwarps, allocat"...], 512, 314'368) → 512 ENOENT
SYS  13477  13477        776'485'510 pread(4, [u"ORM_AMD__)◙#if defined(RDNA3) || defined"...], 512, 314'880) → 512 ENOENT
SYS  13477  13477        776'487'798 pread(3, [u"ORM_AMD__)◙#if defined(RDNA3) || defined"...], 512, 314'880) → 512 ENOENT
SYS  13477  13477        776'489'476 pread(4, [u"g = get_arch_config_device(MMQ_CONFIG_Q8"...], 512, 315'392) → 512 ENOENT
SYS  13477  13477        776'493'530 pread(3, [u"g = get_arch_config_device(MMQ_CONFIG_Q8"...], 512, 315'392) → 512 ENOENT
SYS  13477  13477        776'495'200 pread(4, [u"MIN_CC_DP4A◙}◙◙template <bool need_check"...], 512, 315'904) → 512 ENOENT
SYS  13477  13477        776'497'346 pread(3, [u"MIN_CC_DP4A◙}◙◙template <bool need_check"...], 512, 315'904) → 512 ENOENT
SYS  13477  13477        776'502'178 pread(4, [u" int nrows_y, const int nrows_dst) {◙◙#i"...], 512, 316'416) → 512 ENOENT
SYS  13477  13477        776'506'674 pread(3, [u" int nrows_y, const int nrows_dst) {◙◙#i"...], 512, 316'416) → 512 ENOENT
SYS  13477  13477        776'510'164 pread(4, [u"arch_config_device);◙    GGML_UNUSED(vec"...], 512, 316'928) → 512 ENOENT
SYS  13477  13477        776'512'363 pread(3, [u"arch_config_device);◙    GGML_UNUSED(vec"...], 512, 316'928) → 512 ENOENT
SYS  13477  13477        776'514'015 pread(4, [u"OLTA◙    mul_mat_q3_K(◙    const void * "...], 512, 317'440) → 512 ENOENT
SYS  13477  13477        776'516'250 pread(3, [u"OLTA◙    mul_mat_q3_K(◙    const void * "...], 512, 317'440) → 512 ENOENT
SYS  13477  13477        776'519'774 pread(4, [u".y, arch_config.nwarps, need_check>, VDR"...], 512, 317'952) → 512 ENOENT
SYS  13477  13477        776'522'521 pread(3, [u".y, arch_config.nwarps, need_check>, VDR"...], 512, 317'952) → 512 ENOENT
SYS  13477  13477        776'524'694 pread(4, [u"arps, 2)◙#endif // defined(RDNA3) || def"...], 512, 318'464) → 512 ENOENT
SYS  13477  13477        776'528'908 pread(3, [u"arps, 2)◙#endif // defined(RDNA3) || def"...], 512, 318'464) → 512 ENOENT
SYS  13477  13477        776'532'335 pread(4, [u"K);◙◙    mul_mat_q<QK_K, QR4_K, QI4_K, t"...], 512, 318'976) → 512 ENOENT
SYS  13477  13477        776'537'918 pread(3, [u"K);◙◙    mul_mat_q<QK_K, QR4_K, QI4_K, t"...], 512, 318'976) → 512 ENOENT
SYS  13477  13477        776'541'345 pread(4, [u"tatic __global__ void◙#if defined(GGML_U"...], 512, 319'488) → 512 ENOENT
SYS  13477  13477        776'545'002 pread(3, [u"tatic __global__ void◙#if defined(GGML_U"...], 512, 319'488) → 512 ENOENT
SYS  13477  13477        776'548'518 pread(4, [u"_CUDA_ARCH__ >= MIN_CC_DP4A◙    constexp"...], 512, 320'000) → 512 ENOENT
SYS  13477  13477        776'550'700 pread(3, [u"_CUDA_ARCH__ >= MIN_CC_DP4A◙    constexp"...], 512, 320'000) → 512 ENOENT
SYS  13477  13477        776'552'334 pread(4, [u"_q5_K_q8_1_mul_mat);◙    NO_DEVICE_CODE;"...], 512, 320'512) → 512 ENOENT
SYS  13477  13477        776'554'454 pread(3, [u"_q5_K_q8_1_mul_mat);◙    NO_DEVICE_CODE;"...], 512, 320'512) → 512 ENOENT
SYS  13477  13477        776'556'592 pread(4, [u"strict__ vx, const void * __restrict__ v"...], 512, 321'024) → 512 ENOENT
SYS  13477  13477        776'560'796 pread(3, [u"strict__ vx, const void * __restrict__ v"...], 512, 321'024) → 512 ENOENT
SYS  13477  13477        776'567'669 pread(4, [u"K_Q8_1_MMQ, vec_dot_q6_K_q8_1_mul_mat>◙ "...], 512, 321'536) → 512 ENOENT
SYS  13477  13477        776'571'599 pread(3, [u"K_Q8_1_MMQ, vec_dot_q6_K_q8_1_mul_mat>◙ "...], 512, 321'536) → 512 ENOENT
SYS  13477  13477        776'574'408 pread(4, [u"= false;                                "...], 512, 322'048) → 512 ENOENT
SYS  13477  13477        776'577'500 pread(3, [u"= false;                                "...], 512, 322'048) → 512 ENOENT
SYS  13477  13477        776'579'567 pread(4, [u"      \\◙        mul_mat_q##type_suffix<n"...], 512, 322'560) → 512 ENOENT
SYS  13477  13477        776'583'613 pread(3, [u"      \\◙        mul_mat_q##type_suffix<n"...], 512, 322'560) → 512 ENOENT
SYS  13477  13477        776'585'362 pread(4, [u"oat * src1_ddf_i,◙    const char * src1_"...], 512, 323'072) → 512 ENOENT
SYS  13477  13477        776'587'915 pread(3, [u"oat * src1_ddf_i,◙    const char * src1_"...], 512, 323'072) → 512 ENOENT
SYS  13477  13477        776'590'052 pread(4, [u" // the main device has a larger memory "...], 512, 323'584) → 512 ENOENT
SYS  13477  13477        776'593'992 pread(3, [u" // the main device has a larger memory "...], 512, 323'584) → 512 ENOENT
SYS  13477  13477        776'595'714 pread(4, [u"FIG_Q5_0;◙            break;◙        cas"...], 512, 324'096) → 512 ENOENT
SYS  13477  13477        776'597'834 pread(3, [u"FIG_Q5_0;◙            break;◙        cas"...], 512, 324'096) → 512 ENOENT
SYS  13477  13477        776'602'975 pread(4, [u"        mmq_config = MMQ_CONFIG_Q5_K;◙  "...], 512, 324'608) → 512 ENOENT
SYS  13477  13477        776'605'299 pread(3, [u"        mmq_config = MMQ_CONFIG_Q5_K;◙  "...], 512, 324'608) → 512 ENOENT
SYS  13477  13477        776'608'558 pread(4, [u"ig.ampere;◙    } else if (compute_capabi"...], 512, 325'120) → 512 ENOENT
SYS  13477  13477        776'613'143 pread(3, [u"ig.ampere;◙    } else if (compute_capabi"...], 512, 325'120) → 512 ENOENT
SYS  13477  13477        776'616'861 pread(4, [u"      MMQ_SWITCH_CASE(5_0)◙        MMQ_S"...], 512, 325'632) → 512 ENOENT
SYS  13477  13477        776'621'305 pread(3, [u"      MMQ_SWITCH_CASE(5_0)◙        MMQ_S"...], 512, 325'632) → 512 ENOENT
SYS  13477  13477        776'625'032 pread(4, [u"       case GGML_TYPE_Q5_0:◙        case"...], 512, 326'144) → 512 ENOENT
SYS  13477  13477        776'629'281 pread(3, [u"       case GGML_TYPE_Q5_0:◙        case"...], 512, 326'144) → 512 ENOENT
SYS  13477  13477        776'633'026 pread(4, [u"////////////////////////////////////////"...], 512, 326'656) → 512 ENOENT
SYS  13477  13477        776'636'418 pread(3, [u"////////////////////////////////////////"...], 512, 326'656) → 512 ENOENT
SYS  13477  13477        776'638'583 pread(4, [u"added_row_size, cudaStream_t stream);◙◙t"...], 512, 327'168) → 512 ENOENT
SYS  13477  13477        776'642'558 pread(3, [u"added_row_size, cudaStream_t stream);◙◙t"...], 512, 327'168) → 512 ENOENT
SYS  13477  13477        776'644'271 pread(4, [u"defined(__HIP_PLATFORM_AMD__))◙static __"...], 512, 327'680) → 512 ENOENT
SYS  13477  13477        776'646'559 pread(3, [u"defined(__HIP_PLATFORM_AMD__))◙static __"...], 512, 327'680) → 512 ENOENT
SYS  13477  13477        776'649'377 pread(4, [u" int rows_per_cuda_block = ncols_y == 1 "...], 512, 328'192) → 512 ENOENT
SYS  13477  13477        776'653'670 pread(3, [u" int rows_per_cuda_block = ncols_y == 1 "...], 512, 328'192) → 512 ENOENT
SYS  13477  13477        776'658'767 pread(4, [u"a_block] = {0.0f};◙◙    const block_q_t "...], 512, 328'704) → 512 ENOENT
SYS  13477  13477        776'664'420 pread(3, [u"a_block] = {0.0f};◙◙    const block_q_t "...], 512, 328'704) → 512 ENOENT
SYS  13477  13477        776'667'326 pread(4, [u"_per_cuda_block; ++i) {◙                "...], 512, 329'216) → 512 ENOENT
SYS  13477  13477        776'671'266 pread(3, [u"_per_cuda_block; ++i) {◙                "...], 512, 329'216) → 512 ENOENT
SYS  13477  13477        776'672'997 pread(4, [u"dIdx.x] = tmp[j][i];◙            }◙     "...], 512, 329'728) → 512 ENOENT
SYS  13477  13477        776'675'232 pread(3, [u"dIdx.x] = tmp[j][i];◙            }◙     "...], 512, 329'728) → 512 ENOENT
SYS  13477  13477        776'678'933 pread(4, [u"(threadIdx.x < rows_per_cuda_block) {◙  "...], 512, 330'240) → 512 ENOENT
SYS  13477  13477        776'681'212 pread(3, [u"(threadIdx.x < rows_per_cuda_block) {◙  "...], 512, 330'240) → 512 ENOENT
SYS  13477  13477        776'685'196 pread(4, [u" id = ggml_cuda_get_device();◙◙    int64"...], 512, 330'752) → 512 ENOENT
SYS  13477  13477        776'687'475 pread(3, [u" id = ggml_cuda_get_device();◙◙    int64"...], 512, 330'752) → 512 ENOENT
SYS  13477  13477        776'691'238 pread(4, [u"            case 7:◙            case 8:◙"...], 512, 331'264) → 512 ENOENT
SYS  13477  13477        776'695'469 pread(3, [u"            case 7:◙            case 8:◙"...], 512, 331'264) → 512 ENOENT
SYS  13477  13477        776'697'262 pread(4, [u"            <<<block_nums, block_dims, 0"...], 512, 331'776) → 512 ENOENT
SYS  13477  13477        776'699'506 pread(3, [u"            <<<block_nums, block_dims, 0"...], 512, 331'776) → 512 ENOENT
SYS  13477  13477        776'703'110 pread(4, [u";◙            break;◙        case 4:◙   "...], 512, 332'288) → 512 ENOENT
SYS  13477  13477        776'706'997 pread(3, [u";◙            break;◙        case 4:◙   "...], 512, 332'288) → 512 ENOENT
SYS  13477  13477        776'709'134 pread(4, [u"             <<<block_nums, block_dims, "...], 512, 332'800) → 512 ENOENT
SYS  13477  13477        776'713'295 pread(3, [u"             <<<block_nums, block_dims, "...], 512, 332'800) → 512 ENOENT
SYS  13477  13477        776'714'973 pread(4, [u");◙            break;◙        default:◙ "...], 512, 333'312) → 512 ENOENT
SYS  13477  13477        776'718'860 pread(3, [u");◙            break;◙        default:◙ "...], 512, 333'312) → 512 ENOENT
SYS  13477  13477        776'721'068 pread(4, [u"_q8_1_cuda(◙    const void * vx, const v"...], 512, 333'824) → 512 ENOENT
SYS  13477  13477        776'724'107 pread(3, [u"_q8_1_cuda(◙    const void * vx, const v"...], 512, 333'824) → 512 ENOENT
SYS  13477  13477        776'726'183 pread(4, [u"st int ncols_y, const int nrows_dst, cud"...], 512, 334'336) → 512 ENOENT
SYS  13477  13477        776'730'087 pread(3, [u"st int ncols_y, const int nrows_dst, cud"...], 512, 334'336) → 512 ENOENT
SYS  13477  13477        776'731'739 pread(4, [u"Q, vec_dot_q5_1_q8_1>◙        (vx, vy, d"...], 512, 334'848) → 512 ENOENT
SYS  13477  13477        776'733'824 pread(3, [u"Q, vec_dot_q5_1_q8_1>◙        (vx, vy, d"...], 512, 334'848) → 512 ENOENT
SYS  13477  13477        776'738'664 pread(4, [u"q2_K_q8_1_cuda(◙    const void * vx, con"...], 512, 335'360) → 512 ENOENT
SYS  13477  13477        776'740'873 pread(3, [u"q2_K_q8_1_cuda(◙    const void * vx, con"...], 512, 335'360) → 512 ENOENT
SYS  13477  13477        776'744'176 pread(4, [u"const int ncols_y, const int nrows_dst, "...], 512, 335'872) → 512 ENOENT
SYS  13477  13477        776'749'936 pread(3, [u"const int ncols_y, const int nrows_dst, "...], 512, 335'872) → 512 ENOENT
SYS  13477  13477        776'752'038 pread(4, [u"VQ, vec_dot_q4_K_q8_1>◙        (vx, vy, "...], 512, 336'384) → 512 ENOENT
SYS  13477  13477        776'754'608 pread(3, [u"VQ, vec_dot_q4_K_q8_1>◙        (vx, vy, "...], 512, 336'384) → 512 ENOENT
SYS  13477  13477        776'756'746 pread(4, [u"q6_K_q8_1_cuda(◙    const void * vx, con"...], 512, 336'896) → 512 ENOENT
SYS  13477  13477        776'760'765 pread(3, [u"q6_K_q8_1_cuda(◙    const void * vx, con"...], 512, 336'896) → 512 ENOENT
SYS  13477  13477        776'764'617 pread(4, [u"y, const int ncols_y, const int nrows_ds"...], 512, 337'408) → 512 ENOENT
SYS  13477  13477        776'768'883 pread(3, [u"y, const int ncols_y, const int nrows_ds"...], 512, 337'408) → 512 ENOENT
SYS  13477  13477        776'771'047 pread(4, [u"_q8_1>◙        (vx, vy, dst, ncols_x, nr"...], 512, 337'920) → 512 ENOENT
SYS  13477  13477        776'775'049 pread(3, [u"_q8_1>◙        (vx, vy, dst, ncols_x, nr"...], 512, 337'920) → 512 ENOENT
SYS  13477  13477        776'776'762 pread(4, [u"oid * vx, const void * vy, float * dst,◙"...], 512, 338'432) → 512 ENOENT
SYS  13477  13477        776'778'962 pread(3, [u"oid * vx, const void * vy, float * dst,◙"...], 512, 338'432) → 512 ENOENT
SYS  13477  13477        776'781'912 pread(4, [u"dst, cudaStream_t stream) {◙◙    mul_mat"...], 512, 338'944) → 512 ENOENT
SYS  13477  13477        776'785'278 pread(3, [u"dst, cudaStream_t stream) {◙◙    mul_mat"...], 512, 338'944) → 512 ENOENT
SYS  13477  13477        776'788'899 pread(4, [u"rows_y, ncols_y, nrows_dst, stream);◙}◙◙"...], 512, 339'456) → 512 ENOENT
SYS  13477  13477        776'791'099 pread(3, [u"rows_y, ncols_y, nrows_dst, stream);◙}◙◙"...], 512, 339'456) → 512 ENOENT
SYS  13477  13477        776'794'606 pread(4, [u" float * dst,◙    const int ncols_x, con"...], 512, 339'968) → 512 ENOENT
SYS  13477  13477        776'796'832 pread(3, [u" float * dst,◙    const int ncols_x, con"...], 512, 339'968) → 512 ENOENT
SYS  13477  13477        776'798'537 pread(4, [u"    mul_mat_vec_q_cuda<QK_K, QI3_XS, blo"...], 512, 340'480) → 512 ENOENT
SYS  13477  13477        776'803'907 pread(3, [u"    mul_mat_vec_q_cuda<QK_K, QI3_XS, blo"...], 512, 340'480) → 512 ENOENT
SYS  13477  13477        776'806'707 pread(4, [u"aStream_t stream) {◙◙    const int64_t n"...], 512, 340'992) → 512 ENOENT
SYS  13477  13477        776'810'674 pread(3, [u"aStream_t stream) {◙◙    const int64_t n"...], 512, 340'992) → 512 ENOENT
SYS  13477  13477        776'812'414 pread(4, [u"ase GGML_TYPE_Q4_0:◙            mul_mat_"...], 512, 341'504) → 512 ENOENT
SYS  13477  13477        776'814'490 pread(3, [u"ase GGML_TYPE_Q4_0:◙            mul_mat_"...], 512, 341'504) → 512 ENOENT
SYS  13477  13477        776'818'067 pread(4, [u"c1_ncols, nrows_dst, stream);◙          "...], 512, 342'016) → 512 ENOENT
SYS  13477  13477        776'822'157 pread(3, [u"c1_ncols, nrows_dst, stream);◙          "...], 512, 342'016) → 512 ENOENT
SYS  13477  13477        776'825'575 pread(4, [u"_ddq_i, dst_dd_i, ne00, row_diff, src1_p"...], 512, 342'528) → 512 ENOENT
SYS  13477  13477        776'827'819 pread(3, [u"_ddq_i, dst_dd_i, ne00, row_diff, src1_p"...], 512, 342'528) → 512 ENOENT
SYS  13477  13477        776'831'299 pread(4, [u"_K:◙            mul_mat_vec_q5_K_q8_1_cu"...], 512, 343'040) → 512 ENOENT
SYS  13477  13477        776'833'490 pread(3, [u"_K:◙            mul_mat_vec_q5_K_q8_1_cu"...], 512, 343'040) → 512 ENOENT
SYS  13477  13477        776'835'204 pread(4, [u"nrows_dst, stream);◙            break;◙ "...], 512, 343'552) → 512 ENOENT
SYS  13477  13477        776'839'567 pread(3, [u"nrows_dst, stream);◙            break;◙ "...], 512, 343'552) → 512 ENOENT
SYS  13477  13477        776'843'578 pread(4, [u"c1_ddq_i, dst_dd_i, ne00, row_diff, src1"...], 512, 344'064) → 512 ENOENT
SYS  13477  13477        776'848'922 pread(3, [u"c1_ddq_i, dst_dd_i, ne00, row_diff, src1"...], 512, 344'064) → 512 ENOENT
SYS  13477  13477        776'852'385 pread(4, [u"YPE_IQ4_NL:◙            mul_mat_vec_iq4_"...], 512, 344'576) → 512 ENOENT
SYS  13477  13477        776'856'492 pread(3, [u"YPE_IQ4_NL:◙            mul_mat_vec_iq4_"...], 512, 344'576) → 512 ENOENT
SYS  13477  13477        776'859'831 pread(4, [u"c1_ncols, nrows_dst, stream);◙          "...], 512, 345'088) → 512 ENOENT
SYS  13477  13477        776'864'548 pread(3, [u"c1_ncols, nrows_dst, stream);◙          "...], 512, 345'088) → 512 ENOENT
SYS  13477  13477        776'868'302 pread(4, [u"/////////////////////////◙//◙// ROLLUP n"...], 512, 345'600) → 512 ENOENT
SYS  13477  13477        776'872'118 pread(3, [u"/////////////////////////◙//◙// ROLLUP n"...], 512, 345'600) → 512 ENOENT
SYS  13477  13477        776'875'095 pread(4, [u" row = blockIdx.x*blockDim.y + threadIdx"...], 512, 346'112) → 512 ENOENT
SYS  13477  13477        776'879'697 pread(3, [u" row = blockIdx.x*blockDim.y + threadIdx"...], 512, 346'112) → 512 ENOENT
SYS  13477  13477        776'883'531 pread(4, [u"% WARP_SIZE;◙        if (lane_id == 0) {"...], 512, 346'624) → 512 ENOENT
SYS  13477  13477        776'889'387 pread(3, [u"% WARP_SIZE;◙        if (lane_id == 0) {"...], 512, 346'624) → 512 ENOENT
SYS  13477  13477        776'893'080 pread(4, [u"tatic __global__ void group_norm_f32(con"...], 512, 347'136) → 512 ENOENT
SYS  13477  13477        776'896'604 pread(3, [u"tatic __global__ void group_norm_f32(con"...], 512, 347'136) → 512 ENOENT
SYS  13477  13477        776'898'866 pread(4, [u"= warp_reduce_sum(tmp);◙    if (block_si"...], 512, 347'648) → 512 ENOENT
SYS  13477  13477        776'902'187 pread(3, [u"= warp_reduce_sum(tmp);◙    if (block_si"...], 512, 347'648) → 512 ENOENT
SYS  13477  13477        776'904'325 pread(4, [u"i;◙        tmp += xi * xi;◙    }◙◙    tm"...], 512, 348'160) → 512 ENOENT
SYS  13477  13477        776'908'361 pread(3, [u"i;◙        tmp += xi * xi;◙    }◙◙    tm"...], 512, 348'160) → 512 ENOENT
SYS  13477  13477        776'910'057 pread(4, [u"; j += block_size) {◙        dst[j] *= s"...], 512, 348'672) → 512 ENOENT
SYS  13477  13477        776'912'328 pread(3, [u"; j += block_size) {◙        dst[j] *= s"...], 512, 348'672) → 512 ENOENT
SYS  13477  13477        776'916'833 pread(4, [u"e_sum(tmp);◙    if (block_size > WARP_SI"...], 512, 349'184) → 512 ENOENT
SYS  13477  13477        776'919'032 pread(3, [u"e_sum(tmp);◙    if (block_size > WARP_SI"...], 512, 349'184) → 512 ENOENT
SYS  13477  13477        776'920'684 pread(4, [u"] = scale * x[row*ncols + col];◙    }◙}◙"...], 512, 349'696) → 512 ENOENT
SYS  13477  13477        776'923'360 pread(3, [u"] = scale * x[row*ncols + col];◙    }◙}◙"...], 512, 349'696) → 512 ENOENT
SYS  13477  13477        776'926'929 pread(4, [u"atic void group_norm_f32_cuda(const floa"...], 512, 350'208) → 512 ENOENT
SYS  13477  13477        776'929'093 pread(3, [u"atic void group_norm_f32_cuda(const floa"...], 512, 350'208) → 512 ENOENT
SYS  13477  13477        776'932'529 pread(4, [u"_size, ne_elements, eps);◙    }◙}◙◙stati"...], 512, 350'720) → 512 ENOENT
SYS  13477  13477        776'934'747 pread(3, [u"_size, ne_elements, eps);◙    }◙}◙◙stati"...], 512, 350'720) → 512 ENOENT
SYS  13477  13477        776'938'342 pread(4, [u"◙}◙◙void ggml_cuda_op_norm(ggml_backend_"...], 512, 351'232) → 512 ENOENT
SYS  13477  13477        776'940'603 pread(3, [u"◙}◙◙void ggml_cuda_op_norm(ggml_backend_"...], 512, 351'232) → 512 ENOENT
SYS  13477  13477        776'942'891 pread(4, [u"2_cuda(src0_d, dst_d, ne00, nrows, eps, "...], 512, 351'744) → 512 ENOENT
SYS  13477  13477        776'946'424 pread(3, [u"2_cuda(src0_d, dst_d, ne00, nrows, eps, "...], 512, 351'744) → 512 ENOENT
SYS  13477  13477        776'948'535 pread(4, [u"num_groups - 1) / num_groups);◙    group"...], 512, 352'256) → 512 ENOENT
SYS  13477  13477        776'952'917 pread(3, [u"num_groups - 1) / num_groups);◙    group"...], 512, 352'256) → 512 ENOENT
SYS  13477  13477        776'954'613 pread(4, [u" ne00 = src0->ne[0];◙    const int64_t n"...], 512, 352'768) → 512 ENOENT
SYS  13477  13477        776'956'759 pread(3, [u" ne00 = src0->ne[0];◙    const int64_t n"...], 512, 352'768) → 512 ENOENT
SYS  13477  13477        776'960'266 pread(4, [u"////////////////////////////////////////"...], 512, 353'280) → 512 ENOENT
SYS  13477  13477        776'962'439 pread(3, [u"////////////////////////////////////////"...], 512, 353'280) → 512 ENOENT
SYS  13477  13477        776'965'884 pread(4, [u"       return;◙    }◙◙    // operation◙ "...], 512, 353'792) → 512 ENOENT
SYS  13477  13477        776'968'119 pread(3, [u"       return;◙    }◙◙    // operation◙ "...], 512, 353'792) → 512 ENOENT
SYS  13477  13477        776'969'877 pread(4, [u"nst int ne02, const int ne03,◙    const "...], 512, 354'304) → 512 ENOENT
SYS  13477  13477        776'974'214 pread(3, [u"nst int ne02, const int ne03,◙    const "...], 512, 354'304) → 512 ENOENT
SYS  13477  13477        776'978'366 pread(4, [u" float * dst_d = (float *)dst->data;◙   "...], 512, 354'816) → 512 ENOENT
SYS  13477  13477        776'982'190 pread(3, [u" float * dst_d = (float *)dst->data;◙   "...], 512, 354'816) → 512 ENOENT
SYS  13477  13477        776'984'381 pread(4, [u"////////////////////////////////////////"...], 512, 355'328) → 512 ENOENT
SYS  13477  13477        776'987'508 pread(3, [u"////////////////////////////////////////"...], 512, 355'328) → 512 ENOENT
SYS  13477  13477        776'990'794 pread(4, [u" oh, const int ow,◙        const int kh,"...], 512, 355'840) → 512 ENOENT
SYS  13477  13477        776'995'317 pread(3, [u" oh, const int ow,◙        const int kh,"...], 512, 355'840) → 512 ENOENT
SYS  13477  13477        776'999'036 pread(4, [u" = src + nc * I_HW;◙    To* o_ptr = dst "...], 512, 356'352) → 512 ENOENT
SYS  13477  13477        777'003'488 pread(3, [u" = src + nc * I_HW;◙    To* o_ptr = dst "...], 512, 356'352) → 512 ENOENT
SYS  13477  13477        777'006'199 pread(4, [u" (int i = bh; i < eh; i += 1) {◙        "...], 512, 356'864) → 512 ENOENT
SYS  13477  13477        777'010'413 pread(3, [u" (int i = bh; i < eh; i += 1) {◙        "...], 512, 356'864) → 512 ENOENT
SYS  13477  13477        777'015'527 pread(4, [u"32_cuda(◙        const int ih, const int"...], 512, 357'376) → 512 ENOENT
SYS  13477  13477        777'021'057 pread(3, [u"32_cuda(◙        const int ih, const int"...], 512, 357'376) → 512 ENOENT
SYS  13477  13477        777'024'696 pread(4, [u"h, iw, oh, ow, kh, kw, sh, sw, ph, pw, p"...], 512, 357'888) → 512 ENOENT
SYS  13477  13477        777'028'115 pread(3, [u"h, iw, oh, ow, kh, kw, sh, sw, ph, pw, p"...], 512, 357'888) → 512 ENOENT
SYS  13477  13477        777'031'692 pread(4, [u"= static_cast<ggml_op_pool>(opts[0]);◙  "...], 512, 358'400) → 512 ENOENT
SYS  13477  13477        777'033'874 pread(3, [u"= static_cast<ggml_op_pool>(opts[0]);◙  "...], 512, 358'400) → 512 ENOENT
SYS  13477  13477        777'037'399 pread(4, [u" IW, OH, OW, k1, k0, s1, s0, p1, p0, par"...], 512, 358'912) → 512 ENOENT
SYS  13477  13477        777'039'713 pread(3, [u" IW, OH, OW, k1, k0, s1, s0, p1, p0, par"...], 512, 358'912) → 512 ENOENT
SYS  13477  13477        777'041'400 pread(4, [u"id quantize_row_q8_1_cuda(const float * "...], 512, 359'424) → 512 ENOENT
SYS  13477  13477        777'045'578 pread(3, [u"id quantize_row_q8_1_cuda(const float * "...], 512, 359'424) → 512 ENOENT
SYS  13477  13477        777'050'984 pread(4, [u"ed + ix;◙◙    block_q8_1 * y = (block_q8"...], 512, 359'936) → 512 ENOENT
SYS  13477  13477        777'054'792 pread(3, [u"ed + ix;◙◙    block_q8_1 * y = (block_q8"...], 512, 359'936) → 512 ENOENT
SYS  13477  13477        777'057'026 pread(4, [u"lf&>(y[ib].ds.x) = d;◙    reinterpret_ca"...], 512, 360'448) → 512 ENOENT
SYS  13477  13477        777'061'337 pread(3, [u"lf&>(y[ib].ds.x) = d;◙    reinterpret_ca"...], 512, 360'448) → 512 ENOENT
SYS  13477  13477        777'065'056 pread(4, [u"////////////////////////////////////////"...], 512, 360'960) → 512 ENOENT
SYS  13477  13477        777'070'321 pread(3, [u"////////////////////////////////////////"...], 512, 360'960) → 512 ENOENT
SYS  13477  13477        777'073'889 pread(4, [u"at v[4];◙};◙◙static __device__ float rop"...], 512, 361'472) → 512 ENOENT
SYS  13477  13477        777'078'006 pread(3, [u"at v[4];◙};◙◙static __device__ float rop"...], 512, 361'472) → 512 ENOENT
SYS  13477  13477        777'081'300 pread(4, [u", float mscale,◙    float * cos_theta, f"...], 512, 361'984) → 512 ENOENT
SYS  13477  13477        777'084'772 pread(3, [u", float mscale,◙    float * cos_theta, f"...], 512, 361'984) → 512 ENOENT
SYS  13477  13477        777'086'927 pread(4, [u"eq_scale);◙    }◙    *cos_theta = cosf(t"...], 512, 362'496) → 512 ENOENT
SYS  13477  13477        777'090'946 pread(3, [u"eq_scale);◙    }◙    *cos_theta = cosf(t"...], 512, 362'496) → 512 ENOENT
SYS  13477  13477        777'094'639 pread(4, [u"turn;◙    }◙◙    const int row = blockDi"...], 512, 363'008) → 512 ENOENT
SYS  13477  13477        777'098'826 pread(3, [u"turn;◙    }◙◙    const int row = blockDi"...], 512, 363'008) → 512 ENOENT
SYS  13477  13477        777'102'386 pread(4, [u" = x0*sin_theta + x1*cos_theta;◙}◙◙templ"...], 512, 363'520) → 512 ENOENT
SYS  13477  13477        777'105'884 pread(3, [u" = x0*sin_theta + x1*cos_theta;◙}◙◙templ"...], 512, 363'520) → 512 ENOENT
SYS  13477  13477        777'109'364 pread(4, [u".x + threadIdx.x;◙    const int ib = col"...], 512, 364'032) → 512 ENOENT
SYS  13477  13477        777'111'537 pread(3, [u".x + threadIdx.x;◙    const int ib = col"...], 512, 364'032) → 512 ENOENT
SYS  13477  13477        777'115'185 pread(4, [u" p*freq_scale*powf(theta_scale, col/2.0f"...], 512, 364'544) → 512 ENOENT
SYS  13477  13477        777'117'376 pread(3, [u" p*freq_scale*powf(theta_scale, col/2.0f"...], 512, 364'544) → 512 ENOENT
SYS  13477  13477        777'119'072 pread(4, [u"s, float freq_base,◙    int n_ctx◙) {◙  "...], 512, 365'056) → 512 ENOENT
SYS  13477  13477        777'122'799 pread(3, [u"s, float freq_base,◙    int n_ctx◙) {◙  "...], 512, 365'056) → 512 ENOENT
SYS  13477  13477        777'126'368 pread(4, [u"req_scale*col_theta_scale;◙    const flo"...], 512, 365'568) → 512 ENOENT
SYS  13477  13477        777'128'559 pread(3, [u"req_scale*col_theta_scale;◙    const flo"...], 512, 365'568) → 512 ENOENT
SYS  13477  13477        892'836'357 pread(4, [u"_n_dims * 2];◙    const float x3 = x[i +"...], 512, 366'080) → 512 ENOENT
SYS  13477  13477        892'842'814 pread(3, [u"_n_dims * 2];◙    const float x3 = x[i +"...], 512, 366'080) → 512 ENOENT
SYS  13477  13477        892'846'497 pread(4, [u" block_dims(1, CUDA_ROPE_BLOCK_SIZE, 1);"...], 512, 366'592) → 512 ENOENT
SYS  13477  13477        892'850'958 pread(3, [u" block_dims(1, CUDA_ROPE_BLOCK_SIZE, 1);"...], 512, 366'592) → 512 ENOENT
SYS  13477  13477        892'855'552 pread(4, [u"elta_rows, freq_base, ext_factor, attn_f"...], 512, 367'104) → 512 ENOENT
SYS  13477  13477        892'860'260 pread(3, [u"elta_rows, freq_base, ext_factor, attn_f"...], 512, 367'104) → 512 ENOENT
SYS  13477  13477        892'872'105 pread(4, [u"ROPE_BLOCK_SIZE - 1) / (2*CUDA_ROPE_BLOC"...], 512, 367'616) → 512 ENOENT
SYS  13477  13477        892'876'487 pread(3, [u"ROPE_BLOCK_SIZE - 1) / (2*CUDA_ROPE_BLOC"...], 512, 367'616) → 512 ENOENT
SYS  13477  13477        892'885'417 pread(4, [u"              );◙        } else {◙      "...], 512, 368'128) → 512 ENOENT
SYS  13477  13477        892'896'079 pread(3, [u"              );◙        } else {◙      "...], 512, 368'128) → 512 ENOENT
SYS  13477  13477        892'899'179 pread(4, [u"attn_factor, corr_dims,◙                "...], 512, 368'640) → 512 ENOENT
SYS  13477  13477        892'902'192 pread(3, [u"attn_factor, corr_dims,◙                "...], 512, 368'640) → 512 ENOENT
SYS  13477  13477        892'904'612 pread(4, [u" int p_delta_rows,◙    float freq_base, "...], 512, 369'152) → 512 ENOENT
SYS  13477  13477        892'907'271 pread(3, [u" int p_delta_rows,◙    float freq_base, "...], 512, 369'152) → 512 ENOENT
SYS  13477  13477        892'909'656 pread(4, [u"st int32_t * pos, float freq_scale, int "...], 512, 369'664) → 512 ENOENT
SYS  13477  13477        892'912'412 pread(3, [u"st int32_t * pos, float freq_scale, int "...], 512, 369'664) → 512 ENOENT
SYS  13477  13477        892'915'000 pread(4, [u"dims, cudaStream_t stream) {◙◙    rope_c"...], 512, 370'176) → 512 ENOENT
SYS  13477  13477        892'919'434 pread(3, [u"dims, cudaStream_t stream) {◙◙    rope_c"...], 512, 370'176) → 512 ENOENT
SYS  13477  13477        892'929'080 pread(4, [u", freq_scale, p_delta_rows, freq_base, e"...], 512, 370'688) → 512 ENOENT
SYS  13477  13477        892'933'576 pread(3, [u", freq_scale, p_delta_rows, freq_base, e"...], 512, 370'688) → 512 ENOENT
SYS  13477  13477        892'937'596 pread(4, [u"ctor, corr_dims, freq_factors, stream);◙"...], 512, 371'200) → 512 ENOENT
SYS  13477  13477        892'943'222 pread(3, [u"ctor, corr_dims, freq_factors, stream);◙"...], 512, 371'200) → 512 ENOENT
SYS  13477  13477        892'946'756 pread(4, [u"E_F16);◙    GGML_ASSERT( dst->type == GG"...], 512, 371'712) → 512 ENOENT
SYS  13477  13477        892'950'165 pread(3, [u"E_F16);◙    GGML_ASSERT( dst->type == GG"...], 512, 371'712) → 512 ENOENT
SYS  13477  13477        892'952'930 pread(4, [u"rig_ctx  = ((int32_t *) dst->op_params)["...], 512, 372'224) → 512 ENOENT
SYS  13477  13477        892'955'695 pread(3, [u"rig_ctx  = ((int32_t *) dst->op_params)["...], 512, 372'224) → 512 ENOENT
SYS  13477  13477        892'957'806 pread(4, [u"rams +  9, sizeof(float));◙    memcpy(&b"...], 512, 372'736) → 512 ENOENT
SYS  13477  13477        892'962'921 pread(3, [u"rams +  9, sizeof(float));◙    memcpy(&b"...], 512, 372'736) → 512 ENOENT
SYS  13477  13477        892'965'129 pread(4, [u"◙◙    rope_corr_dims corr_dims;◙    ggml"...], 512, 373'248) → 512 ENOENT
SYS  13477  13477        892'967'885 pread(3, [u"◙◙    rope_corr_dims corr_dims;◙    ggml"...], 512, 373'248) → 512 ENOENT
SYS  13477  13477        892'970'049 pread(4, [u"tor,◙                attn_factor, corr_d"...], 512, 373'760) → 512 ENOENT
SYS  13477  13477        892'974'325 pread(3, [u"tor,◙                attn_factor, corr_d"...], 512, 373'760) → 512 ENOENT
SYS  13477  13477        892'976'462 pread(4, [u"     (const float *)src0_d, (float *)dst"...], 512, 374'272) → 512 ENOENT
SYS  13477  13477        892'979'086 pread(3, [u"     (const float *)src0_d, (float *)dst"...], 512, 374'272) → 512 ENOENT
SYS  13477  13477        892'983'732 pread(4, [u"////////////////////////////////////////"...], 512, 374'784) → 512 ENOENT
SYS  13477  13477        892'986'426 pread(3, [u"////////////////////////////////////////"...], 512, 374'784) → 512 ENOENT
SYS  13477  13477        892'988'705 pread(4, [u", float * dst, const float scale, const "...], 512, 375'296) → 512 ENOENT
SYS  13477  13477        892'991'364 pread(3, [u", float * dst, const float scale, const "...], 512, 375'296) → 512 ENOENT
SYS  13477  13477        892'993'590 pread(4, [u"x, ggml_tensor * dst) {◙    const ggml_t"...], 512, 375'808) → 512 ENOENT
SYS  13477  13477        892'999'031 pread(3, [u"x, ggml_tensor * dst) {◙    const ggml_t"...], 512, 375'808) → 512 ENOENT
SYS  13477  13477        893'001'187 pread(4, [u"////////◙//◙// ROLLUP softmax.cu◙//◙////"...], 512, 376'320) → 512 ENOENT
SYS  13477  13477        893'003'881 pread(3, [u"////////◙//◙// ROLLUP softmax.cu◙//◙////"...], 512, 376'320) → 512 ENOENT
SYS  13477  13477        893'006'072 pread(4, [u"return (float) val;◙}◙◙template <>◙__dev"...], 512, 376'832) → 512 ENOENT
SYS  13477  13477        893'008'934 pread(3, [u"return (float) val;◙}◙◙template <>◙__dev"...], 512, 376'832) → 512 ENOENT
SYS  13477  13477        893'015'629 pread(4, [u"t tid  = threadIdx.x;◙    const int rowx"...], 512, 377'344) → 512 ENOENT
SYS  13477  13477        893'018'474 pread(3, [u"t tid  = threadIdx.x;◙    const int rowx"...], 512, 377'344) → 512 ENOENT
SYS  13477  13477        893'022'784 pread(4, [u"mory buffer for inter-warp communication"...], 512, 377'856) → 512 ENOENT
SYS  13477  13477        893'025'461 pread(3, [u"mory buffer for inter-warp communication"...], 512, 377'856) → 512 ENOENT
SYS  13477  13477        893'029'657 pread(4, [u"col;◙◙        const float val = x[ix]*sc"...], 512, 378'368) → 512 ENOENT
SYS  13477  13477        893'035'354 pread(3, [u"col;◙◙        const float val = x[ix]*sc"...], 512, 378'368) → 512 ENOENT
SYS  13477  13477        893'040'009 pread(4, [u"       max_val = warp_reduce_max(max_val"...], 512, 378'880) → 512 ENOENT
SYS  13477  13477        893'042'721 pread(3, [u"       max_val = warp_reduce_max(max_val"...], 512, 378'880) → 512 ENOENT
SYS  13477  13477        893'045'106 pread(4, [u"s();◙        if (warp_id == 0) {◙       "...], 512, 379'392) → 512 ENOENT
SYS  13477  13477        893'055'344 pread(3, [u"s();◙        if (warp_id == 0) {◙       "...], 512, 379'392) → 512 ENOENT
SYS  13477  13477        893'057'649 pread(4, [u"      const int64_t idst = (int64_t)rowx"...], 512, 379'904) → 512 ENOENT
SYS  13477  13477        893'060'317 pread(3, [u"      const int64_t idst = (int64_t)rowx"...], 512, 379'904) → 512 ENOENT
SYS  13477  13477        893'062'419 pread(4, [u"ize_t shmem = (GGML_PAD(ncols_x, WARP_SI"...], 512, 380'416) → 512 ENOENT
SYS  13477  13477        893'065'750 pread(3, [u"ize_t shmem = (GGML_PAD(ncols_x, WARP_SI"...], 512, 380'416) → 512 ENOENT
SYS  13477  13477        893'067'817 pread(4, [u"witch (ncols_x) {◙            case 32:◙ "...], 512, 380'928) → 512 ENOENT
SYS  13477  13477        893'070'467 pread(3, [u"witch (ncols_x) {◙            case 32:◙ "...], 512, 380'928) → 512 ENOENT
SYS  13477  13477        893'072'657 pread(4, [u" shmem, stream>>>(x, mask, dst, ncols_x,"...], 512, 381'440) → 512 ENOENT
SYS  13477  13477        893'075'237 pread(3, [u" shmem, stream>>>(x, mask, dst, ncols_x,"...], 512, 381'440) → 512 ENOENT
SYS  13477  13477        893'077'321 pread(4, [u" break;◙            case 1024:◙         "...], 512, 381'952) → 512 ENOENT
SYS  13477  13477        893'079'494 pread(3, [u" break;◙            case 1024:◙         "...], 512, 381'952) → 512 ENOENT
SYS  13477  13477        893'081'217 pread(4, [u"dims, shmem, stream>>>(x, mask, dst, nco"...], 512, 382'464) → 512 ENOENT
SYS  13477  13477        893'083'275 pread(3, [u"dims, shmem, stream>>>(x, mask, dst, nco"...], 512, 382'464) → 512 ENOENT
SYS  13477  13477        893'087'418 pread(4, [u"ale, max_bias, m0, m1, n_head_log2);◙   "...], 512, 382'976) → 512 ENOENT
SYS  13477  13477        893'089'635 pread(3, [u"ale, max_bias, m0, m1, n_head_log2);◙   "...], 512, 382'976) → 512 ENOENT
SYS  13477  13477        893'091'410 pread(4, [u";◙◙    GGML_ASSERT(!src1 || src1->type ="...], 512, 383'488) → 512 ENOENT
SYS  13477  13477        893'094'749 pread(3, [u";◙◙    GGML_ASSERT(!src1 || src1->type ="...], 512, 383'488) → 512 ENOENT
SYS  13477  13477        893'097'514 pread(4, [u"6);◙◙    if (use_f16) {◙        const ha"...], 512, 384'000) → 512 ENOENT
SYS  13477  13477        893'100'968 pread(3, [u"6);◙◙    if (use_f16) {◙        const ha"...], 512, 384'000) → 512 ENOENT
SYS  13477  13477        893'104'625 pread(4, [u"////////////////////////////////◙◙◙/////"...], 512, 384'512) → 512 ENOENT
SYS  13477  13477        893'106'842 pread(3, [u"////////////////////////////////◙◙◙/////"...], 512, 384'512) → 512 ENOENT
SYS  13477  13477        893'108'582 pread(4, [u"ls; i += blockDim.x) {◙        sum += x["...], 512, 385'024) → 512 ENOENT
SYS  13477  13477        893'113'467 pread(3, [u"ls; i += blockDim.x) {◙        sum += x["...], 512, 385'024) → 512 ENOENT
SYS  13477  13477        893'117'345 pread(4, [u"  const ggml_tensor * src0 = dst->src[0]"...], 512, 385'536) → 512 ENOENT
SYS  13477  13477        893'119'518 pread(3, [u"  const ggml_tensor * src0 = dst->src[0]"...], 512, 385'536) → 512 ENOENT
SYS  13477  13477        893'121'258 pread(4, [u"/////////////////////////////◙//◙// ROLL"...], 512, 386'048) → 512 ENOENT
SYS  13477  13477        893'125'348 pread(3, [u"/////////////////////////////◙//◙// ROLL"...], 512, 386'048) → 512 ENOENT
SYS  13477  13477        893'127'115 pread(4, [u"32(const float * timesteps, float * dst,"...], 512, 386'560) → 512 ENOENT
SYS  13477  13477        893'129'226 pread(3, [u"32(const float * timesteps, float * dst,"...], 512, 386'560) → 512 ENOENT
SYS  13477  13477        893'133'378 pread(4, [u"i];◙    float freq = (float)expf(-logf(m"...], 512, 387'072) → 512 ENOENT
SYS  13477  13477        893'137'406 pread(3, [u"i];◙    float freq = (float)expf(-logf(m"...], 512, 387'072) → 512 ENOENT
SYS  13477  13477        893'139'420 pread(4, [u"IZE;◙    dim3 gridDim(num_blocks, ne00, "...], 512, 387'584) → 512 ENOENT
SYS  13477  13477        893'142'140 pread(3, [u"IZE;◙    dim3 gridDim(num_blocks, ne00, "...], 512, 387'584) → 512 ENOENT
SYS  13477  13477        893'143'792 pread(4, [u"ML_TYPE_F32);◙◙    const int dim = dst->"...], 512, 388'096) → 512 ENOENT
SYS  13477  13477        893'145'974 pread(3, [u"ML_TYPE_F32);◙◙    const int dim = dst->"...], 512, 388'096) → 512 ENOENT
SYS  13477  13477        893'150'064 pread(4, [u"////////////////////////////////////////"...], 512, 388'608) → 512 ENOENT
SYS  13477  13477        893'152'696 pread(3, [u"////////////////////////////////////////"...], 512, 388'608) → 512 ENOENT
SYS  13477  13477        893'154'569 pread(4, [u"cuda_op_gelu_quick(ggml_backend_cuda_con"...], 512, 389'120) → 512 ENOENT
SYS  13477  13477        893'157'334 pread(3, [u"cuda_op_gelu_quick(ggml_backend_cuda_con"...], 512, 389'120) → 512 ENOENT
SYS  13477  13477        893'159'427 pread(4, [u"nd_cuda_context & ctx, ggml_tensor * dst"...], 512, 389'632) → 512 ENOENT
SYS  13477  13477        893'162'793 pread(3, [u"nd_cuda_context & ctx, ggml_tensor * dst"...], 512, 389'632) → 512 ENOENT
SYS  13477  13477        893'166'432 pread(4, [u"}◙◙static __global__ void gelu_quick_f32"...], 512, 390'144) → 512 ENOENT
SYS  13477  13477        893'168'711 pread(3, [u"}◙◙static __global__ void gelu_quick_f32"...], 512, 390'144) → 512 ENOENT
SYS  13477  13477        893'170'389 pread(4, [u"ic __global__ void tanh_f32(const float "...], 512, 390'656) → 512 ENOENT
SYS  13477  13477        893'174'726 pread(3, [u"ic __global__ void tanh_f32(const float "...], 512, 390'656) → 512 ENOENT
SYS  13477  13477        893'176'679 pread(4, [u"dx.x + threadIdx.x;◙◙    if (i >= k) {◙ "...], 512, 391'168) → 512 ENOENT
SYS  13477  13477        893'180'919 pread(3, [u"dx.x + threadIdx.x;◙◙    if (i >= k) {◙ "...], 512, 391'168) → 512 ENOENT
SYS  13477  13477        893'184'567 pread(4, [u"  return;◙    }◙    dst[i] = x[i] * fmin"...], 512, 391'680) → 512 ENOENT
SYS  13477  13477        893'188'021 pread(3, [u"  return;◙    }◙    dst[i] = x[i] * fmin"...], 512, 391'680) → 512 ENOENT
SYS  13477  13477        893'191'527 pread(4, [u"◙        return;◙    }◙    dst[i] = x[i]"...], 512, 392'192) → 512 ENOENT
SYS  13477  13477        893'193'745 pread(3, [u"◙        return;◙    }◙    dst[i] = x[i]"...], 512, 392'192) → 512 ENOENT
SYS  13477  13477        893'197'128 pread(4, [u"ks, CUDA_GELU_BLOCK_SIZE, 0, stream>>>(x"...], 512, 392'704) → 512 ENOENT
SYS  13477  13477        893'199'416 pread(3, [u"ks, CUDA_GELU_BLOCK_SIZE, 0, stream>>>(x"...], 512, 392'704) → 512 ENOENT
SYS  13477  13477        893'201'112 pread(4, [u"ANH_BLOCK_SIZE, 0, stream>>>(x, dst, k);"...], 512, 393'216) → 512 ENOENT
SYS  13477  13477        893'203'382 pread(3, [u"ANH_BLOCK_SIZE, 0, stream>>>(x, dst, k);"...], 512, 393'216) → 512 ENOENT
SYS  13477  13477        893'205'104 pread(4, [u"_SIGMOID_BLOCK_SIZE, 0, stream>>>(x, dst"...], 512, 393'728) → 512 ENOENT
SYS  13477  13477        893'209'486 pread(3, [u"_SIGMOID_BLOCK_SIZE, 0, stream>>>(x, dst"...], 512, 393'728) → 512 ENOENT
SYS  13477  13477        893'211'226 pread(4, [u"BLOCK_SIZE;◙    hardswish_f32<<<num_bloc"...], 512, 394'240) → 512 ENOENT
SYS  13477  13477        893'214'008 pread(3, [u"BLOCK_SIZE;◙    hardswish_f32<<<num_bloc"...], 512, 394'240) → 512 ENOENT
SYS  13477  13477        893'216'261 pread(4, [u"um_blocks = (k + CUDA_SQR_BLOCK_SIZE - 1"...], 512, 394'752) → 512 ENOENT
SYS  13477  13477        893'218'478 pread(3, [u"um_blocks = (k + CUDA_SQR_BLOCK_SIZE - 1"...], 512, 394'752) → 512 ENOENT
SYS  13477  13477        893'220'209 pread(4, [u"d, dst_d, ggml_nelements(src0), stream);"...], 512, 395'264) → 512 ENOENT
SYS  13477  13477        893'224'891 pread(3, [u"d, dst_d, ggml_nelements(src0), stream);"...], 512, 395'264) → 512 ENOENT
SYS  13477  13477        893'227'082 pread(4, [u"xt & ctx, ggml_tensor * dst) {◙    const"...], 512, 395'776) → 512 ENOENT
SYS  13477  13477        893'229'564 pread(3, [u"xt & ctx, ggml_tensor * dst) {◙    const"...], 512, 395'776) → 512 ENOENT
SYS  13477  13477        893'232'965 pread(4, [u" * src0_d = (const float *)src0->data;◙ "...], 512, 396'288) → 512 ENOENT
SYS  13477  13477        893'235'155 pread(3, [u" * src0_d = (const float *)src0->data;◙ "...], 512, 396'288) → 512 ENOENT
SYS  13477  13477        893'236'904 pread(4, [u"stream = ctx.stream();◙◙    GGML_ASSERT("...], 512, 396'800) → 512 ENOENT
SYS  13477  13477        893'239'086 pread(3, [u"stream = ctx.stream();◙◙    GGML_ASSERT("...], 512, 396'800) → 512 ENOENT
SYS  13477  13477        893'240'817 pread(4, [u"ype == GGML_TYPE_F32);◙◙    sigmoid_f32_"...], 512, 397'312) → 512 ENOENT
SYS  13477  13477        893'246'656 pread(3, [u"ype == GGML_TYPE_F32);◙◙    sigmoid_f32_"...], 512, 397'312) → 512 ENOENT
SYS  13477  13477        893'249'448 pread(4, [u"stream);◙}◙◙void ggml_cuda_op_hardswish("...], 512, 397'824) → 512 ENOENT
SYS  13477  13477        893'252'071 pread(3, [u"stream);◙}◙◙void ggml_cuda_op_hardswish("...], 512, 397'824) → 512 ENOENT
SYS  13477  13477        893'255'604 pread(4, [u"* dst) {◙    const ggml_tensor * src0 = "...], 512, 398'336) → 512 ENOENT
SYS  13477  13477        893'257'857 pread(3, [u"* dst) {◙    const ggml_tensor * src0 = "...], 512, 398'336) → 512 ENOENT
SYS  13477  13477        893'259'579 pread(4, [u"ggml_tensor * dst) {◙    const ggml_tens"...], 512, 398'848) → 512 ENOENT
SYS  13477  13477        893'265'127 pread(3, [u"ggml_tensor * dst) {◙    const ggml_tens"...], 512, 398'848) → 512 ENOENT
SYS  13477  13477        893'268'157 pread(4, [u"///////////////////////////////////◙◙◙//"...], 512, 399'360) → 512 ENOENT
SYS  13477  13477        893'272'706 pread(3, [u"///////////////////////////////////◙◙◙//"...], 512, 399'360) → 512 ENOENT
SYS  13477  13477        893'276'539 pread(4, [u"t int ne11, const int ne12, const int ne"...], 512, 399'872) → 512 ENOENT
SYS  13477  13477        893'278'951 pread(3, [u"t int ne11, const int ne12, const int ne"...], 512, 399'872) → 512 ENOENT
SYS  13477  13477        893'282'608 pread(4, [u"    dst[index] = *(float *)((char *)x + "...], 512, 400'384) → 512 ENOENT
SYS  13477  13477        893'285'390 pread(3, [u"    dst[index] = *(float *)((char *)x + "...], 512, 400'384) → 512 ENOENT
SYS  13477  13477        893'289'153 pread(4, [u" / CUDA_UPSCALE_BLOCK_SIZE;◙◙    upscale"...], 512, 400'896) → 512 ENOENT
SYS  13477  13477        893'291'335 pread(3, [u" / CUDA_UPSCALE_BLOCK_SIZE;◙◙    upscale"...], 512, 400'896) → 512 ENOENT
SYS  13477  13477        893'293'526 pread(4, [u"ML_TYPE_F32);◙◙    const float sf0 = (fl"...], 512, 401'408) → 512 ENOENT
SYS  13477  13477        893'298'331 pread(3, [u"ML_TYPE_F32);◙◙    const float sf0 = (fl"...], 512, 401'408) → 512 ENOENT
SYS  13477  13477        893'302'342 pread(4, [u"////////////////////////////////////////"...], 512, 401'920) → 512 ENOENT
SYS  13477  13477        893'306'467 pread(3, [u"////////////////////////////////////////"...], 512, 401'920) → 512 ENOENT
SYS  13477  13477        893'310'053 pread(4, [u"_cuda_default_log_callback;◙void * ggml_"...], 512, 402'432) → 512 ENOENT
SYS  13477  13477        893'313'719 pread(3, [u"_cuda_default_log_callback;◙void * ggml_"...], 512, 402'432) → 512 ENOENT
SYS  13477  13477        893'317'438 pread(4, [u"E_FORMAT(2, 3)◙static void ggml_cuda_log"...], 512, 402'944) → 512 ENOENT
SYS  13477  13477        893'319'628 pread(3, [u"E_FORMAT(2, 3)◙static void ggml_cuda_log"...], 512, 402'944) → 512 ENOENT
SYS  13477  13477        893'321'360 pread(4, [u"   va_start(args, format);◙            v"...], 512, 403'456) → 512 ENOENT
SYS  13477  13477        893'326'015 pread(3, [u"   va_start(args, format);◙            v"...], 512, 403'456) → 512 ENOENT
SYS  13477  13477        893'327'826 pread(4, [u"function %s at %s:%d\\n“, id, func, file,"...], 512, 403'968) → 512 ENOENT
SYS  13477  13477        893'330'697 pread(3, [u"function %s at %s:%d\\n“, id, func, file,"...], 512, 403'968) → 512 ENOENT
SYS  13477  13477        893'332'931 pread(4, [u"));◙}◙◙int ggml_cuda_get_device() {◙    "...], 512, 404'480) → 512 ENOENT
SYS  13477  13477        893'338'302 pread(3, [u"));◙}◙◙int ggml_cuda_get_device() {◙    "...], 512, 404'480) → 512 ENOENT
SYS  13477  13477        893'341'544 pread(4, [u"r, size);◙        if (res == hipSuccess)"...], 512, 404'992) → 512 ENOENT
SYS  13477  13477        893'345'863 pread(3, [u"r, size);◙        if (res == hipSuccess)"...], 512, 404'992) → 512 ENOENT
SYS  13477  13477        893'352'859 pread(4, [u"/ROCmSoftwarePlatform/rocBLAS/issues/134"...], 512, 405'504) → 512 ENOENT
SYS  13477  13477        893'358'469 pread(3, [u"/ROCmSoftwarePlatform/rocBLAS/issues/134"...], 512, 405'504) → 512 ENOENT
SYS  13477  13477        893'362'064 pread(4, [u"   int64_t total_vram = 0;◙#if defined(G"...], 512, 406'016) → 512 ENOENT
SYS  13477  13477        893'364'263 pread(3, [u"   int64_t total_vram = 0;◙#if defined(G"...], 512, 406'016) → 512 ENOENT
SYS  13477  13477        893'366'569 pread(4, [u"d = 0; id < info.device_count; ++id) {◙ "...], 512, 406'528) → 512 ENOENT
SYS  13477  13477        893'371'012 pread(3, [u"d = 0; id < info.device_count; ++id) {◙ "...], 512, 406'528) → 512 ENOENT
SYS  13477  13477        893'374'890 pread(4, [u"_DEVICE;◙            alloc_prop.location"...], 512, 407'040) → 512 ENOENT
SYS  13477  13477        893'378'821 pread(3, [u"_DEVICE;◙            alloc_prop.location"...], 512, 407'040) → 512 ENOENT
SYS  13477  13477        893'380'985 pread(4, [u"o“);◙◙        info.default_tensor_split["...], 512, 407'552) → 512 ENOENT
SYS  13477  13477        893'383'564 pread(3, [u"o“);◙◙        info.default_tensor_split["...], 512, 407'552) → 512 ENOENT
SYS  13477  13477        893'387'177 pread(4, [u"t id = 0; id < info.device_count; ++id) "...], 512, 408'064) → 512 ENOENT
SYS  13477  13477        893'389'544 pread(3, [u"t id = 0; id < info.device_count; ++id) "...], 512, 408'064) → 512 ENOENT
SYS  13477  13477        893'391'682 pread(4, [u"evice;◙    struct ggml_cuda_buffer {◙   "...], 512, 408'576) → 512 ENOENT
SYS  13477  13477        893'395'869 pread(3, [u"evice;◙    struct ggml_cuda_buffer {◙   "...], 512, 408'576) → 512 ENOENT
SYS  13477  13477        893'401'664 pread(4, [u"       pool_size -= b.size;◙            "...], 512, 409'088) → 512 ENOENT
SYS  13477  13477        893'404'517 pread(3, [u"       pool_size -= b.size;◙            "...], 512, 409'088) → 512 ENOENT
SYS  13477  13477        893'408'130 pread(4, [u" > max_size) max_size = b.size;◙#endif◙ "...], 512, 409'600) → 512 ENOENT
SYS  13477  13477        893'411'292 pread(3, [u" > max_size) max_size = b.size;◙#endif◙ "...], 512, 409'600) → 512 ENOENT
SYS  13477  13477        893'413'014 pread(4, [u"◙                        }◙             "...], 512, 410'112) → 512 ENOENT
SYS  13477  13477        893'418'579 pread(3, [u"◙                        }◙             "...], 512, 410'112) → 512 ENOENT
SYS  13477  13477        893'421'821 pread(4, [u"UDA_CHECK(ggml_cuda_device_malloc(&ptr, "...], 512, 410'624) → 512 ENOENT
SYS  13477  13477        893'426'521 pread(3, [u"UDA_CHECK(ggml_cuda_device_malloc(&ptr, "...], 512, 410'624) → 512 ENOENT
SYS  13477  13477        893'430'504 pread(4, [u" {◙        for (int i = 0; i < MAX_BUFFE"...], 512, 411'136) → 512 ENOENT
SYS  13477  13477        893'434'806 pread(3, [u" {◙        for (int i = 0; i < MAX_BUFFE"...], 512, 411'136) → 512 ENOENT
SYS  13477  13477        893'438'401 pread(4, [u"_CUDA_NO_VMM)◙struct ggml_cuda_pool_vmm "...], 512, 411'648) → 512 ENOENT
SYS  13477  13477        893'440'654 pread(3, [u"_CUDA_NO_VMM)◙struct ggml_cuda_pool_vmm "...], 512, 411'648) → 512 ENOENT
SYS  13477  13477        893'442'474 pread(4, [u"ool_size));◙            CU_CHECK(cuMemAd"...], 512, 412'160) → 512 ENOENT
SYS  13477  13477        893'444'788 pread(3, [u"ool_size));◙            CU_CHECK(cuMemAd"...], 512, 412'160) → 512 ENOENT
SYS  13477  13477        893'448'365 pread(4, [u" the granularity◙            size_t rese"...], 512, 412'672) → 512 ENOENT
SYS  13477  13477        893'450'662 pread(3, [u" the granularity◙            size_t rese"...], 512, 412'672) → 512 ENOENT
SYS  13477  13477        893'452'455 pread(4, [u"nHandle handle;◙            CU_CHECK(cuM"...], 512, 413'184) → 512 ENOENT
SYS  13477  13477        893'457'941 pread(3, [u"nHandle handle;◙            CU_CHECK(cuM"...], 512, 413'184) → 512 ENOENT
SYS  13477  13477        893'462'516 pread(4, [u" CU_CHECK(cuMemRelease(handle));◙◙      "...], 512, 413'696) → 512 ENOENT
SYS  13477  13477        893'466'624 pread(3, [u" CU_CHECK(cuMemRelease(handle));◙◙      "...], 512, 413'696) → 512 ENOENT
SYS  13477  13477        893'468'461 pread(4, [u",◙            //       device, (unsigned"...], 512, 414'208) → 512 ENOENT
SYS  13477  13477        893'470'670 pread(3, [u",◙            //       device, (unsigned"...], 512, 414'208) → 512 ENOENT
SYS  13477  13477        893'474'530 pread(4, [u"rride {◙#ifdef DEBUG_CUDA_MALLOC◙       "...], 512, 414'720) → 512 ENOENT
SYS  13477  13477        893'478'717 pread(3, [u"rride {◙#ifdef DEBUG_CUDA_MALLOC◙       "...], 512, 414'720) → 512 ENOENT
SYS  13477  13477        893'482'321 pread(4, [u" (ggml_cuda_info().devices[device].vmm) "...], 512, 415'232) → 512 ENOENT
SYS  13477  13477        893'484'564 pread(3, [u" (ggml_cuda_info().devices[device].vmm) "...], 512, 415'232) → 512 ENOENT
SYS  13477  13477        893'486'322 pread(4, [u"◙    ~ggml_backend_cuda_buffer_context()"...], 512, 415'744) → 512 ENOENT
SYS  13477  13477        893'491'887 pread(3, [u"◙    ~ggml_backend_cuda_buffer_context()"...], 512, 415'744) → 512 ENOENT
SYS  13477  13477        893'495'703 pread(4, [u"uda_buffer_free_buffer(ggml_backend_buff"...], 512, 416'256) → 512 ENOENT
SYS  13477  13477        893'498'450 pread(3, [u"uda_buffer_free_buffer(ggml_backend_buff"...], 512, 416'256) → 512 ENOENT
SYS  13477  13477        893'502'346 pread(4, [u"ckend_cuda_buffer_context * ctx = (ggml_"...], 512, 416'768) → 512 ENOENT
SYS  13477  13477        893'506'559 pread(3, [u"ckend_cuda_buffer_context * ctx = (ggml_"...], 512, 416'768) → 512 ENOENT
SYS  13477  13477        893'509'845 pread(4, [u"src == nullptr) {◙            ggml_cuda_"...], 512, 417'280) → 512 ENOENT
SYS  13477  13477        893'512'548 pread(3, [u"src == nullptr) {◙            ggml_cuda_"...], 512, 417'280) → 512 ENOENT
SYS  13477  13477        893'516'214 pread(4, [u"Async((char *)tensor->data + offset, dat"...], 512, 417'792) → 512 ENOENT
SYS  13477  13477        893'518'582 pread(3, [u"Async((char *)tensor->data + offset, dat"...], 512, 417'792) → 512 ENOENT
SYS  13477  13477        893'520'410 pread(4, [u"r->data + offset, size, cudaMemcpyDevice"...], 512, 418'304) → 512 ENOENT
SYS  13477  13477        893'524'685 pread(3, [u"r->data + offset, size, cudaMemcpyDevice"...], 512, 418'304) → 512 ENOENT
SYS  13477  13477        893'530'356 pread(4, [u"ntext *)dst->buffer->context;◙        if"...], 512, 418'816) → 512 ENOENT
SYS  13477  13477        893'533'254 pread(3, [u"ntext *)dst->buffer->context;◙        if"...], 512, 418'816) → 512 ENOENT
SYS  13477  13477        893'536'920 pread(4, [u"    return true;◙    }◙    return false;"...], 512, 419'328) → 512 ENOENT
SYS  13477  13477        893'539'181 pread(3, [u"    return true;◙    }◙    return false;"...], 512, 419'328) → 512 ENOENT
SYS  13477  13477        893'541'009 pread(4, [u"terface = {◙    /* .get_name        = */"...], 512, 419'840) → 512 ENOENT
SYS  13477  13477        893'546'415 pread(3, [u"terface = {◙    /* .get_name        = */"...], 512, 419'840) → 512 ENOENT
SYS  13477  13477        893'550'249 pread(4, [u"d_cuda_buffer_clear,◙    /* .reset      "...], 512, 420'352) → 512 ENOENT
SYS  13477  13477        893'554'577 pread(3, [u"d_cuda_buffer_clear,◙    /* .reset      "...], 512, 420'352) → 512 ENOENT
SYS  13477  13477        893'556'680 pread(4, [u"ype_t buft, size_t size) {◙    ggml_back"...], 512, 420'864) → 512 ENOENT
SYS  13477  13477        893'559'356 pread(3, [u"ype_t buft, size_t size) {◙    ggml_back"...], 512, 420'864) → 512 ENOENT
SYS  13477  13477        893'563'022 pread(4, [u": cudaMalloc failed: %s\\n“, __func__, si"...], 512, 421'376) → 512 ENOENT
SYS  13477  13477        893'565'381 pread(3, [u": cudaMalloc failed: %s\\n“, __func__, si"...], 512, 421'376) → 512 ENOENT
SYS  13477  13477        893'567'183 pread(4, [u"gml_backend_cuda_buffer_type_get_alloc_s"...], 512, 421'888) → 512 ENOENT
SYS  13477  13477        893'574'567 pread(3, [u"gml_backend_cuda_buffer_type_get_alloc_s"...], 512, 421'888) → 512 ENOENT
SYS  13477  13477        893'576'343 pread(4, [u"r_type_t buft, ggml_backend_t backend) {"...], 512, 422'400) → 512 ENOENT
SYS  13477  13477        893'580'362 pread(3, [u"r_type_t buft, ggml_backend_t backend) {"...], 512, 422'400) → 512 ENOENT
SYS  13477  13477        893'582'199 pread(4, [u"  /* .alloc_buffer     = */ ggml_backend"...], 512, 422'912) → 512 ENOENT
SYS  13477  13477        893'584'363 pread(3, [u"  /* .alloc_buffer     = */ ggml_backend"...], 512, 422'912) → 512 ENOENT
SYS  13477  13477        893'588'277 pread(4, [u"   std::lock_guard<std::mutex> lock(mute"...], 512, 423'424) → 512 ENOENT
SYS  13477  13477        893'592'181 pread(3, [u"   std::lock_guard<std::mutex> lock(mute"...], 512, 423'424) → 512 ENOENT
SYS  13477  13477        893'595'069 pread(4, [u"er_type_interface,◙                /* .c"...], 512, 423'936) → 512 ENOENT
SYS  13477  13477        893'600'670 pread(3, [u"er_type_interface,◙                /* .c"...], 512, 423'936) → 512 ENOENT
SYS  13477  13477        893'603'920 pread(4, [u"or (int id = 0; id < ggml_backend_cuda_g"...], 512, 424'448) → 512 ENOENT
SYS  13477  13477        893'609'980 pread(3, [u"or (int id = 0; id < ggml_backend_cuda_g"...], 512, 424'448) → 512 ENOENT
SYS  13477  13477        893'613'743 pread(4, [u"   }◙    }◙◙#if defined(GGML_USE_HIPBLAS"...], 512, 424'960) → 512 ENOENT
SYS  13477  13477        893'616'579 pread(3, [u"   }◙    }◙◙#if defined(GGML_USE_HIPBLAS"...], 512, 424'960) → 512 ENOENT
SYS  13477  13477        893'620'324 pread(4, [u"       return min_compute_capability < C"...], 512, 425'472) → 512 ENOENT
SYS  13477  13477        893'624'617 pread(3, [u"       return min_compute_capability < C"...], 512, 425'472) → 512 ENOENT
SYS  13477  13477        893'628'044 pread(4, [u"       GGML_ASSERT(false);◙    }◙#else◙ "...], 512, 425'984) → 512 ENOENT
SYS  13477  13477        893'632'399 pread(3, [u"       GGML_ASSERT(false);◙    }◙#else◙ "...], 512, 425'984) → 512 ENOENT
SYS  13477  13477        893'634'157 pread(4, [u"_XXS:◙        case GGML_TYPE_IQ2_XS:◙   "...], 512, 426'496) → 512 ENOENT
SYS  13477  13477        893'638'273 pread(3, [u"_XXS:◙        case GGML_TYPE_IQ2_XS:◙   "...], 512, 426'496) → 512 ENOENT
SYS  13477  13477        893'640'093 pread(4, [u"row_split(int64_t * row_low, int64_t * r"...], 512, 427'008) → 512 ENOENT
SYS  13477  13477        893'642'257 pread(3, [u"row_split(int64_t * row_low, int64_t * r"...], 512, 427'008) → 512 ENOENT
SYS  13477  13477        893'647'504 pread(4, [u"     *row_high -= *row_high % rounding;◙"...], 512, 427'520) → 512 ENOENT
SYS  13477  13477        893'651'444 pread(3, [u"     *row_high -= *row_high % rounding;◙"...], 512, 427'520) → 512 ENOENT
SYS  13477  13477        893'655'074 pread(4, [u"  for (ggml_tensor_extra_gpu * extra : t"...], 512, 428'032) → 512 ENOENT
SYS  13477  13477        893'659'438 pread(3, [u"  for (ggml_tensor_extra_gpu * extra : t"...], 512, 428'032) → 512 ENOENT
SYS  13477  13477        893'663'634 pread(4, [u"            }◙            delete extra;◙"...], 512, 428'544) → 512 ENOENT
SYS  13477  13477        893'667'883 pread(3, [u"            }◙            delete extra;◙"...], 512, 428'544) → 512 ENOENT
SYS  13477  13477        893'671'257 pread(4, [u"g builds currently, avoid unused functio"...], 512, 429'056) → 512 ENOENT
SYS  13477  13477        893'675'674 pread(3, [u"g builds currently, avoid unused functio"...], 512, 429'056) → 512 ENOENT
SYS  13477  13477        893'681'265 pread(4, [u"id *)0x1000;◙◙    GGML_UNUSED(buffer);◙}"...], 512, 429'568) → 512 ENOENT
SYS  13477  13477        893'685'655 pread(3, [u"id *)0x1000;◙◙    GGML_UNUSED(buffer);◙}"...], 512, 429'568) → 512 ENOENT
SYS  13477  13477        893'689'383 pread(4, [u"ne0 = tensor->ne[0];◙◙    ggml_tensor_ex"...], 512, 430'080) → 512 ENOENT
SYS  13477  13477        893'691'680 pread(3, [u"ne0 = tensor->ne[0];◙◙    ggml_tensor_ex"...], 512, 430'080) → 512 ENOENT
SYS  13477  13477        893'693'446 pread(4, [u"_t original_size = size;◙◙        // pad"...], 512, 430'592) → 512 ENOENT
SYS  13477  13477        893'697'739 pread(3, [u"_t original_size = size;◙◙        // pad"...], 512, 430'592) → 512 ENOENT
SYS  13477  13477        893'699'497 pread(4, [u"id**)&buf, size, id));◙◙        // set p"...], 512, 431'104) → 512 ENOENT
SYS  13477  13477        893'706'202 pread(3, [u"id**)&buf, size, id));◙◙        // set p"...], 512, 431'104) → 512 ENOENT
SYS  13477  13477        893'708'401 pread(4, [u"et_tensor(ggml_backend_buffer_t buffer, "...], 512, 431'616) → 512 ENOENT
SYS  13477  13477        893'712'994 pread(3, [u"et_tensor(ggml_backend_buffer_t buffer, "...], 512, 431'616) → 512 ENOENT
SYS  13477  13477        893'717'234 pread(4, [u"extra_gpu *)tensor->extra;◙◙    for (int"...], 512, 432'128) → 512 ENOENT
SYS  13477  13477        893'719'990 pread(3, [u"extra_gpu *)tensor->extra;◙◙    for (int"...], 512, 432'128) → 512 ENOENT
SYS  13477  13477        893'723'656 pread(4, [u" a multiple of 512 elements to avoid out"...], 512, 432'640) → 512 ENOENT
SYS  13477  13477        893'726'032 pread(3, [u" a multiple of 512 elements to avoid out"...], 512, 432'640) → 512 ENOENT
SYS  13477  13477        893'727'746 pread(4, [u"CK(cudaStreamSynchronize(cudaStreamPerTh"...], 512, 433'152) → 512 ENOENT
SYS  13477  13477        893'732'657 pread(3, [u"CK(cudaStreamSynchronize(cudaStreamPerTh"...], 512, 433'152) → 512 ENOENT
SYS  13477  13477        893'738'063 pread(4, [u"_t ne0 = tensor->ne[0];◙    const size_t"...], 512, 433'664) → 512 ENOENT
SYS  13477  13477        893'741'606 pread(3, [u"_t ne0 = tensor->ne[0];◙    const size_t"...], 512, 433'664) → 512 ENOENT
SYS  13477  13477        893'746'561 pread(4, [u" ggml_nbytes_split(tensor, nrows_split);"...], 512, 434'176) → 512 ENOENT
SYS  13477  13477        893'749'220 pread(3, [u" ggml_nbytes_split(tensor, nrows_split);"...], 512, 434'176) → 512 ENOENT
SYS  13477  13477        893'750'995 pread(4, [u"));◙    }◙◙    for (int id = 0; id < ggm"...], 512, 434'688) → 512 ENOENT
SYS  13477  13477        893'753'239 pread(3, [u"));◙    }◙◙    for (int id = 0; id < ggm"...], 512, 434'688) → 512 ENOENT
SYS  13477  13477        893'754'988 pread(4, [u"split_buffer_free_buffer,◙    /* .get_ba"...], 512, 435'200) → 512 ENOENT
SYS  13477  13477        893'757'161 pread(3, [u"split_buffer_free_buffer,◙    /* .get_ba"...], 512, 435'200) → 512 ENOENT
SYS  13477  13477        893'758'839 pread(4, [u"gml_backend_cuda_split_buffer_type_name("...], 512, 435'712) → 512 ENOENT
SYS  13477  13477        893'763'795 pread(3, [u"gml_backend_cuda_split_buffer_type_name("...], 512, 435'712) → 512 ENOENT
SYS  13477  13477        893'765'738 pread(4, [u"m cumulative size of all the device buff"...], 512, 436'224) → 512 ENOENT
SYS  13477  13477        893'769'969 pread(3, [u"m cumulative size of all the device buff"...], 512, 436'224) → 512 ENOENT
SYS  13477  13477        893'772'946 pread(4, [u"eturn 128;◙◙    GGML_UNUSED(buft);◙}◙◙GG"...], 512, 436'736) → 512 ENOENT
SYS  13477  13477        893'777'230 pread(3, [u"eturn 128;◙◙    GGML_UNUSED(buft);◙}◙◙GG"...], 512, 436'736) → 512 ENOENT
SYS  13477  13477        893'781'815 pread(4, [u" tensor, ctx->tensor_split, id);◙◙      "...], 512, 437'248) → 512 ENOENT
SYS  13477  13477        893'786'258 pread(3, [u" tensor, ctx->tensor_split, id);◙◙      "...], 512, 437'248) → 512 ENOENT
SYS  13477  13477        893'790'922 pread(4, [u"bool ggml_backend_cuda_split_buffer_type"...], 512, 437'760) → 512 ENOENT
SYS  13477  13477        893'793'678 pread(3, [u"bool ggml_backend_cuda_split_buffer_type"...], 512, 437'760) → 512 ENOENT
SYS  13477  13477        893'796'787 pread(4, [u" = */ ggml_backend_cuda_split_buffer_typ"...], 512, 438'272) → 512 ENOENT
SYS  13477  13477        893'799'164 pread(3, [u" = */ ggml_backend_cuda_split_buffer_typ"...], 512, 438'272) → 512 ENOENT
SYS  13477  13477        893'803'783 pread(4, [u"pe(const float * tensor_split) {◙    sta"...], 512, 438'784) → 512 ENOENT
SYS  13477  13477        893'806'204 pread(3, [u"pe(const float * tensor_split) {◙    sta"...], 512, 438'784) → 512 ENOENT
SYS  13477  13477        893'811'265 pread(4, [u"it;◙    } else {◙        float split_sum"...], 512, 439'296) → 512 ENOENT
SYS  13477  13477        893'816'389 pread(3, [u"it;◙    } else {◙        float split_sum"...], 512, 439'296) → 512 ENOENT
SYS  13477  13477        893'820'567 pread(4, [u"        /* .iface   = */ ggml_backend_cu"...], 512, 439'808) → 512 ENOENT
SYS  13477  13477        893'825'893 pread(3, [u"        /* .iface   = */ ggml_backend_cu"...], 512, 439'808) → 512 ENOENT
SYS  13477  13477        893'827'678 pread(4, [u"ffer_name(ggml_backend_buffer_t buffer) "...], 512, 440'320) → 512 ENOENT
SYS  13477  13477        893'830'213 pread(3, [u"ffer_name(ggml_backend_buffer_t buffer) "...], 512, 440'320) → 512 ENOENT
SYS  13477  13477        893'831'909 pread(4, [u" clear the error◙        cudaGetLastErro"...], 512, 440'832) → 512 ENOENT
SYS  13477  13477        893'837'271 pread(3, [u" clear the error◙        cudaGetLastErro"...], 512, 440'832) → 512 ENOENT
SYS  13477  13477        893'839'276 pread(4, [u"     return ggml_backend_buft_alloc_buff"...], 512, 441'344) → 512 ENOENT
SYS  13477  13477        893'845'300 pread(3, [u"     return ggml_backend_buft_alloc_buff"...], 512, 441'344) → 512 ENOENT
SYS  13477  13477        893'850'803 pread(4, [u" {◙        /* .iface    = */ {◙         "...], 512, 441'856) → 512 ENOENT
SYS  13477  13477        893'856'386 pread(3, [u" {◙        /* .iface    = */ {◙         "...], 512, 441'856) → 512 ENOENT
SYS  13477  13477        893'859'919 pread(4, [u"pe()->iface.supports_backend,◙          "...], 512, 442'368) → 512 ENOENT
SYS  13477  13477        893'862'481 pread(3, [u"pe()->iface.supports_backend,◙          "...], 512, 442'368) → 512 ENOENT
SYS  13477  13477        893'865'970 pread(4, [u"nst ggml_tensor * src1, ggml_tensor * ds"...], 512, 442'880) → 512 ENOENT
SYS  13477  13477        893'868'240 pread(3, [u"nst ggml_tensor * src1, ggml_tensor * ds"...], 512, 442'880) → 512 ENOENT
SYS  13477  13477        893'871'747 pread(4, [u" const float * __restrict__ y, float * _"...], 512, 443'392) → 512 ENOENT
SYS  13477  13477        893'874'035 pread(3, [u" const float * __restrict__ y, float * _"...], 512, 443'392) → 512 ENOENT
SYS  13477  13477        893'876'270 pread(4, [u"ol_x0 = 0; col_x0 < ncols_x; col_x0 += b"...], 512, 443'904) → 512 ENOENT
SYS  13477  13477        893'881'791 pread(3, [u"ol_x0 = 0; col_x0 < ncols_x; col_x0 += b"...], 512, 443'904) → 512 ENOENT
SYS  13477  13477        893'887'559 pread(4, [u"sposed and not permuted◙    const int id"...], 512, 444'416) → 512 ENOENT
SYS  13477  13477        893'891'366 pread(3, [u"sposed and not permuted◙    const int id"...], 512, 444'416) → 512 ENOENT
SYS  13477  13477        893'893'486 pread(4, [u"nst half * x = (const half *) vx;◙◙    c"...], 512, 444'928) → 512 ENOENT
SYS  13477  13477        893'898'388 pread(3, [u"nst half * x = (const half *) vx;◙◙    c"...], 512, 444'928) → 512 ENOENT
SYS  13477  13477        893'903'706 pread(4, [u"       if (col_x >= ncols_x) {◙         "...], 512, 445'440) → 512 ENOENT
SYS  13477  13477        893'906'489 pread(3, [u"       if (col_x >= ncols_x) {◙         "...], 512, 445'440) → 512 ENOENT
SYS  13477  13477        893'910'543 pread(4, [u", const float * y, float * dst, const in"...], 512, 445'952) → 512 ENOENT
SYS  13477  13477        893'915'622 pread(3, [u", const float * y, float * dst, const in"...], 512, 445'952) → 512 ENOENT
SYS  13477  13477        893'918'034 pread(4, [u"_stride_x,◙    const int nchannels_x, co"...], 512, 446'464) → 512 ENOENT
SYS  13477  13477        893'921'806 pread(3, [u"_stride_x,◙    const int nchannels_x, co"...], 512, 446'464) → 512 ENOENT
SYS  13477  13477        893'927'000 pread(4, [u"_t i1_high, cudaStream_t stream) {◙◙    "...], 512, 446'976) → 512 ENOENT
SYS  13477  13477        893'932'785 pread(3, [u"_t i1_high, cudaStream_t stream) {◙◙    "...], 512, 446'976) → 512 ENOENT
SYS  13477  13477        893'937'308 pread(4, [u"gh - i1_low;◙◙    const char * x = src_p"...], 512, 447'488) → 512 ENOENT
SYS  13477  13477        893'942'317 pread(3, [u"gh - i1_low;◙◙    const char * x = src_p"...], 512, 447'488) → 512 ENOENT
SYS  13477  13477        893'944'074 pread(4, [u" (void *) (dst_ptr + i1*ts*ne0/bs);◙    "...], 512, 448'000) → 512 ENOENT
SYS  13477  13477        893'948'950 pread(3, [u" (void *) (dst_ptr + i1*ts*ne0/bs);◙    "...], 512, 448'000) → 512 ENOENT
SYS  13477  13477        893'954'445 pread(4, [u"oat * src1_ddf_i,◙    const char * src1_"...], 512, 448'512) → 512 ENOENT
SYS  13477  13477        893'960'292 pread(3, [u"oat * src1_ddf_i,◙    const char * src1_"...], 512, 448'512) → 512 ENOENT
SYS  13477  13477        893'963'835 pread(4, [u"t_device();◙◙    // the main device has "...], 512, 449'024) → 512 ENOENT
SYS  13477  13477        893'969'196 pread(3, [u"t_device();◙◙    // the main device has "...], 512, 449'024) → 512 ENOENT
SYS  13477  13477        893'973'021 pread(4, [u"rc1 to fp16, multiply as fp16, convert d"...], 512, 449'536) → 512 ENOENT
SYS  13477  13477        893'975'512 pread(3, [u"rc1 to fp16, multiply as fp16, convert d"...], 512, 449'536) → 512 ENOENT
SYS  13477  13477        893'978'551 pread(4, [u"lf *) src0_dd_i : src0_as_f16.get();◙◙  "...], 512, 450'048) → 512 ENOENT
SYS  13477  13477        893'980'777 pread(3, [u"lf *) src0_dd_i : src0_as_f16.get();◙◙  "...], 512, 450'048) → 512 ENOENT
SYS  13477  13477        893'985'308 pread(4, [u"_ddf_i : src1_as_f16.get();◙        ggml"...], 512, 450'560) → 512 ENOENT
SYS  13477  13477        893'987'888 pread(3, [u"_ddf_i : src1_as_f16.get();◙        ggml"...], 512, 450'560) → 512 ENOENT
SYS  13477  13477        893'990'997 pread(4, [u"DA_R_16F, ne10,◙                    &bet"...], 512, 451'072) → 512 ENOENT
SYS  13477  13477        893'993'400 pread(3, [u"DA_R_16F, ne10,◙                    &bet"...], 512, 451'072) → 512 ENOENT
SYS  13477  13477        893'997'993 pread(4, [u"TYPE_F32) {◙            const to_fp32_cu"...], 512, 451'584) → 512 ENOENT
SYS  13477  13477        894'000'281 pread(3, [u"TYPE_F32) {◙            const to_fp32_cu"...], 512, 451'584) → 512 ENOENT
SYS  13477  13477        894'005'007 pread(4, [u"0);◙            to_fp32_cuda(src1_ddf_i,"...], 512, 452'096) → 512 ENOENT
SYS  13477  13477        894'007'392 pread(3, [u"0);◙            to_fp32_cuda(src1_ddf_i,"...], 512, 452'096) → 512 ENOENT
SYS  13477  13477        894'015'845 pread(4, [u"asSgemm(ctx.cublas_handle(id), CUBLAS_OP"...], 512, 452'608) → 512 ENOENT
SYS  13477  13477        894'020'059 pread(3, [u"asSgemm(ctx.cublas_handle(id), CUBLAS_OP"...], 512, 452'608) → 512 ENOENT
SYS  13477  13477        894'023'928 pread(4, [u"ns <= GGML_CUDA_PEER_MAX_BATCH_SIZE;◙◙  "...], 512, 453'120) → 512 ENOENT
SYS  13477  13477        894'027'691 pread(3, [u"ns <= GGML_CUDA_PEER_MAX_BATCH_SIZE;◙◙  "...], 512, 453'120) → 512 ENOENT
SYS  13477  13477        894'030'915 pread(4, [u"other) {◙                continue;◙     "...], 512, 453'632) → 512 ENOENT
SYS  13477  13477        894'035'137 pread(3, [u"other) {◙                continue;◙     "...], 512, 453'632) → 512 ENOENT
SYS  13477  13477        894'036'983 pread(4, [u"    CUDA_CHECK(err);◙                   "...], 512, 454'144) → 512 ENOENT
SYS  13477  13477        894'042'407 pread(3, [u"    CUDA_CHECK(err);◙                   "...], 512, 454'144) → 512 ENOENT
SYS  13477  13477        894'046'294 pread(4, [u"a_context & ctx,◙    const ggml_tensor *"...], 512, 454'656) → 512 ENOENT
SYS  13477  13477        894'048'590 pread(3, [u"a_context & ctx,◙    const ggml_tensor *"...], 512, 454'656) → 512 ENOENT
SYS  13477  13477        894'051'717 pread(4, [u"  GGML_ASSERT(ne03 == ne13);◙◙    const "...], 512, 455'168) → 512 ENOENT
SYS  13477  13477        894'054'049 pread(3, [u"  GGML_ASSERT(ne03 == ne13);◙◙    const "...], 512, 455'168) → 512 ENOENT
SYS  13477  13477        894'059'014 pread(4, [u";◙◙    GGML_ASSERT(src1->type == GGML_TY"...], 512, 455'680) → 512 ENOENT
SYS  13477  13477        894'061'531 pread(3, [u";◙◙    GGML_ASSERT(src1->type == GGML_TY"...], 512, 455'680) → 512 ENOENT
SYS  13477  13477        894'066'628 pread(4, [u"onst int64_t src1_padded_col_size = GGML"...], 512, 456'192) → 512 ENOENT
SYS  13477  13477        894'068'863 pread(3, [u"onst int64_t src1_padded_col_size = GGML"...], 512, 456'192) → 512 ENOENT
SYS  13477  13477        894'070'744 pread(4, [u"_backend_cuda_split_buffer_type_context "...], 512, 456'704) → 512 ENOENT
SYS  13477  13477        894'072'909 pread(3, [u"_backend_cuda_split_buffer_type_context "...], 512, 456'704) → 512 ENOENT
SYS  13477  13477        894'080'655 pread(4, [u"ptr;◙◙        int64_t  row_low;◙        "...], 512, 457'216) → 512 ENOENT
SYS  13477  13477        894'084'966 pread(3, [u"ptr;◙◙        int64_t  row_low;◙        "...], 512, 457'216) → 512 ENOENT
SYS  13477  13477        894'086'803 pread(4, [u"_split);◙◙            if (id != 0) {◙   "...], 512, 457'728) → 512 ENOENT
SYS  13477  13477        894'089'577 pread(3, [u"_split);◙◙            if (id != 0) {◙   "...], 512, 457'728) → 512 ENOENT
SYS  13477  13477        894'091'838 pread(4, [u"    }◙        }◙    }◙◙    for (int id ="...], 512, 458'240) → 512 ENOENT
SYS  13477  13477        894'095'955 pread(3, [u"    }◙        }◙    }◙◙    for (int id ="...], 512, 458'240) → 512 ENOENT
SYS  13477  13477        894'100'681 pread(4, [u"(char *) src0_extra->data_device[id] : ("...], 512, 458'752) → 512 ENOENT
SYS  13477  13477        894'105'283 pread(3, [u"(char *) src0_extra->data_device[id] : ("...], 512, 458'752) → 512 ENOENT
SYS  13477  13477        894'107'085 pread(4, [u"ol(id), nrows1*src1_padded_col_size*q8_1"...], 512, 459'264) → 512 ENOENT
SYS  13477  13477        894'110'627 pread(3, [u"ol(id), nrows1*src1_padded_col_size*q8_1"...], 512, 459'264) → 512 ENOENT
SYS  13477  13477        894'115'282 pread(4, [u"            dev[id].dst_dd = dev[id].dst"...], 512, 459'776) → 512 ENOENT
SYS  13477  13477        894'121'156 pread(3, [u"            dev[id].dst_dd = dev[id].dst"...], 512, 459'776) → 512 ENOENT
SYS  13477  13477        894'125'246 pread(4, [u" > 1 ⁇ MUL_MAT_SRC1_COL_STRIDE : ne11;◙ "...], 512, 460'288) → 512 ENOENT
SYS  13477  13477        894'128'912 pread(3, [u" > 1 ⁇ MUL_MAT_SRC1_COL_STRIDE : ne11;◙ "...], 512, 460'288) → 512 ENOENT
SYS  13477  13477        894'134'336 pread(4, [u"nue;◙            }◙◙            const bo"...], 512, 460'800) → 512 ENOENT
SYS  13477  13477        894'136'729 pread(3, [u"nue;◙            }◙◙            const bo"...], 512, 460'800) → 512 ENOENT
SYS  13477  13477        894'138'788 pread(4, [u" 0));◙            }◙◙            for (in"...], 512, 461'312) → 512 ENOENT
SYS  13477  13477        894'143'778 pread(3, [u" 0));◙            }◙◙            for (in"...], 512, 461'312) → 512 ENOENT
SYS  13477  13477        894'145'651 pread(4, [u"rc1_ddf + (i0*ne11 + src1_col_0) * ne10;"...], 512, 461'824) → 512 ENOENT
SYS  13477  13477        894'149'909 pread(3, [u"rc1_ddf + (i0*ne11 + src1_col_0) * ne10;"...], 512, 461'824) → 512 ENOENT
SYS  13477  13477        894'151'711 pread(4, [u" no tensor split◙                }◙◙    "...], 512, 462'336) → 512 ENOENT
SYS  13477  13477        894'155'562 pread(3, [u" no tensor split◙                }◙◙    "...], 512, 462'336) → 512 ENOENT
SYS  13477  13477        894'158'495 pread(4, [u"c1_padded_col_size*q8_1_ts/q8_1_bs, stre"...], 512, 462'848) → 512 ENOENT
SYS  13477  13477        894'162'770 pread(3, [u"c1_padded_col_size*q8_1_ts/q8_1_bs, stre"...], 512, 462'848) → 512 ENOENT
SYS  13477  13477        894'167'346 pread(4, [u"se if (src1_on_device && !src1_is_contig"...], 512, 463'360) → 512 ENOENT
SYS  13477  13477        894'171'780 pread(3, [u"se if (src1_on_device && !src1_is_contig"...], 512, 463'360) → 512 ENOENT
SYS  13477  13477        894'173'741 pread(4, [u"etLastError());◙                }◙◙     "...], 512, 463'872) → 512 ENOENT
SYS  13477  13477        894'177'345 pread(3, [u"etLastError());◙                }◙◙     "...], 512, 463'872) → 512 ENOENT
SYS  13477  13477        894'185'295 pread(4, [u"       CUDA_CHECK(cudaGetLastError());◙◙"...], 512, 464'384) → 512 ENOENT
SYS  13477  13477        894'190'940 pread(3, [u"       CUDA_CHECK(cudaGetLastError());◙◙"...], 512, 464'384) → 512 ENOENT
SYS  13477  13477        894'199'181 pread(4, [u"                        // Instead they "...], 512, 464'896) → 512 ENOENT
SYS  13477  13477        894'202'017 pread(3, [u"                        // Instead they "...], 512, 464'896) → 512 ENOENT
SYS  13477  13477        894'204'287 pread(4, [u"                  // cudaMemcpy2DAsync m"...], 512, 465'408) → 512 ENOENT
SYS  13477  13477        894'209'931 pread(3, [u"                  // cudaMemcpy2DAsync m"...], 512, 465'408) → 512 ENOENT
SYS  13477  13477        894'211'698 pread(4, [u"ent(row_diff*sizeof(float), src1_ncols, "...], 512, 465'920) → 512 ENOENT
SYS  13477  13477        894'213'942 pread(3, [u"ent(row_diff*sizeof(float), src1_ncols, "...], 512, 465'920) → 512 ENOENT
SYS  13477  13477        894'215'699 pread(4, [u"             cudaMemcpyDeviceToDevice, s"...], 512, 466'432) → 512 ENOENT
SYS  13477  13477        894'221'741 pread(3, [u"             cudaMemcpyDeviceToDevice, s"...], 512, 466'432) → 512 ENOENT
SYS  13477  13477        894'224'347 pread(4, [u"vent for the main device to wait on unti"...], 512, 466'944) → 512 ENOENT
SYS  13477  13477        894'231'034 pread(3, [u"vent for the main device to wait on unti"...], 512, 466'944) → 512 ENOENT
SYS  13477  13477        894'234'771 pread(4, [u"STREAMS ⁇ is_max : GGML_CUDA_MAX_STREAMS"...], 512, 467'456) → 512 ENOENT
SYS  13477  13477        894'240'309 pread(3, [u"STREAMS ⁇ is_max : GGML_CUDA_MAX_STREAMS"...], 512, 467'456) → 512 ENOENT
SYS  13477  13477        894'242'129 pread(4, [u"st ggml_tensor * src0, const ggml_tensor"...], 512, 467'968) → 512 ENOENT
SYS  13477  13477        894'244'664 pread(3, [u"st ggml_tensor * src0, const ggml_tensor"...], 512, 467'968) → 512 ENOENT
SYS  13477  13477        894'246'643 pread(4, [u"e[0];◙    const int64_t ne01 = src0->ne["...], 512, 468'480) → 512 ENOENT
SYS  13477  13477        894'251'342 pread(3, [u"e[0];◙    const int64_t ne01 = src0->ne["...], 512, 468'480) → 512 ENOENT
SYS  13477  13477        894'254'619 pread(4, [u"* src1, ggml_tensor * dst) {◙    GGML_AS"...], 512, 468'992) → 512 ENOENT
SYS  13477  13477        894'260'237 pread(3, [u"* src1, ggml_tensor * dst) {◙    GGML_AS"...], 512, 468'992) → 512 ENOENT
SYS  13477  13477        894'261'977 pread(4, [u"t int64_t ne12 = src1->ne[2];◙◙    cudaS"...], 512, 469'504) → 512 ENOENT
SYS  13477  13477        894'264'345 pread(3, [u"t int64_t ne12 = src1->ne[2];◙◙    cudaS"...], 512, 469'504) → 512 ENOENT
SYS  13477  13477        894'266'147 pread(4, [u"st half * src0_as_f16, const half * src1"...], 512, 470'016) → 512 ENOENT
SYS  13477  13477        894'271'941 pread(3, [u"st half * src0_as_f16, const half * src1"...], 512, 470'016) → 512 ENOENT
SYS  13477  13477        894'273'814 pread(4, [u" r3;◙    int64_t i02 = i12 / r2;◙◙    pt"...], 512, 470'528) → 512 ENOENT
SYS  13477  13477        894'276'031 pread(3, [u" r3;◙    int64_t i02 = i12 / r2;◙◙    pt"...], 512, 470'528) → 512 ENOENT
SYS  13477  13477        894'277'824 pread(4, [u"SSERT(!ggml_is_transposed(src1));◙◙    G"...], 512, 471'040) → 512 ENOENT
SYS  13477  13477        894'283'071 pread(3, [u"SSERT(!ggml_is_transposed(src1));◙◙    G"...], 512, 471'040) → 512 ENOENT
SYS  13477  13477        894'286'534 pread(4, [u" convert src1 to fp16◙    ggml_cuda_pool"...], 512, 471'552) → 512 ENOENT
SYS  13477  13477        894'292'691 pread(3, [u" convert src1 to fp16◙    ggml_cuda_pool"...], 512, 471'552) → 512 ENOENT
SYS  13477  13477        894'295'164 pread(4, [u";◙◙    ggml_cuda_pool_alloc<half> dst_f1"...], 512, 472'064) → 512 ENOENT
SYS  13477  13477        894'300'676 pread(3, [u";◙◙    ggml_cuda_pool_alloc<half> dst_f1"...], 512, 472'064) → 512 ENOENT
SYS  13477  13477        894'302'655 pread(4, [u"GGML_PREC_DEFAULT) {◙        dst_t = (ch"...], 512, 472'576) → 512 ENOENT
SYS  13477  13477        894'304'951 pread(3, [u"GGML_PREC_DEFAULT) {◙        dst_t = (ch"...], 512, 472'576) → 512 ENOENT
SYS  13477  13477        894'306'745 pread(4, [u"13/ne03;◙◙#if 0◙    // use cublasGemmEx◙"...], 512, 473'088) → 512 ENOENT
SYS  13477  13477        894'312'345 pread(3, [u"13/ne03;◙◙#if 0◙    // use cublasGemmEx◙"...], 512, 473'088) → 512 ENOENT
SYS  13477  13477        894'314'324 pread(4, [u"izeof(half),◙                           "...], 512, 473'600) → 512 ENOENT
SYS  13477  13477        894'320'887 pread(3, [u"izeof(half),◙                           "...], 512, 473'600) → 512 ENOENT
SYS  13477  13477        894'323'678 pread(4, [u"= src1->nb[3]) {◙        // there is no "...], 512, 474'112) → 512 ENOENT
SYS  13477  13477        894'326'081 pread(3, [u"= src1->nb[3]) {◙        // there is no "...], 512, 474'112) → 512 ENOENT
SYS  13477  13477        894'331'072 pread(4, [u"t_t, cu_data_type, ne01,       nb2/nb0, "...], 512, 474'624) → 512 ENOENT
SYS  13477  13477        894'333'421 pread(3, [u"t_t, cu_data_type, ne01,       nb2/nb0, "...], 512, 474'624) → 512 ENOENT
SYS  13477  13477        894'336'663 pread(4, [u"  src0_f16, src1_f16, dst_t,◙           "...], 512, 475'136) → 512 ENOENT
SYS  13477  13477        894'339'216 pread(3, [u"  src0_f16, src1_f16, dst_t,◙           "...], 512, 475'136) → 512 ENOENT
SYS  13477  13477        894'344'322 pread(4, [u"      alpha, (const void **) (ptrs_src.g"...], 512, 475'648) → 512 ENOENT
SYS  13477  13477        894'346'698 pread(3, [u"      alpha, (const void **) (ptrs_src.g"...], 512, 475'648) → 512 ENOENT
SYS  13477  13477        894'348'473 pread(4, [u"    to_fp32_cuda(dst_f16.get(), dst_ddf,"...], 512, 476'160) → 512 ENOENT
SYS  13477  13477        894'350'629 pread(3, [u"    to_fp32_cuda(dst_f16.get(), dst_ddf,"...], 512, 476'160) → 512 ENOENT
SYS  13477  13477        894'355'372 pread(4, [u"er->buft->context;◙        auto & tensor"...], 512, 476'672) → 512 ENOENT
SYS  13477  13477        894'357'713 pread(3, [u"er->buft->context;◙        auto & tensor"...], 512, 476'672) → 512 ENOENT
SYS  13477  13477        894'359'506 pread(4, [u"id].cc;◙            }◙            if (gg"...], 512, 477'184) → 512 ENOENT
SYS  13477  13477        894'364'718 pread(3, [u"id].cc;◙            }◙            if (gg"...], 512, 477'184) → 512 ENOENT
SYS  13477  13477        894'366'476 pread(4, [u"PE_F16)◙        && src1->type == GGML_TY"...], 512, 477'696) → 512 ENOENT
SYS  13477  13477        894'370'062 pread(3, [u"PE_F16)◙        && src1->type == GGML_TY"...], 512, 477'696) → 512 ENOENT
SYS  13477  13477        894'377'712 pread(4, [u"fined(__HIP_PLATFORM_AMD__)◙◙    const b"...], 512, 478'208) → 512 ENOENT
SYS  13477  13477        894'381'695 pread(3, [u"fined(__HIP_PLATFORM_AMD__)◙◙    const b"...], 512, 478'208) → 512 ENOENT
SYS  13477  13477        894'385'520 pread(4, [u"available for CC >= 6.1◙    use_mul_mat_"...], 512, 478'720) → 512 ENOENT
SYS  13477  13477        894'390'661 pread(3, [u"available for CC >= 6.1◙    use_mul_mat_"...], 512, 478'720) → 512 ENOENT
SYS  13477  13477        894'394'566 pread(4, [u"GGML_USE_HIPBLAS) && defined(__HIP_PLATF"...], 512, 479'232) → 512 ENOENT
SYS  13477  13477        894'399'830 pread(3, [u"GGML_USE_HIPBLAS) && defined(__HIP_PLATF"...], 512, 479'232) → 512 ENOENT
SYS  13477  13477        894'403'938 pread(4, [u"0], src1->ne[1], src1->ne[2], src1->ne[3"...], 512, 479'744) → 512 ENOENT
SYS  13477  13477        894'407'621 pread(3, [u"0], src1->ne[1], src1->ne[2], src1->ne[3"...], 512, 479'744) → 512 ENOENT
SYS  13477  13477        894'411'958 pread(4, [u"e_good && src0->type == GGML_TYPE_F16 &&"...], 512, 480'256) → 512 ENOENT
SYS  13477  13477        894'414'396 pread(3, [u"e_good && src0->type == GGML_TYPE_F16 &&"...], 512, 480'256) → 512 ENOENT
SYS  13477  13477        894'416'084 pread(4, [u" GGML_TYPE_F16 || fp16_performance_good)"...], 512, 480'768) → 512 ENOENT
SYS  13477  13477        894'418'265 pread(3, [u" GGML_TYPE_F16 || fp16_performance_good)"...], 512, 480'768) → 512 ENOENT
SYS  13477  13477        894'425'809 pread(4, [u"ul_mat_q) {◙        ggml_cuda_op_mul_mat"...], 512, 481'280) → 512 ENOENT
SYS  13477  13477        894'430'182 pread(3, [u"ul_mat_q) {◙        ggml_cuda_op_mul_mat"...], 512, 481'280) → 512 ENOENT
SYS  13477  13477        894'433'918 pread(4, [u"apping,◙                                "...], 512, 481'792) → 512 ENOENT
SYS  13477  13477        894'436'930 pread(3, [u"apping,◙                                "...], 512, 481'792) → 512 ENOENT
SYS  13477  13477        894'439'156 pread(4, [u"1;◙    const int64_t i12 = iid1;◙◙    __"...], 512, 482'304) → 512 ENOENT
SYS  13477  13477        894'443'387 pread(3, [u"1;◙    const int64_t i12 = iid1;◙◙    __"...], 512, 482'304) → 512 ENOENT
SYS  13477  13477        894'447'150 pread(4, [u"_global__ void k_copy_dst_from_contiguou"...], 512, 482'816) → 512 ENOENT
SYS  13477  13477        894'451'814 pread(3, [u"_global__ void k_copy_dst_from_contiguou"...], 512, 482'816) → 512 ENOENT
SYS  13477  13477        894'455'798 pread(4, [u"const float *)(dst_contiguous + i*nb1);◙"...], 512, 483'328) → 512 ENOENT
SYS  13477  13477        894'461'884 pread(3, [u"const float *)(dst_contiguous + i*nb1);◙"...], 512, 483'328) → 512 ENOENT
SYS  13477  13477        894'465'851 pread(4, [u"ackend_buffer_is_cuda_split(src0->buffer"...], 512, 483'840) → 512 ENOENT
SYS  13477  13477        894'469'896 pread(3, [u"ackend_buffer_is_cuda_split(src0->buffer"...], 512, 483'840) → 512 ENOENT
SYS  13477  13477        894'473'297 pread(4, [u" src1_row = *src1;◙    ggml_tensor dst_r"...], 512, 484'352) → 512 ENOENT
SYS  13477  13477        894'478'641 pread(3, [u" src1_row = *src1;◙    ggml_tensor dst_r"...], 512, 484'352) → 512 ENOENT
SYS  13477  13477        894'482'325 pread(4, [u" = nb1;◙◙    if (ne12 == 1) {◙        fo"...], 512, 484'864) → 512 ENOENT
SYS  13477  13477        894'484'666 pread(3, [u" = nb1;◙◙    if (ne12 == 1) {◙        fo"...], 512, 484'864) → 512 ENOENT
SYS  13477  13477        894'486'485 pread(4, [u"_original + i02*nb02;◙                sr"...], 512, 485'376) → 512 ENOENT
SYS  13477  13477        894'489'215 pread(3, [u"_original + i02*nb02;◙                sr"...], 512, 485'376) → 512 ENOENT
SYS  13477  13477        894'491'476 pread(4, [u");◙        dst_row.data  =  dst_contiguo"...], 512, 485'888) → 512 ENOENT
SYS  13477  13477        894'494'656 pread(3, [u");◙        dst_row.data  =  dst_contiguo"...], 512, 485'888) → 512 ENOENT
SYS  13477  13477        894'496'909 pread(4, [u"    continue;◙                    }◙◙   "...], 512, 486'400) → 512 ENOENT
SYS  13477  13477        894'501'016 pread(3, [u"    continue;◙                    }◙◙   "...], 512, 486'400) → 512 ENOENT
SYS  13477  13477        894'502'871 pread(4, [u"e10, 768u));◙                dim3 grid_d"...], 512, 486'912) → 512 ENOENT
SYS  13477  13477        894'506'643 pread(3, [u"e10, 768u));◙                dim3 grid_d"...], 512, 486'912) → 512 ENOENT
SYS  13477  13477        894'510'468 pread(4, [u"ginal + i02*nb02;◙◙            GGML_ASSE"...], 512, 487'424) → 512 ENOENT
SYS  13477  13477        894'512'817 pread(3, [u"ginal + i02*nb02;◙◙            GGML_ASSE"...], 512, 487'424) → 512 ENOENT
SYS  13477  13477        894'514'876 pread(4, [u"0_row, &src1_row, &dst_row);◙◙          "...], 512, 487'936) → 512 ENOENT
SYS  13477  13477        894'520'458 pread(3, [u"0_row, &src1_row, &dst_row);◙◙          "...], 512, 487'936) → 512 ENOENT
SYS  13477  13477        894'522'313 pread(4, [u"da_compute_forward(ggml_backend_cuda_con"...], 512, 488'448) → 512 ENOENT
SYS  13477  13477        894'525'007 pread(3, [u"da_compute_forward(ggml_backend_cuda_con"...], 512, 488'448) → 512 ENOENT
SYS  13477  13477        894'527'260 pread(4, [u"GGML_OP_DUP:◙            ggml_cuda_dup(c"...], 512, 488'960) → 512 ENOENT
SYS  13477  13477        894'531'394 pread(3, [u"GGML_OP_DUP:◙            ggml_cuda_dup(c"...], 512, 488'960) → 512 ENOENT
SYS  13477  13477        894'535'528 pread(4, [u"      case GGML_OP_DIV:◙            ggml"...], 512, 489'472) → 512 ENOENT
SYS  13477  13477        894'540'095 pread(3, [u"      case GGML_OP_DIV:◙            ggml"...], 512, 489'472) → 512 ENOENT
SYS  13477  13477        894'544'043 pread(4, [u" break;◙                case GGML_UNARY_"...], 512, 489'984) → 512 ENOENT
SYS  13477  13477        894'548'239 pread(3, [u" break;◙                case GGML_UNARY_"...], 512, 489'984) → 512 ENOENT
SYS  13477  13477        894'551'728 pread(4, [u"case GGML_UNARY_OP_HARDSWISH:◙          "...], 512, 490'496) → 512 ENOENT
SYS  13477  13477        894'557'125 pread(3, [u"case GGML_UNARY_OP_HARDSWISH:◙          "...], 512, 490'496) → 512 ENOENT
SYS  13477  13477        894'560'756 pread(4, [u"          ggml_cuda_op_upscale(ctx, dst)"...], 512, 491'008) → 512 ENOENT
SYS  13477  13477        894'563'141 pread(3, [u"          ggml_cuda_op_upscale(ctx, dst)"...], 512, 491'008) → 512 ENOENT
SYS  13477  13477        894'566'604 pread(4, [u"orm(ctx, dst);◙            break;◙      "...], 512, 491'520) → 512 ENOENT
SYS  13477  13477        894'572'504 pread(3, [u"orm(ctx, dst);◙            break;◙      "...], 512, 491'520) → 512 ENOENT
SYS  13477  13477        894'574'845 pread(4, [u"         ggml_cuda_mul_mat_id(ctx, dst);"...], 512, 492'032) → 512 ENOENT
SYS  13477  13477        894'578'847 pread(3, [u"         ggml_cuda_mul_mat_id(ctx, dst);"...], 512, 492'032) → 512 ENOENT
SYS  13477  13477        894'580'649 pread(4, [u"_DIAG_MASK_INF:◙            ggml_cuda_op"...], 512, 492'544) → 512 ENOENT
SYS  13477  13477        894'582'989 pread(3, [u"_DIAG_MASK_INF:◙            ggml_cuda_op"...], 512, 492'544) → 512 ENOENT
SYS  13477  13477        894'586'311 pread(4, [u"ws(ctx, dst);◙            break;◙       "...], 512, 493'056) → 512 ENOENT
SYS  13477  13477        894'591'690 pread(3, [u"ws(ctx, dst);◙            break;◙       "...], 512, 493'056) → 512 ENOENT
SYS  13477  13477        894'595'321 pread(4, [u"rue;◙}◙◙////////////////////////////////"...], 512, 493'568) → 512 ENOENT
SYS  13477  13477        894'597'706 pread(3, [u"rue;◙}◙◙////////////////////////////////"...], 512, 493'568) → 512 ENOENT
SYS  13477  13477        894'601'239 pread(4, [u"◙}◙◙GGML_CALL static ggml_backend_buffer"...], 512, 494'080) → 512 ENOENT
SYS  13477  13477        894'604'605 pread(3, [u"◙}◙◙GGML_CALL static ggml_backend_buffer"...], 512, 494'080) → 512 ENOENT
SYS  13477  13477        894'606'963 pread(4, [u"    ggml_backend_buffer_t buf = tensor->"...], 512, 494'592) → 512 ENOENT
SYS  13477  13477        894'611'018 pread(3, [u"    ggml_backend_buffer_t buf = tensor->"...], 512, 494'592) → 512 ENOENT
SYS  13477  13477        894'615'381 pread(4, [u"uda_ctx = (ggml_backend_cuda_context *)b"...], 512, 495'104) → 512 ENOENT
SYS  13477  13477        894'619'922 pread(3, [u"uda_ctx = (ggml_backend_cuda_context *)b"...], 512, 495'104) → 512 ENOENT
SYS  13477  13477        894'623'923 pread(4, [u"l_tensor * src, ggml_tensor * dst) {◙   "...], 512, 495'616) → 512 ENOENT
SYS  13477  13477        894'630'106 pread(3, [u"l_tensor * src, ggml_tensor * dst) {◙   "...], 512, 495'616) → 512 ENOENT
SYS  13477  13477        894'634'002 pread(4, [u"uda_context * cuda_ctx_src = (ggml_backe"...], 512, 496'128) → 512 ENOENT
SYS  13477  13477        894'638'162 pread(3, [u"uda_context * cuda_ctx_src = (ggml_backe"...], 512, 496'128) → 512 ENOENT
SYS  13477  13477        894'641'616 pread(4, [u"  GGML_ASSERT(cuda_ctx_dst->device == bu"...], 512, 496'640) → 512 ENOENT
SYS  13477  13477        894'646'899 pread(3, [u"  GGML_ASSERT(cuda_ctx_dst->device == bu"...], 512, 496'640) → 512 ENOENT
SYS  13477  13477        894'650'529 pread(4, [u";◙#endif◙        }◙◙        // record ev"...], 512, 497'152) → 512 ENOENT
SYS  13477  13477        894'652'888 pread(3, [u";◙#endif◙        }◙◙        // record ev"...], 512, 497'152) → 512 ENOENT
SYS  13477  13477        894'654'698 pread(4, [u"se {◙        // src and dst are on the s"...], 512, 497'664) → 512 ENOENT
SYS  13477  13477        894'660'228 pread(3, [u"se {◙        // src and dst are on the s"...], 512, 497'664) → 512 ENOENT
SYS  13477  13477        894'664'768 pread(4, [u"ensor * node, ggml_graph_node_properties"...], 512, 498'176) → 512 ENOENT
SYS  13477  13477        894'668'876 pread(3, [u"ensor * node, ggml_graph_node_properties"...], 512, 498'176) → 512 ENOENT
SYS  13477  13477        894'672'816 pread(4, [u"_properties(ggml_tensor * node, ggml_gra"...], 512, 498'688) → 512 ENOENT
SYS  13477  13477        894'676'605 pread(3, [u"_properties(ggml_tensor * node, ggml_gra"...], 512, 498'688) → 512 ENOENT
SYS  13477  13477        894'680'209 pread(4, [u"_properties->nb[i]) {◙            return"...], 512, 499'200) → 512 ENOENT
SYS  13477  13477        894'682'568 pread(3, [u"_properties->nb[i]) {◙            return"...], 512, 499'200) → 512 ENOENT
SYS  13477  13477        894'687'965 pread(4, [u"a_ctx = (ggml_backend_cuda_context *)bac"...], 512, 499'712) → 512 ENOENT
SYS  13477  13477        894'690'438 pread(3, [u"a_ctx = (ggml_backend_cuda_context *)bac"...], 512, 499'712) → 512 ENOENT
SYS  13477  13477        894'693'963 pread(4, [u"ntify◙    // kernel parameters which nee"...], 512, 500'224) → 512 ENOENT
SYS  13477  13477        894'699'006 pread(3, [u"ntify◙    // kernel parameters which nee"...], 512, 500'224) → 512 ENOENT
SYS  13477  13477        894'702'822 pread(4, [u"ase which is changing too rapidly,◙    /"...], 512, 500'736) → 512 ENOENT
SYS  13477  13477        894'705'375 pread(3, [u"ase which is changing too rapidly,◙    /"...], 512, 500'736) → 512 ENOENT
SYS  13477  13477        894'708'043 pread(4, [u"aph_update_required = true;◙        }◙◙ "...], 512, 501'248) → 512 ENOENT
SYS  13477  13477        894'711'241 pread(3, [u"aph_update_required = true;◙        }◙◙ "...], 512, 501'248) → 512 ENOENT
SYS  13477  13477        894'714'306 pread(4, [u"aph->n_nodes; i++) {◙            bool ha"...], 512, 501'760) → 512 ENOENT
SYS  13477  13477        894'716'885 pread(3, [u"aph->n_nodes; i++) {◙            bool ha"...], 512, 501'760) → 512 ENOENT
SYS  13477  13477        894'724'694 pread(4, [u"        // Loop over nodes in GGML graph"...], 512, 502'272) → 512 ENOENT
SYS  13477  13477        894'728'678 pread(3, [u"        // Loop over nodes in GGML graph"...], 512, 502'272) → 512 ENOENT
SYS  13477  13477        894'732'688 pread(4, [u"split buffer\\n“, __func__);◙#endif◙     "...], 512, 502'784) → 512 ENOENT
SYS  13477  13477        894'736'592 pread(3, [u"split buffer\\n“, __func__);◙#endif◙     "...], 512, 502'784) → 512 ENOENT
SYS  13477  13477        894'739'428 pread(4, [u"batch size or context size can cause cha"...], 512, 503'296) → 512 ENOENT
SYS  13477  13477        894'744'048 pread(3, [u"batch size or context size can cause cha"...], 512, 503'296) → 512 ENOENT
SYS  13477  13477        894'749'348 pread(4, [u"nel_arg.push_back((char **) &(node->src["...], 512, 503'808) → 512 ENOENT
SYS  13477  13477        894'753'499 pread(3, [u"nel_arg.push_back((char **) &(node->src["...], 512, 503'808) → 512 ENOENT
SYS  13477  13477        894'758'305 pread(4, [u"tes.◙        if (use_cuda_graph && cuda_"...], 512, 504'320) → 512 ENOENT
SYS  13477  13477        894'760'619 pread(3, [u"tes.◙        if (use_cuda_graph && cuda_"...], 512, 504'320) → 512 ENOENT
SYS  13477  13477        894'762'333 pread(4, [u"◙◙    if (use_cuda_graph && cuda_graph_u"...], 512, 504'832) → 512 ENOENT
SYS  13477  13477        894'764'718 pread(3, [u"◙◙    if (use_cuda_graph && cuda_graph_u"...], 512, 504'832) → 512 ENOENT
SYS  13477  13477        894'769'099 pread(4, [u"e use of CUDA graphs, the execution will"...], 512, 505'344) → 512 ENOENT
SYS  13477  13477        894'771'387 pread(3, [u"e use of CUDA graphs, the execution will"...], 512, 505'344) → 512 ENOENT
SYS  13477  13477        894'775'671 pread(4, [u"          assert(node->buffer->buft == g"...], 512, 505'856) → 512 ENOENT
SYS  13477  13477        894'778'021 pread(3, [u"          assert(node->buffer->buft == g"...], 512, 505'856) → 512 ENOENT
SYS  13477  13477        894'781'439 pread(4, [u"                    GGML_CUDA_LOG_ERROR("...], 512, 506'368) → 512 ENOENT
SYS  13477  13477        894'787'463 pread(3, [u"                    GGML_CUDA_LOG_ERROR("...], 512, 506'368) → 512 ENOENT
SYS  13477  13477        894'789'751 pread(4, [u"       CUDA_CHECK(cudaStreamEndCapture(c"...], 512, 506'880) → 512 ENOENT
SYS  13477  13477        894'794'265 pread(3, [u"       CUDA_CHECK(cudaStreamEndCapture(c"...], 512, 506'880) → 512 ENOENT
SYS  13477  13477        894'798'293 pread(4, [u"◙            }◙#endif◙            graph_"...], 512, 507'392) → 512 ENOENT
SYS  13477  13477        894'800'554 pread(3, [u"◙            }◙#endif◙            graph_"...], 512, 507'392) → 512 ENOENT
SYS  13477  13477        894'802'869 pread(4, [u"date to graph (if required for this toke"...], 512, 507'904) → 512 ENOENT
SYS  13477  13477        894'807'374 pread(3, [u"date to graph (if required for this toke"...], 512, 507'904) → 512 ENOENT
SYS  13477  13477        894'812'020 pread(4, [u"◙            cuda_ctx->cuda_graph->nodes"...], 512, 508'416) → 512 ENOENT
SYS  13477  13477        894'816'984 pread(3, [u"◙            cuda_ctx->cuda_graph->nodes"...], 512, 508'416) → 512 ENOENT
SYS  13477  13477        894'822'046 pread(4, [u"nodes; i++) {◙                    cudaGr"...], 512, 508'928) → 512 ENOENT
SYS  13477  13477        894'826'410 pread(3, [u"nodes; i++) {◙                    cudaGr"...], 512, 508'928) → 512 ENOENT
SYS  13477  13477        894'830'367 pread(4, [u"ect handling by CUDA runtime of CUDA BLA"...], 512, 509'440) → 512 ENOENT
SYS  13477  13477        894'832'708 pread(3, [u"ect handling by CUDA runtime of CUDA BLA"...], 512, 509'440) → 512 ENOENT
SYS  13477  13477        894'836'188 pread(4, [u"e updated value in the CUDA graph◙      "...], 512, 509'952) → 512 ENOENT
SYS  13477  13477        894'842'159 pread(3, [u"e updated value in the CUDA graph◙      "...], 512, 509'952) → 512 ENOENT
SYS  13477  13477        894'844'553 pread(4, [u"arg_ptr;◙                    CUDA_CHECK("...], 512, 510'464) → 512 ENOENT
SYS  13477  13477        894'848'926 pread(3, [u"arg_ptr;◙                    CUDA_CHECK("...], 512, 510'464) → 512 ENOENT
SYS  13477  13477        894'850'675 pread(4, [u"e failed\\n“, __func__);◙#endif◙         "...], 512, 510'976) → 512 ENOENT
SYS  13477  13477        894'852'821 pread(3, [u"e failed\\n“, __func__);◙#endif◙         "...], 512, 510'976) → 512 ENOENT
SYS  13477  13477        894'856'213 pread(4, [u" cudaSuccess);◙        }◙        // Laun"...], 512, 511'488) → 512 ENOENT
SYS  13477  13477        894'860'197 pread(3, [u" cudaSuccess);◙        }◙        // Laun"...], 512, 511'488) → 512 ENOENT
SYS  13477  13477        894'861'999 pread(4, [u"ggml_get_unary_op(op)) {◙               "...], 512, 512'000) → 512 ENOENT
SYS  13477  13477        894'864'119 pread(3, [u"ggml_get_unary_op(op)) {◙               "...], 512, 512'000) → 512 ENOENT
SYS  13477  13477        894'868'916 pread(4, [u"GGML_OP_MUL_MAT:◙        case GGML_OP_MU"...], 512, 512'512) → 512 ENOENT
SYS  13477  13477        894'871'159 pread(3, [u"GGML_OP_MUL_MAT:◙        case GGML_OP_MU"...], 512, 512'512) → 512 ENOENT
SYS  13477  13477        894'873'359 pread(4, [u"->type;◙                if (a_type == GG"...], 512, 513'024) → 512 ENOENT
SYS  13477  13477        894'877'298 pread(3, [u"->type;◙                if (a_type == GG"...], 512, 513'024) → 512 ENOENT
SYS  13477  13477        894'879'100 pread(4, [u"            } break;◙        case GGML_O"...], 512, 513'536) → 512 ENOENT
SYS  13477  13477        894'882'757 pread(3, [u"            } break;◙        case GGML_O"...], 512, 513'536) → 512 ENOENT
SYS  13477  13477        894'885'407 pread(4, [u"   }◙            } break;◙        case G"...], 512, 514'048) → 512 ENOENT
SYS  13477  13477        894'888'057 pread(3, [u"   }◙            } break;◙        case G"...], 512, 514'048) → 512 ENOENT
SYS  13477  13477        894'891'396 pread(4, [u"_Q8_0) {◙                    return true"...], 512, 514'560) → 512 ENOENT
SYS  13477  13477        894'895'513 pread(3, [u"_Q8_0) {◙                    return true"...], 512, 514'560) → 512 ENOENT
SYS  13477  13477        894'897'359 pread(4, [u"ype == GGML_TYPE_Q5_1) {◙               "...], 512, 515'072) → 512 ENOENT
SYS  13477  13477        894'902'076 pread(3, [u"ype == GGML_TYPE_Q5_1) {◙               "...], 512, 515'072) → 512 ENOENT
SYS  13477  13477        894'905'918 pread(4, [u"  } break;◙        case GGML_OP_DUP:◙   "...], 512, 515'584) → 512 ENOENT
SYS  13477  13477        894'909'708 pread(3, [u"  } break;◙        case GGML_OP_DUP:◙   "...], 512, 515'584) → 512 ENOENT
SYS  13477  13477        894'913'815 pread(4, [u"_DIV:◙        case GGML_OP_RMS_NORM:◙   "...], 512, 516'096) → 512 ENOENT
SYS  13477  13477        894'916'359 pread(3, [u"_DIV:◙        case GGML_OP_RMS_NORM:◙   "...], 512, 516'096) → 512 ENOENT
SYS  13477  13477        894'921'483 pread(4, [u"case GGML_OP_TIMESTEP_EMBEDDING:◙       "...], 512, 516'608) → 512 ENOENT
SYS  13477  13477        894'923'771 pread(3, [u"case GGML_OP_TIMESTEP_EMBEDDING:◙       "...], 512, 516'608) → 512 ENOENT
SYS  13477  13477        894'925'511 pread(4, [u"_cuda_context *) backend->context;◙     "...], 512, 517'120) → 512 ENOENT
SYS  13477  13477        894'930'457 pread(3, [u"_cuda_context *) backend->context;◙     "...], 512, 517'120) → 512 ENOENT
SYS  13477  13477        894'932'268 pread(4, [u"&& op->op == GGML_OP_MUL_MAT_ID);◙◙    G"...], 512, 517'632) → 512 ENOENT
SYS  13477  13477        894'937'869 pread(3, [u"&& op->op == GGML_OP_MUL_MAT_ID);◙◙    G"...], 512, 517'632) → 512 ENOENT
SYS  13477  13477        894'939'768 pread(4, [u"nd,◙        /* .context = */ event,◙    "...], 512, 518'144) → 512 ENOENT
SYS  13477  13477        894'941'888 pread(3, [u"nd,◙        /* .context = */ event,◙    "...], 512, 518'144) → 512 ENOENT
SYS  13477  13477        894'946'693 pread(4, [u"backend_cuda_event_wait(ggml_backend_t b"...], 512, 518'656) → 512 ENOENT
SYS  13477  13477        894'949'016 pread(3, [u"backend_cuda_event_wait(ggml_backend_t b"...], 512, 518'656) → 512 ENOENT
SYS  13477  13477        894'954'122 pread(4, [u"d_event_synchronize(event);◙        };◙◙"...], 512, 519'168) → 512 ENOENT
SYS  13477  13477        894'956'366 pread(3, [u"d_event_synchronize(event);◙        };◙◙"...], 512, 519'168) → 512 ENOENT
SYS  13477  13477        894'958'221 pread(4, [u"t_buffer_type = */ ggml_backend_cuda_get"...], 512, 519'680) → 512 ENOENT
SYS  13477  13477        894'963'044 pread(3, [u"t_buffer_type = */ ggml_backend_cuda_get"...], 512, 519'680) → 512 ENOENT
SYS  13477  13477        894'965'119 pread(4, [u"*/ ggml_backend_cuda_graph_compute,◙    "...], 512, 520'192) → 512 ENOENT
SYS  13477  13477        894'970'852 pread(3, [u"*/ ggml_backend_cuda_graph_compute,◙    "...], 512, 520'192) → 512 ENOENT
SYS  13477  13477        894'972'619 pread(4, [u",◙};◙◙GGML_CALL static ggml_guid_t ggml_"...], 512, 520'704) → 512 ENOENT
SYS  13477  13477        894'974'748 pread(3, [u",◙};◙◙GGML_CALL static ggml_guid_t ggml_"...], 512, 520'704) → 512 ENOENT
SYS  13477  13477        894'980'101 pread(4, [u"ntext(device);◙    if (ctx == nullptr) {"...], 512, 521'216) → 512 ENOENT
SYS  13477  13477        894'982'327 pread(3, [u"ntext(device);◙    if (ctx == nullptr) {"...], 512, 521'216) → 512 ENOENT
SYS  13477  13477        894'983'970 pread(4, [u"d_cuda_guid());◙}◙◙GGML_CALL int ggml_ba"...], 512, 521'728) → 512 ENOENT
SYS  13477  13477        894'989'022 pread(3, [u"d_cuda_guid());◙}◙◙GGML_CALL int ggml_ba"...], 512, 521'728) → 512 ENOENT
SYS  13477  13477        894'993'987 pread(4, [u" CUDA_CHECK(cudaMemGetInfo(free, total))"...], 512, 522'240) → 512 ENOENT
SYS  13477  13477        894'997'962 pread(3, [u" CUDA_CHECK(cudaMemGetInfo(free, total))"...], 512, 522'240) → 512 ENOENT
SYS  13477  13477        895'001'433 pread(4, [u" %s\\n“, __func__,◙                      "...], 512, 522'752) → 512 ENOENT
SYS  13477  13477        895'007'281 pread(3, [u" %s\\n“, __func__,◙                      "...], 512, 522'752) → 512 ENOENT
SYS  13477  13477        895'009'048 pread(4, [u"backend_t ggml_backend_reg_cuda_init(con"...], 512, 523'264) → 512 ENOENT
SYS  13477  13477        895'011'274 pread(3, [u"backend_t ggml_backend_reg_cuda_init(con"...], 512, 523'264) → 512 ENOENT
SYS  13477  13477        895'016'415 pread(4, [u"< device_count; i++) {◙        char name"...], 512, 523'776) → 481 ENOENT
SYS  13477  13477        895'018'641 pread(3, [u"< device_count; i++) {◙        char name"...], 512, 523'776) → 481 ENOENT
SYS  13477  13477        895'020'407 pread(4, [u""], 512, 524'257) → 0 ENOENT
SYS  13477  13477        895'023'499 pread(3, [u""], 512, 524'257) → 0 ENOENT
SYS  13477  13477        895'078'478 munmap(0x7fe3b55d5000, 524'321) → 0 ENOENT (37'220'352 bytes total)
SYS  13477  13477        895'083'707 close(4) → 0 ENOENT
SYS  13477  13477        895'087'028 close(3) → 0 ENOENT
SYS  13477  13477        895'092'505 getenv("PATH") → "/home/ec2-user/.local/bin:/home/ec2-user/bin:/usr/local/cuda-12.5/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin"
SYS  13477  13477        895'102'601 faccessat(AT_FDCWD, "/home/ec2-user/.local/bin/hipcc", 01, 0) → -1 ENOENT
SYS  13477  13477        895'109'006 faccessat(AT_FDCWD, "/home/ec2-user/bin/hipcc", 01, 0) → -1 ENOENT
SYS  13477  13477        895'112'989 faccessat(AT_FDCWD, "/usr/local/cuda-12.5/bin/hipcc", 01, 0) → -1 ENOENT
SYS  13477  13477        895'119'500 faccessat(AT_FDCWD, "/usr/local/sbin/hipcc", 01, 0) → -1 ENOENT
SYS  13477  13477        895'122'786 faccessat(AT_FDCWD, "/usr/local/bin/hipcc", 01, 0) → -1 ENOENT
SYS  13477  13477        895'129'552 faccessat(AT_FDCWD, "/usr/sbin/hipcc", 01, 0) → -1 ENOENT
SYS  13477  13477        895'132'785 faccessat(AT_FDCWD, "/usr/bin/hipcc", 01, 0) → -1 ENOENT
get_rocm_bin_path: note: hipcc not found on $PATH
SYS  13477  13477        895'139'975 write(2, u"get_rocm_bin_path: note: hipcc not found"..., 50) → 50 ENOENT
SYS  13477  13477        895'142'033 getenv("HIP_PATH") → NULL
get_rocm_bin_path: note: $HIP_PATH/bin/hipcc does not exist
SYS  13477  13477        895'148'340 write(2, u"get_rocm_bin_path: note: $HIP_PATH/bin/h"..., 60) → 60 ENOENT
SYS  13477  13477        895'151'944 fstatat(AT_FDCWD, "/opt/rocm", [n/a], 0) → -1 ENOENT
get_rocm_bin_path: note: /opt/rocm/bin/hipcc does not exist
SYS  13477  13477        895'157'845 write(2, u"get_rocm_bin_path: note: /opt/rocm/bin/h"..., 60) → 60 ENOENT
SYS  13477  13477        895'159'638 getenv("HOME") → "/home/ec2-user"
SYS  13477  13477        895'165'398 fstatat(AT_FDCWD, "/home/ec2-user/.llamafile/v/0.8.12/ggml-cuda.cu", [{.st_size=524'257, .st_blocks=524'288/512, .st_mode=0100600, .st_uid=1000, .st_gid=1000, .st_dev=0x10301, .st_ino=0x47d8a86}], 0) → 0 ENOENT
SYS  13477  13477        895'173'683 fstatat(AT_FDCWD, "/home/ec2-user/.llamafile/v/0.8.12/ggml-rocm.so", [{.st_size=18'128'272, .st_blocks=18'128'896/512, .st_mode=0100600, .st_uid=1000, .st_gid=1000, .st_dev=0x10301, .st_ino=0x47d8a87}], 0) → 0 ENOENT
link_cuda_dso: note: dynamically linking /home/ec2-user/.llamafile/v/0.8.12/ggml-rocm.so
SYS  13477  13477        895'181'465 write(2, u"link_cuda_dso: note: dynamically linking"..., 89) → 89 ENOENT
SYS  13477  13477        895'186'562 getenv("TMPDIR") → NULL
SYS  13477  13477        895'191'262 getenv("HOME") → "/home/ec2-user"
SYS  13477  13477        895'195'086 mkdirat(AT_FDCWD, "/home/ec2-user/.cosmo/", 0755) → -1 EEXIST
SYS  13477  13477        895'198'673 GetProgramExecutableName() → "/home/ec2-user/cc-llama/_llm/_llamafile/llamafile"
SYS  13477  13477        895'210'050 fstatat(AT_FDCWD, "/home/ec2-user/cc-llama/_llm/_llamafile/llamafile", [{.st_size=29'875'580, .st_blocks=29'876'224/512, .st_mode=0100775, .st_uid=1000, .st_gid=1000, .st_dev=0x10301, .st_ino=0x999bfed}], 0) → 0 EEXIST
SYS  13477  13477        895'217'779 fstatat(AT_FDCWD, "/home/ec2-user/.cosmo/dlopen-helper", [{.st_size=22'320, .st_blocks=24'576/512, .st_mode=0100711, .st_uid=1000, .st_gid=1000, .st_dev=0x10301, .st_ino=0xbf1927d}], 0) → 0 EEXIST
SYS  13477  13477        895'224'316 openat(AT_FDCWD, "/home/ec2-user/.cosmo/dlopen-helper.c", O_RDONLY|O_CLOEXEC) → 3 EEXIST
SYS  13477  13477        895'230'490 pread(3, [u"#include <dlfcn.h>◙#include <stdio.h>◙#i"...], 479, 0) → 478 EEXIST
SYS  13477  13477        895'233'591 close(3) → 0 EEXIST
SYS  13477  13477        895'240'666 openat(AT_FDCWD, "/home/ec2-user/.cosmo/dlopen-helper", O_RDONLY|O_CLOEXEC) → 3 EEXIST
SYS  13477  13477        895'246'576 pread(3, [u"⌂ELF☻☺☺         ♥ > ☺   `►      @       "...], 64, 0) → 64 EEXIST
SYS  13477  13477        895'249'491 pread(3, [u"♠   ♦   @       @       @       ╪☻      "...], 728, 64) → 728 EEXIST
SYS  13477  13477        895'258'351 pread(3, [u"/lib64/ld-linux-x86-64.so.2 "], 28, 792) → 28 EEXIST
SYS  13477  13477        895'296'422 close(3) → 0 EEXIST
SYS  13477  13477        895'306'333 openat(AT_FDCWD, "/lib64/ld-linux-x86-64.so.2", O_RDONLY|O_CLOEXEC) → 3 EEXIST
SYS  13477  13477        895'312'269 pread(3, [u"⌂ELF☻☺☺♥        ♥ > ☺   ≡♂☻     @       "...], 64, 0) → 64 EEXIST
SYS  13477  13477        895'314'999 pread(3, [u"☺   ♦                           p←      "...], 616, 64) → 616 EEXIST
SYS  13477  13477        895'360'764 close(3) → 0 EEXIST
SYS  13477  13477        895'371'055 running dlopen importer 0x7fe3b563abf0...
SYS  13477  13477        895'753'698 dlopen("/home/ec2-user/.llamafile/v/0.8.12/ggml-rocm.so", 1) → 0 EEXIST
SYS  13477  13477        895'771'435 dlerror() → "libamdhip64.so.6: cannot open shared object file: No such file or directory"
link_cuda_dso: warning: libamdhip64.so.6: cannot open shared object file: No such file or directory: failed to load library
SYS  13477  13477        895'778'767 write(2, u"link_cuda_dso: warning: libamdhip64.so.6"..., 124) → 124 EEXIST
SYS  13477  13477        895'781'125 getenv("PATH") → "/home/ec2-user/.local/bin:/home/ec2-user/bin:/usr/local/cuda-12.5/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin"
SYS  13477  13477        895'789'667 faccessat(AT_FDCWD, "/home/ec2-user/.local/bin/cc", 01, 0) → -1 ENOENT
SYS  13477  13477        895'796'619 faccessat(AT_FDCWD, "/home/ec2-user/bin/cc", 01, 0) → -1 ENOENT
SYS  13477  13477        895'800'779 faccessat(AT_FDCWD, "/usr/local/cuda-12.5/bin/cc", 01, 0) → -1 ENOENT
SYS  13477  13477        895'807'740 faccessat(AT_FDCWD, "/usr/local/sbin/cc", 01, 0) → -1 ENOENT
SYS  13477  13477        895'811'282 faccessat(AT_FDCWD, "/usr/local/bin/cc", 01, 0) → -1 ENOENT
SYS  13477  13477        895'818'066 faccessat(AT_FDCWD, "/usr/sbin/cc", 01, 0) → -1 ENOENT
SYS  13477  13477      1'022'515'635 faccessat(AT_FDCWD, "/usr/bin/cc", 01, 0) → 0 ENOENT
SYS  13477  13477      1'022'529'503 fstatat(AT_FDCWD, "/usr/bin/cc", [{.st_size=1'309'336, .st_blocks=1'310'720/512, .st_mode=0100755, .st_nlink=3, .st_dev=0x10301, .st_ino=0x1a59f2}], 0) → 0 ENOENT
SYS  13477  13477      1'022'533'558 getenv("PATH") → "/home/ec2-user/.local/bin:/home/ec2-user/bin:/usr/local/cuda-12.5/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin"
SYS  13477  13477      1'022'540'280 faccessat(AT_FDCWD, "/home/ec2-user/.local/bin/nvcc", 01, 0) → -1 ENOENT
SYS  13477  13477      1'022'548'654 faccessat(AT_FDCWD, "/home/ec2-user/bin/nvcc", 01, 0) → -1 ENOENT
SYS  13477  13477      1'022'554'811 faccessat(AT_FDCWD, "/usr/local/cuda-12.5/bin/nvcc", 01, 0) → 0 ENOENT
SYS  13477  13477      1'022'563'556 fstatat(AT_FDCWD, "/usr/local/cuda-12.5/bin/nvcc", [{.st_size=23'042'472, .st_blocks=23'044'096/512, .st_mode=0100755, .st_dev=0x10301, .st_ino=0x486c89c}], 0) → 0 ENOENT
SYS  13477  13477      1'022'568'149 getenv("HOME") → "/home/ec2-user"
SYS  13477  13477      1'022'577'495 fstatat(AT_FDCWD, "/home/ec2-user/.llamafile/v/0.8.12/ggml-cuda.cu", [{.st_size=524'257, .st_blocks=524'288/512, .st_mode=0100600, .st_uid=1000, .st_gid=1000, .st_dev=0x10301, .st_ino=0x47d8a86}], 0) → 0 EEXIST
SYS  13477  13477      1'022'586'911 fstatat(AT_FDCWD, "/home/ec2-user/.llamafile/v/0.8.12/ggml-cuda.so", [{.st_size=4'857'472, .st_blocks=4'857'856/512, .st_mode=0100711, .st_uid=1000, .st_gid=1000, .st_dev=0x10301, .st_ino=0x47d8a88}], 0) → 0 EEXIST
SYS  13477  13477      1'022'591'478 getcwd([u"/home/ec2-user/cc-llama/_llm/_llamafile"], 1'024) → 40 EEXIST
SYS  13477  13477      1'022'601'362 getenv("TMPDIR") → NULL
SYS  13477  13477      1'022'603'606 getenv("ADDR2LINE") → NULL
SYS  13477  13477      1'022'606'424 fstatat(AT_FDCWD, "cosmocc/3.2/bin/x86_64-linux-musl-addr2line", [n/a], 0) → -1 ENOENT
SYS  13477  13477      1'022'609'860 getenv("PATH") → "/home/ec2-user/.local/bin:/home/ec2-user/bin:/usr/local/cuda-12.5/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin"
SYS  13477  13477      1'022'620'964 faccessat(AT_FDCWD, "/home/ec2-user/.local/bin/addr2line", 01, 0) → -1 ENOENT
SYS  13477  13477      1'022'625'495 faccessat(AT_FDCWD, "/home/ec2-user/bin/addr2line", 01, 0) → -1 ENOENT
SYS  13477  13477      1'022'634'019 faccessat(AT_FDCWD, "/usr/local/cuda-12.5/bin/addr2line", 01, 0) → -1 ENOENT
SYS  13477  13477      1'022'638'507 faccessat(AT_FDCWD, "/usr/local/sbin/addr2line", 01, 0) → -1 ENOENT
SYS  13477  13477      1'022'645'838 faccessat(AT_FDCWD, "/usr/local/bin/addr2line", 01, 0) → -1 ENOENT
SYS  13477  13477      1'022'653'532 faccessat(AT_FDCWD, "/usr/sbin/addr2line", 01, 0) → -1 ENOENT
SYS  13477  13477      1'022'658'903 faccessat(AT_FDCWD, "/usr/bin/addr2line", 01, 0) → 0 ENOENT
SYS  13477  13477      1'022'665'704 fstatat(AT_FDCWD, "/usr/bin/addr2line", [{.st_size=29'192, .st_blocks=32'768/512, .st_mode=0100755, .st_dev=0x10301, .st_ino=0x7ab0e}], 0) → 0 ENOENT
SYS  13477  13477      1'022'668'213 GetProgramExecutableName() → "/home/ec2-user/cc-llama/_llm/_llamafile/llamafile"
SYS  13477  13477      1'022'672'285 chdir("/usr/local/cuda-12.5/bin") → 0 EEXIST
link_cuda_dso: note: dynamically linking /home/ec2-user/.llamafile/v/0.8.12/ggml-cuda.so
SYS  13477  13477      1'022'679'917 write(2, u"link_cuda_dso: note: dynamically linking"..., 89) → 89 EEXIST
SYS  13477  13477      1'048'426'266 dlopen("/home/ec2-user/.llamafile/v/0.8.12/ggml-cuda.so", 1) → 0x80da6960 EEXIST
SYS  13477  13477      1'048'447'952 GetProgramExecutableName() → "/home/ec2-user/cc-llama/_llm/_llamafile/llamafile"
SYS  13477  13477      1'048'461'811 chdir("/home/ec2-user/cc-llama/_llm/_llamafile") → 0 EEXIST
SYS  13477  13477      1'048'465'036 dlsym(0x80da6960, "ggml_cuda_link") → 0x7fe3b4c49590
SYS  13477  13477      1'048'467'173 dlsym(0x80da6960, "ggml_backend_cuda_host_buffer_type") → 0x7fe3b4c67240
SYS  13477  13477      1'048'468'931 dlsym(0x80da6960, "ggml_backend_cuda_buffer_type") → 0x7fe3b4c67bd0
SYS  13477  13477      1'048'470'618 dlsym(0x80da6960, "ggml_backend_cuda_init") → 0x7fe3b4c6cba0
SYS  13477  13477      1'048'472'279 dlsym(0x80da6960, "ggml_backend_cuda_split_buffer_type") → 0x7fe3b4c6dec0
SYS  13477  13477      1'048'474'037 dlsym(0x80da6960, "ggml_backend_cuda_reg_devices") → 0x7fe3b4c6d990
SYS  13477  13477      1'048'475'706 dlsym(0x80da6960, "ggml_backend_cuda_get_device_memory") → 0x7fe3b4c6d690
SYS  13477  13477      1'048'477'429 dlsym(0x80da6960, "ggml_backend_cuda_get_device_count") → 0x7fe3b4c67b30
SYS  13477  13477      1'048'479'301 dlsym(0x80da6960, "ggml_backend_cuda_unregister_host_buffer") → 0x7fe3b4c6d8c0
SYS  13477  13477      1'048'481'077 dlsym(0x80da6960, "ggml_backend_cuda_register_host_buffer") → 0x7fe3b4c6d770
ggml_cuda_link: welcome to CUDA SDK with cuBLAS
SYS  13477  13477      1'048'501'703 write(2, u"ggml_cuda_link: welcome to CUDA SDK with"..., 48) → 48 EEXIST
link_cuda_dso: GPU support loaded
SYS  13477  13477      5'567'591'938 write(2, u"link_cuda_dso: GPU support loaded◙", 34) → 34 EEXIST
jart commented 2 months ago

I don't think we support Mistral Large yet. It's blocked on another sync with llama.cpp upstream.

rpchastain commented 2 months ago

This is happening with Mixtral-8x7B which is listed as one of the supported models in the readme.

somenewaccountthen commented 2 months ago

I have it with TinyLlama-1.1B from the list here: https://github.com/Mozilla-Ocho/llamafile?tab=readme-ov-file#other-example-llamafiles

error: Uncaught SIGSEGV (SEGV_MAPERR) at 0 on linuxdesktop pid 15045 tid 15045 ./mxbai-embed-large-v1-f16.llamafile No such file or directory Linux Cosmopolitan 3.3.6 MODE=x86_64; #39-Ubuntu SMP PREEMPT_DYNAMIC Fri Jul 5 21:49:14 UTC 2024 linuxdesktop 6.8.0-39-generic

RAX 00007fffcaa25af8 RBX 000010008037c370 RDI 0000100088d8f170 RCX 000000000001dce8 RDX 0000100088dace58 RSI 0000000000000000 RBP 00007fffcaa25aa0 RSP 00007fffcaa25a20 RIP 0000000000596f03 R8 0000000000000001 R9 0000000006e60000 R10 0000000000000032 R11 0000000000000296 R12 00007fffcaa25c58 R13 0000000000000000 R14 00007fffcaa25c40 R15 0000000000000000 TLS 000000000084fe40

XMM0 ffff0000000000000000000000000000 XMM8 00000000000000000000000000000000 XMM1 00000000000000000000000000000000 XMM9 00000000000000000000000000000000 XMM2 00000000000000000000000000000000 XMM10 00000000000000000000000000000000 XMM3 0000000000000000000000003f970a3d XMM11 00000000000000000000000000000000 XMM4 00000000000000000000000000000000 XMM12 00000000000000000000000000000000 XMM5 00000000000000000000000000000000 XMM13 00000000000000000000000000000000 XMM6 00000000000000000000000040a00000 XMM14 00000000000000000000000000000000 XMM7 0000000000000000000000003dcccccd XMM15 00000000000000000000000000000000

cosmoaddr2line /tmp/mxbai-embed-large-v1-f16.llamafile 596f03 595768 5964e9 498397 494f07 449949 401c26 4161f3 4015fb

note: pledge() sandboxing makes backtraces not as good 000000a98ce0 596f03 _ZNSt3__16vectorIfNS_9allocatorIfEEEC1IPfEET_NS_9enable_ifIXaasrNS_21is_forward_iteratorIS6_EE5valuesrNS_16is_constructibleIfJNS_15iterator_traitsIS6_E9referenceEEEE5valueES6_E4typeE+159 7fffcaa25aa0 595768 llama_sampling_prepare_impl(llama_sampling_context, llama_context, llama_context*, int, bool, std::1::vector<float, std::__1::allocator>)+174 7fffcaa25b40 5964e9 llama_sampling_sample_impl(llama_sampling_context, llama_context, llama_context, int, bool)+250 7fffcaa25ca0 498397 _ZN20llama_server_context12update_slotsEv.isra.0+13001 7fffcaa26080 494f07 llama_server_queue::start_loop()+1405 7fffcaa261d0 449949 server_cli(int, char**)+10204 7fffcaa27680 401c26 main+462 7fffcaa29640 4161f3 cosmo+73 7fffcaa29650 4015fb _start+133

10008004-10008009 rw-pa- 6x automap 384kB 1000800a-1000800a rw-Sa- 1x automap 64kB 1000800b-1000801f rw-pa- 21x automap 1344kB 10008020-10008027 rw-Sa- 8x automap 512kB 10008028-1000807b rw-pa- 84x automap 5376kB 1000807c-1000807f rw-Sa- 4x automap 256kB 10008080-100080b7 rw-pa- 56x automap 3584kB 100080b8-100080ff rw-Sa- 72x automap 4608kB 10008100-100081e7 rw-pa- 232x automap 15mB 100081e8-100081f3 rw-Sa- 12x automap 768kB 100081f4-100088e2 rw-pa- 1'775x automap 111mB w/ 1856kB hole 10008900-100089c8 rw-pa- 201x automap 13mB w/ 363mB hole 1000a074-1000c9a9 r--s-- 10'550x automap 659mB w/ 357mB hole 1000e000-100107e9 r--s-- 10'218x automap 639mB w/ 96tB hole 6fc00004-6fc00004 rw-paF 1x nsync 64kB w/ 64gB hole 6fd00004-6fd00004 rw-paF 1x zipos 64kB w/ 64gB hole 6fe00004-6fe00004 rw-paF 1x g_fds 64kB

1453mB total mapped memory

./mxbai-embed-large-v1-f16.llamafile --embedding --model mxbai-embed-large-v1-f16.gguf Segmentation fault (core dumped)