search

ggerganov / llama.cpp

LLM inference in C/C++
MIT License
65.89k stars 9.46k forks source link

Bug: Build llama.cpp with DGGML_VULKAN=ON on Ubuntu 20.04 of arch64, the result of the program is not correct. #8365

Closed warren-lei closed 2 months ago

warren-lei commented 3 months ago

What happened?

command of compilation: cmake .. -DGGML_VULKAN=ON -DCMAKE_BUILD_TYPE=Release

command of running: ./bin/llama-cli -m ggml-model-q4_k.gguf -c 512 -b 1024 -n 256 --keep 48 --repeat_penalty 1.0 --color -i -r "User:" -f ../prompts/chat-with-bob.txt -ngl 33

result: llama

Name and Version

./bin/llama-cli --version
version: 3341 (6f0dbf6a)
built with cc (Ubuntu 9.4.0-1ubuntu1~20.04.1) 9.4.0 for aarch64-linux-gn

system version:

 lsb_release -a
No LSB modules are available.
Distributor ID: Ubuntu
Description:    Ubuntu 20.04.6 LTS
Release:    20.04
Codename:   focal
arch
aarch64
uname -a
Linux ubuntu 5.10.104-tegra #1 SMP PREEMPT Tue Jan 24 15:09:44 PST 2023 aarch64 aarch64 aarch64 GNU/Linux

Vulkan verison: libvulkan.so.1.3.289. Only compile vulkan loader.

What operating system are you seeing the problem on?

No response

Relevant log output

Log start
main: build = 3341 (6f0dbf6a)
main: built with cc (Ubuntu 9.4.0-1ubuntu1~20.04.1) 9.4.0 for aarch64-linux-gnu
main: seed  = 1720430997
llama_model_loader: loaded meta data with 19 key-value pairs and 291 tensors from ../build/ggml-model-q4_k.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              = LLaMA v2
llama_model_loader: - kv   2:                       llama.context_length u32              = 4096
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              = 11008
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              = 32
llama_model_loader: - kv   9:     llama.attention.layer_norm_rms_epsilon f32              = 0.000010
llama_model_loader: - kv  10:                          general.file_type u32              = 15
llama_model_loader: - kv  11:                       tokenizer.ggml.model str              = llama
llama_model_loader: - kv  12:                      tokenizer.ggml.tokens arr[str,55296]   = ["<unk>", "<s>", "</s>", "<0x00>", "<...
llama_model_loader: - kv  13:                      tokenizer.ggml.scores arr[f32,55296]   = [0.000000, 0.000000, 0.000000, 0.0000...
llama_model_loader: - kv  14:                  tokenizer.ggml.token_type arr[i32,55296]   = [2, 3, 3, 6, 6, 6, 6, 6, 6, 6, 6, 6, ...
llama_model_loader: - kv  15:                tokenizer.ggml.bos_token_id u32              = 1
llama_model_loader: - kv  16:                tokenizer.ggml.eos_token_id u32              = 2
llama_model_loader: - kv  17:            tokenizer.ggml.padding_token_id u32              = 0
llama_model_loader: - kv  18:               general.quantization_version u32              = 2
llama_model_loader: - type  f32:   65 tensors
llama_model_loader: - type q4_K:  193 tensors
llama_model_loader: - type q6_K:   33 tensors
llm_load_vocab: special tokens cache size = 259
llm_load_vocab: token to piece cache size = 0.2971 MB
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          = 55296
llm_load_print_meta: n_merges         = 0
llm_load_print_meta: vocab_only       = 0
llm_load_print_meta: n_ctx_train      = 4096
llm_load_print_meta: n_embd           = 4096
llm_load_print_meta: n_layer          = 32
llm_load_print_meta: n_head           = 32
llm_load_print_meta: n_head_kv        = 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            = 1
llm_load_print_meta: n_embd_k_gqa     = 4096
llm_load_print_meta: n_embd_v_gqa     = 4096
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             = 11008
llm_load_print_meta: n_expert         = 0
llm_load_print_meta: n_expert_used    = 0
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  = 10000.0
llm_load_print_meta: freq_scale_train = 1
llm_load_print_meta: n_ctx_orig_yarn  = 4096
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       = 8B
llm_load_print_meta: model ftype      = Q4_K - Medium
llm_load_print_meta: model params     = 6.93 B
llm_load_print_meta: model size       = 3.92 GiB (4.86 BPW) 
llm_load_print_meta: general.name     = LLaMA v2
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>'
llm_load_print_meta: max token length = 48
ggml_vulkan: Found 1 Vulkan devices:
Vulkan0: NVIDIA Tegra Orin (nvgpu) (NVIDIA) | uma: 1 | fp16: 1 | warp size: 32
llm_load_tensors: ggml ctx size =    0.27 MiB
llm_load_tensors: offloading 32 repeating layers to GPU
llm_load_tensors: offloading non-repeating layers to GPU
llm_load_tensors: offloaded 33/33 layers to GPU
llm_load_tensors: NVIDIA Tegra Orin (nvgpu) buffer size =  3895.58 MiB
llm_load_tensors:        CPU buffer size =   121.50 MiB
..............................................................................................
llama_new_context_with_model: n_ctx      = 512
llama_new_context_with_model: n_batch    = 512
llama_new_context_with_model: n_ubatch   = 512
llama_new_context_with_model: flash_attn = 0
llama_new_context_with_model: freq_base  = 10000.0
llama_new_context_with_model: freq_scale = 1
llama_kv_cache_init: NVIDIA Tegra Orin (nvgpu) KV buffer size =   256.00 MiB
llama_new_context_with_model: KV self size  =  256.00 MiB, K (f16):  128.00 MiB, V (f16):  128.00 MiB
llama_new_context_with_model: Vulkan_Host  output buffer size =     0.21 MiB
llama_new_context_with_model: NVIDIA Tegra Orin (nvgpu) compute buffer size =   116.00 MiB
llama_new_context_with_model: Vulkan_Host compute buffer size =     9.01 MiB
llama_new_context_with_model: graph nodes  = 1030
llama_new_context_with_model: graph splits = 2

system_info: n_threads = 12 / 12 | AVX = 0 | AVX_VNNI = 0 | AVX2 = 0 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | AVX512_BF16 = 0 | FMA = 0 | NEON = 1 | SVE = 0 | ARM_FMA = 1 | F16C = 0 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 0 | SSSE3 = 0 | VSX = 0 | MATMUL_INT8 = 0 | LLAMAFILE = 0 | 
main: interactive mode on.
Reverse prompt: 'User:'
sampling: 
    repeat_last_n = 64, repeat_penalty = 1.000, frequency_penalty = 0.000, presence_penalty = 0.000
    top_k = 40, tfs_z = 1.000, top_p = 0.950, min_p = 0.050, typical_p = 1.000, temp = 0.800
    mirostat = 0, mirostat_lr = 0.100, mirostat_ent = 5.000
sampling order: 
CFG -> Penalties -> top_k -> tfs_z -> typical_p -> top_p -> min_p -> temperature 
generate: n_ctx = 512, n_batch = 1024, n_predict = 256, n_keep = 49

== Running in interactive mode. ==
 - Press Ctrl+C to interject at any time.
 - Press Return to return control to the AI.
 - To return control without starting a new line, end your input with '/'.
 - If you want to submit another line, end your input with '\'.

 Transcript of a dialog, where the User interacts with an Assistant named Bob. Bob is helpful, kind, honest, good at writing, and never fails to answer the User's requests immediately and with precision.

User: Hello, Bob.
Bob: Hello. How may I help you today?
User: Please tell me the largest city in Europe.
Bob: Sure. The largest city in Europe is Moscow, the capital of Russia.
User:Where is Beijing
 graphic design of an open-minded ，我，正在 ，我，我，我正在 ，我是一个，我意识到重新组织我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我，我
warren-lei commented 2 months ago

Currently, it seems that the wrong output of Vulkan may be caused by data type conversion issues. For the Q4 model (4-bit, ggml-model-q4_k.gguf), setting ngl to 11 starts to cause some wrong output, and the higher the setting layers of ngl, the more errors occur. For the F16 model, it can provide correct answers with ngl set to 18, but when ngl is set to 19 , errors begin to occur.

warren-lei commented 2 months ago

I found the cause of the problem. I only install vulkan loader in my PC without installing VULKAN SDK. So glslc is not installed. Thus the VULKAN shaders will not be compiled. After I built and installed glslc from repo Shaderc, then recompiled llama with GGML_VULKAN, the answer was correct.

warren-lei commented 2 months ago

Resolved. Close the issue.