search

vllm-project / vllm

A high-throughput and memory-efficient inference and serving engine for LLMs
https://docs.vllm.ai
Apache License 2.0
29.15k stars 4.35k forks source link

[Bug]: Concurrently captioning images with phi3 Vision can cause the backend to crash #5885

Closed KaraKaraWitch closed 4 months ago

KaraKaraWitch commented 4 months ago

Your current environment

PyTorch version: 2.3.0+cu121
Is debug build: False
CUDA used to build PyTorch: 12.1
ROCM used to build PyTorch: N/A

OS: Ubuntu 22.04.3 LTS (x86_64)
GCC version: (Ubuntu 11.4.0-1ubuntu1~22.04) 11.4.0
Clang version: Could not collect
CMake version: version 3.29.6
Libc version: glibc-2.35

Python version: 3.10.12 (main, Jun 11 2023, 05:26:28) [GCC 11.4.0] (64-bit runtime)
Python platform: Linux-5.15.0-112-generic-x86_64-with-glibc2.35
Is CUDA available: True
CUDA runtime version: 12.1.105
CUDA_MODULE_LOADING set to: LAZY
GPU models and configuration: GPU 0: NVIDIA H100 PCIe
Nvidia driver version: 550.67
cuDNN version: Could not collect
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: True

CPU:
Architecture:                       x86_64
CPU op-mode(s):                     32-bit, 64-bit
Address sizes:                      52 bits physical, 57 bits virtual
Byte Order:                         Little Endian
CPU(s):                             128
On-line CPU(s) list:                0-127
Vendor ID:                          AuthenticAMD
Model name:                         AMD EPYC 9354 32-Core Processor
CPU family:                         25
Model:                              17
Thread(s) per core:                 2
Core(s) per socket:                 32
Socket(s):                          2
Stepping:                           1
Frequency boost:                    enabled
CPU max MHz:                        3799.0720
CPU min MHz:                        1500.0000
BogoMIPS:                           6500.17
Flags:                              fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl nonstop_tsc cpuid extd_apicid aperfmperf rapl pni pclmulqdq monitor ssse3 fma cx16 pcid sse4_1 sse4_2 x2apic movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt tce topoext perfctr_core perfctr_nb bpext perfctr_llc mwaitx cpb cat_l3 cdp_l3 invpcid_single hw_pstate ssbd mba ibrs ibpb stibp vmmcall fsgsbase bmi1 avx2 smep bmi2 erms invpcid cqm rdt_a avx512f avx512dq rdseed adx smap avx512ifma clflushopt clwb avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local avx512_bf16 clzero irperf xsaveerptr rdpru wbnoinvd amd_ppin cppc arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold avic v_vmsave_vmload vgif v_spec_ctrl avx512vbmi umip pku ospke avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg avx512_vpopcntdq la57 rdpid overflow_recov succor smca fsrm flush_l1d
Virtualization:                     AMD-V
L1d cache:                          2 MiB (64 instances)
L1i cache:                          2 MiB (64 instances)
L2 cache:                           64 MiB (64 instances)
L3 cache:                           512 MiB (16 instances)
NUMA node(s):                       2
NUMA node0 CPU(s):                  0-31,64-95
NUMA node1 CPU(s):                  32-63,96-127
Vulnerability Gather data sampling: Not affected
Vulnerability Itlb multihit:        Not affected
Vulnerability L1tf:                 Not affected
Vulnerability Mds:                  Not affected
Vulnerability Meltdown:             Not affected
Vulnerability Mmio stale data:      Not affected
Vulnerability Retbleed:             Not affected
Vulnerability Spec rstack overflow: Mitigation; safe RET
Vulnerability Spec store bypass:    Mitigation; Speculative Store Bypass disabled via prctl and seccomp
Vulnerability Spectre v1:           Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2:           Mitigation; Retpolines; IBPB conditional; IBRS_FW; STIBP always-on; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected
Vulnerability Srbds:                Not affected
Vulnerability Tsx async abort:      Not affected

Versions of relevant libraries:
[pip3] numpy==1.26.4
[pip3] nvidia-nccl-cu12==2.20.5
[pip3] onnxruntime-gpu==1.18.0
[pip3] torch==2.3.0
[pip3] torchaudio==2.1.1
[pip3] torchvision==0.18.0
[pip3] transformers==4.41.2
[pip3] triton==2.3.0
[conda] Could not collect
ROCM Version: Could not collect
Neuron SDK Version: N/A
vLLM Version: 0.5.0.post1
vLLM Build Flags:
CUDA Archs: Not Set; ROCm: Disabled; Neuron: Disabled
GPU Topology:
GPU0    CPU Affinity    NUMA Affinity   GPU NUMA ID
GPU0     X      32-63,96-127    1               N/A

Legend:

  X    = Self
  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
  PIX  = Connection traversing at most a single PCIe bridge
  NV#  = Connection traversing a bonded set of # NVLinks

🐛 Describe the bug

When running multiple requests to phi-3 vision, I sometimes get this error. Causing the app to refuse further requests.

Stacktrace on VLLM side:

  File "/usr/local/lib/python3.10/dist-packages/fastapi/routing.py", line 278, in app
    raw_response = await run_endpoint_function(
  File "/usr/local/lib/python3.10/dist-packages/fastapi/routing.py", line 191, in run_endpoint_function
    return await dependant.call(**values)
  File "/usr/local/lib/python3.10/dist-packages/vllm/entrypoints/openai/api_server.py", line 132, in create_chat_completion
    generator = await openai_serving_chat.create_chat_completion(
  File "/usr/local/lib/python3.10/dist-packages/vllm/entrypoints/openai/serving_chat.py", line 293, in create_chat_completion
    return await self.chat_completion_full_generator(
  File "/usr/local/lib/python3.10/dist-packages/vllm/entrypoints/openai/serving_chat.py", line 493, in chat_completion_full_generator
    async for res in result_generator:
  File "/usr/local/lib/python3.10/dist-packages/vllm/engine/async_llm_engine.py", line 684, in generate
    async for output in self._process_request(
  File "/usr/local/lib/python3.10/dist-packages/vllm/engine/async_llm_engine.py", line 797, in _process_request
    raise e
  File "/usr/local/lib/python3.10/dist-packages/vllm/engine/async_llm_engine.py", line 793, in _process_request
    async for request_output in stream:
  File "/usr/local/lib/python3.10/dist-packages/vllm/engine/async_llm_engine.py", line 90, in __anext__
    raise result
  File "/usr/local/lib/python3.10/dist-packages/vllm/engine/async_llm_engine.py", line 43, in _log_task_completion
    return_value = task.result()
  File "/usr/local/lib/python3.10/dist-packages/vllm/engine/async_llm_engine.py", line 552, in run_engine_loop
    has_requests_in_progress = await self.engine_step()
  File "/usr/local/lib/python3.10/dist-packages/vllm/engine/async_llm_engine.py", line 525, in engine_step
    request_outputs = await self.engine.step_async()
  File "/usr/local/lib/python3.10/dist-packages/vllm/engine/async_llm_engine.py", line 236, in step_async
    output = await self.model_executor.execute_model_async(
  File "/usr/local/lib/python3.10/dist-packages/vllm/executor/gpu_executor.py", line 121, in execute_model_async
    output = await make_async(self.driver_worker.execute_model
  File "/usr/lib/python3.10/concurrent/futures/thread.py", line 58, in run
    result = self.fn(*self.args, **self.kwargs)
  File "/usr/local/lib/python3.10/dist-packages/vllm/worker/worker_base.py", line 255, in execute_model
    output = self.model_runner.execute_model(model_input, self.kv_cache)
  File "/usr/local/lib/python3.10/dist-packages/torch/utils/_contextlib.py", line 115, in decorate_context
    return func(*args, **kwargs)
  File "/usr/local/lib/python3.10/dist-packages/vllm/worker/model_runner.py", line 999, in execute_model
    hidden_states = model_executable(
  File "/usr/local/lib/python3.10/dist-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
  File "/usr/local/lib/python3.10/dist-packages/torch/nn/modules/module.py", line 1541, in _call_impl
    return forward_call(*args, **kwargs)
  File "/usr/local/lib/python3.10/dist-packages/vllm/model_executor/models/phi3v.py", line 376, in forward
    inputs_embeds = self.vision_embed_tokens(
  File "/usr/local/lib/python3.10/dist-packages/torch/nn/modules/module.py", line 1532, in _wrapped_call_impl
    return self._call_impl(*args, **kwargs)
  File "/usr/local/lib/python3.10/dist-packages/torch/nn/modules/module.py", line 1541, in _call_impl
    return forward_call(*args, **kwargs)
  File "/usr/local/lib/python3.10/dist-packages/vllm/model_executor/models/phi3v.py", line 220, in forward
    sub_img = sub_img[:B_]
TypeError: only integer tensors of a single element can be converted to an index

The workaround I used was to send 1 request at a time with a multiprocessing lock from my client process.

DarkLight1337 commented 4 months ago

cc @xwjiang2010

DarkLight1337 commented 4 months ago

@KaraKaraWitch Can you try out #5888 and see if it fixes this issue?

KaraKaraWitch commented 4 months ago

Tested commit https://github.com/ywang96/vllm/commit/7097a204835360890606c362034e0cd847bd97ba and it appears to have resolved the issue.