vllm - ✅(Solved) Fix [Bug]: GPU hang on MI325/300 when serving MiniMax-M2.1-MXFP4 with TP=1 [1 pull requests, 3 comments, 3 participants]

vllm2026-03-18 09:49:42

ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

GitHub issue URL

Helpful · Quick feedback

GitHub stats

vllm-project/vllm#37406•Fetched 2026-04-08 00:57:37

View on GitHub

Comments

Participants

Timeline

Reactions

Author

Participants

Timeline (top)

mentioned ×5subscribed ×5commented ×3labeled ×2

Root Cause

This is because MI300 has no native MXFP4, therefore Quark uses the emulative (QDQ) path, dequantizing MXFP4 to BF16 (~4× memory). With a ~230 GB FP16-equivalent model, a single 192 GB MI300 is under heavy memory pressure, which can lead to hang or invalid access.

Fix Action

Fix / Workaround

============================== CPU Info

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): 256 On-line CPU(s) list: 0-255 Vendor ID: AuthenticAMD Model name: AMD EPYC 9555 64-Core Processor CPU family: 26 Model: 2 Thread(s) per core: 2 Core(s) per socket: 64 Socket(s): 2 Stepping: 1 Frequency boost: enabled CPU max MHz: 4409.3750 CPU min MHz: 1500.0000 BogoMIPS: 6399.98 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 amd_lbr_v2 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 hw_pstate ssbd mba perfmon_v2 ibrs ibpb stibp ibrs_enhanced vmmcall fsgsbase tsc_adjust 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 user_shstk avx_vnni avx512_bf16 clzero irperf xsaveerptr rdpru wbnoinvd amd_ppin cppc amd_ibpb_ret arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold avic v_vmsave_vmload vgif x2avic v_spec_ctrl vnmi avx512vbmi umip pku ospke avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg avx512_vpopcntdq la57 rdpid bus_lock_detect movdiri movdir64b overflow_recov succor smca fsrm avx512_vp2intersect flush_l1d debug_swap Virtualization: AMD-V L1d cache: 6 MiB (128 instances) L1i cache: 4 MiB (128 instances) L2 cache: 128 MiB (128 instances) L3 cache: 512 MiB (16 instances) NUMA node(s): 2 NUMA node0 CPU(s): 0-63,128-191 NUMA node1 CPU(s): 64-127,192-255 Vulnerability Gather data sampling: Not affected Vulnerability Indirect target selection: 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 Reg file data sampling: Not affected Vulnerability Retbleed: Not affected Vulnerability Spec rstack overflow: Not affected Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization Vulnerability Spectre v2: Mitigation; Enhanced / Automatic IBRS; IBPB conditional; STIBP always-on; PBRSB-eIBRS Not affected; BHI Not affected Vulnerability Srbds: Not affected Vulnerability Tsa: Not affected Vulnerability Tsx async abort: Not affected Vulnerability Vmscape: Not affected

PR fix notes

PR #37408: [ROCm][Quantization] fallback trust_remote_code=True in Quark config for some cases

Repository: vllm-project/vllm
Author: xuebwang-amd
State: closed | merged: False
Link: https://github.com/vllm-project/vllm/pull/37408

Description (problem / solution / changelog)

Purpose

Model: amd/MiniMax-M2.1-MXFP4 Transformers: 4.57.6

Error message:

... ...
(APIServer pid=295080)   File "/workspace/xuebwang/vllm/vllm/engine/arg_utils.py", line 1928, in create_engine_config
(APIServer pid=295080)     config = VllmConfig(
(APIServer pid=295080)              ^^^^^^^^^^^
(APIServer pid=295080)   File "/usr/local/lib/python3.12/dist-packages/pydantic/_internal/_dataclasses.py", line 121, in __init__
(APIServer pid=295080)     s.__pydantic_validator__.validate_python(ArgsKwargs(args, kwargs), self_instance=s)
(APIServer pid=295080) pydantic_core._pydantic_core.ValidationError: 1 validation error for VllmConfig
(APIServer pid=295080)   Value error, The repository /workspace/amd/MiniMax-M2.1-MXFP4 contains custom code which must be executed to correctly load the model. You can inspect the repository content at /workspace/amd/MiniMax-M2.1-MXFP4 .
(APIServer pid=295080)  You can inspect the repository content at https://hf.co//workspace/amd/MiniMax-M2.1-MXFP4.
(APIServer pid=295080) Please pass the argument `trust_remote_code=True` to allow custom code to be run. [type=value_error, input_value=ArgsKwargs((), {'model_co... 'shutdown_timeout': 0}), input_type=ArgsKwargs]

Test Plan & Result

After fixing:

vllm (pretrained=/workspace/amd/MiniMax-M2.1-MXFP4,tensor_parallel_size=2,dtype=auto,gpu_memory_utilization=0.9,enforce_eager=True,trust_remote_code=True,max_model_len=32768), gen_kwargs: (None), limit: None, num_fewshot: 5, batch_size: auto
|    Tasks     |Version|     Filter     |n-shot|  Metric   |   |Value |   |Stderr|
|--------------|------:|----------------|-----:|-----------|---|-----:|---|-----:|
|gsm8k_platinum|      3|flexible-extract|     5|exact_match|↑  |0.9603|±  |0.0056|
|              |       |strict-match    |     5|exact_match|↑  |0.9570|±  |0.0058|

Changed files

vllm/model_executor/layers/quantization/quark/quark.py (modified, +33/-6)

Code Example

Collecting environment information...
==============================
        System Info
==============================
OS                           : Ubuntu 22.04.5 LTS (x86_64)
GCC version                  : (Ubuntu 11.4.0-1ubuntu1~22.04.2) 11.4.0
Clang version                : 20.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-7.0.0 25314 f4087f6b428f0e6f575ebac8a8a724dab123d06e)
CMake version                : version 3.31.10
Libc version                 : glibc-2.35

==============================
       PyTorch Info
==============================
PyTorch version              : 2.9.1+git8907517
Is debug build               : False
CUDA used to build PyTorch   : N/A
ROCM used to build PyTorch   : 7.0.51831-a3e329ad8

==============================
      Python Environment
==============================
Python version               : 3.12.12 (main, Oct 10 2025, 08:52:57) [GCC 11.4.0] (64-bit runtime)
Python platform              : Linux-6.8.0-101-generic-x86_64-with-glibc2.35

==============================
       CUDA / GPU Info
==============================
Is CUDA available            : True
CUDA runtime version         : Could not collect
CUDA_MODULE_LOADING set to   : 
GPU models and configuration :  (gfx942:sramecc+:xnack-)
Nvidia driver version        : Could not collect
cuDNN version                : Could not collect
HIP runtime version          : 7.0.51831
MIOpen runtime version       : 3.5.0
Is XNNPACK available         : True

==============================
          CPU Info
==============================
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):                                  256
On-line CPU(s) list:                     0-255
Vendor ID:                               AuthenticAMD
Model name:                              AMD EPYC 9555 64-Core Processor
CPU family:                              26
Model:                                   2
Thread(s) per core:                      2
Core(s) per socket:                      64
Socket(s):                               2
Stepping:                                1
Frequency boost:                         enabled
CPU max MHz:                             4409.3750
CPU min MHz:                             1500.0000
BogoMIPS:                                6399.98
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 amd_lbr_v2 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 hw_pstate ssbd mba perfmon_v2 ibrs ibpb stibp ibrs_enhanced vmmcall fsgsbase tsc_adjust 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 user_shstk avx_vnni avx512_bf16 clzero irperf xsaveerptr rdpru wbnoinvd amd_ppin cppc amd_ibpb_ret arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold avic v_vmsave_vmload vgif x2avic v_spec_ctrl vnmi avx512vbmi umip pku ospke avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg avx512_vpopcntdq la57 rdpid bus_lock_detect movdiri movdir64b overflow_recov succor smca fsrm avx512_vp2intersect flush_l1d debug_swap
Virtualization:                          AMD-V
L1d cache:                               6 MiB (128 instances)
L1i cache:                               4 MiB (128 instances)
L2 cache:                                128 MiB (128 instances)
L3 cache:                                512 MiB (16 instances)
NUMA node(s):                            2
NUMA node0 CPU(s):                       0-63,128-191
NUMA node1 CPU(s):                       64-127,192-255
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: 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 Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; Enhanced / Automatic IBRS; IBPB conditional; STIBP always-on; PBRSB-eIBRS Not affected; BHI Not affected
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

==============================
Versions of relevant libraries
==============================
[pip3] conch-triton-kernels==1.2.1
[pip3] numpy==2.1.3
[pip3] onnx==1.19.0
[pip3] onnx-ir==0.2.0
[pip3] onnxscript==0.6.2
[pip3] onnxslim==0.1.86
[pip3] pyzmq==27.1.0
[pip3] torch==2.9.1+git8907517
[pip3] torchaudio==2.9.0+eaa9e4e
[pip3] torchvision==0.24.1+d801a34
[pip3] **transformers==4.57.6**
[pip3] triton==3.4.0
[pip3] triton_kernels==1.0.0
[conda] Could not collect

==============================
         vLLM Info
==============================
ROCM Version                 : 7.0.51831-a3e329ad8
vLLM Version                 : 0.17.1rc1.dev34+g0b4aea4d4.d20260314 (git sha: 0b4aea4d4, date: 20260314)
vLLM Build Flags:
  CUDA Archs: Not Set; ROCm: Disabled
GPU Topology:
  ============================ ROCm System Management Interface ============================
================================ Weight between two GPUs =================================
       GPU0         GPU1         GPU2         GPU3         GPU4         GPU5         GPU6         GPU7         
GPU0   0            15           15           15           15           15           15           15           
GPU1   15           0            15           15           15           15           15           15           
GPU2   15           15           0            15           15           15           15           15           
GPU3   15           15           15           0            15           15           15           15           
GPU4   15           15           15           15           0            15           15           15           
GPU5   15           15           15           15           15           0            15           15           
GPU6   15           15           15           15           15           15           0            15           
GPU7   15           15           15           15           15           15           15           0            

================================= Hops between two GPUs ==================================
       GPU0         GPU1         GPU2         GPU3         GPU4         GPU5         GPU6         GPU7         
GPU0   0            1            1            1            1            1            1            1            
GPU1   1            0            1            1            1            1            1            1            
GPU2   1            1            0            1            1            1            1            1            
GPU3   1            1            1            0            1            1            1            1            
GPU4   1            1            1            1            0            1            1            1            
GPU5   1            1            1            1            1            0            1            1            
GPU6   1            1            1            1            1            1            0            1            
GPU7   1            1            1            1            1            1            1            0            

=============================== Link Type between two GPUs ===============================
       GPU0         GPU1         GPU2         GPU3         GPU4         GPU5         GPU6         GPU7         
GPU0   0            XGMI         XGMI         XGMI         XGMI         XGMI         XGMI         XGMI         
GPU1   XGMI         0            XGMI         XGMI         XGMI         XGMI         XGMI         XGMI         
GPU2   XGMI         XGMI         0            XGMI         XGMI         XGMI         XGMI         XGMI         
GPU3   XGMI         XGMI         XGMI         0            XGMI         XGMI         XGMI         XGMI         
GPU4   XGMI         XGMI         XGMI         XGMI         0            XGMI         XGMI         XGMI         
GPU5   XGMI         XGMI         XGMI         XGMI         XGMI         0            XGMI         XGMI         
GPU6   XGMI         XGMI         XGMI         XGMI         XGMI         XGMI         0            XGMI         
GPU7   XGMI         XGMI         XGMI         XGMI         XGMI         XGMI         XGMI         0            

======================================= Numa Nodes =======================================
GPU[0]		: (Topology) Numa Node: 0
GPU[0]		: (Topology) Numa Affinity: 0
GPU[1]		: (Topology) Numa Node: 0
GPU[1]		: (Topology) Numa Affinity: 0
GPU[2]		: (Topology) Numa Node: 0
GPU[2]		: (Topology) Numa Affinity: 0
GPU[3]		: (Topology) Numa Node: 0
GPU[3]		: (Topology) Numa Affinity: 0
GPU[4]		: (Topology) Numa Node: 1
GPU[4]		: (Topology) Numa Affinity: 1
GPU[5]		: (Topology) Numa Node: 1
GPU[5]		: (Topology) Numa Affinity: 1
GPU[6]		: (Topology) Numa Node: 1
GPU[6]		: (Topology) Numa Affinity: 1
GPU[7]		: (Topology) Numa Node: 1
GPU[7]		: (Topology) Numa Affinity: 1
================================== End of ROCm SMI Log ===================================

==============================
     Environment Variables
==============================
PYTORCH_ROCM_ARCH=gfx90a;gfx942;gfx950;gfx1100;gfx1101;gfx1200;gfx1201;gfx1150;gfx1151
LD_LIBRARY_PATH=/opt/rocm/lib:/usr/local/lib:
PYTORCH_NVML_BASED_CUDA_CHECK=1
TORCHINDUCTOR_COMPILE_THREADS=1

RAW_BUFFERClick to expand / collapse

Your current environment

<details> <summary>The output of <code>python collect_env.py</code></summary>

Collecting environment information...
==============================
        System Info
==============================
OS                           : Ubuntu 22.04.5 LTS (x86_64)
GCC version                  : (Ubuntu 11.4.0-1ubuntu1~22.04.2) 11.4.0
Clang version                : 20.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-7.0.0 25314 f4087f6b428f0e6f575ebac8a8a724dab123d06e)
CMake version                : version 3.31.10
Libc version                 : glibc-2.35

==============================
       PyTorch Info
==============================
PyTorch version              : 2.9.1+git8907517
Is debug build               : False
CUDA used to build PyTorch   : N/A
ROCM used to build PyTorch   : 7.0.51831-a3e329ad8

==============================
      Python Environment
==============================
Python version               : 3.12.12 (main, Oct 10 2025, 08:52:57) [GCC 11.4.0] (64-bit runtime)
Python platform              : Linux-6.8.0-101-generic-x86_64-with-glibc2.35

==============================
       CUDA / GPU Info
==============================
Is CUDA available            : True
CUDA runtime version         : Could not collect
CUDA_MODULE_LOADING set to   : 
GPU models and configuration :  (gfx942:sramecc+:xnack-)
Nvidia driver version        : Could not collect
cuDNN version                : Could not collect
HIP runtime version          : 7.0.51831
MIOpen runtime version       : 3.5.0
Is XNNPACK available         : True

==============================
          CPU Info
==============================
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):                                  256
On-line CPU(s) list:                     0-255
Vendor ID:                               AuthenticAMD
Model name:                              AMD EPYC 9555 64-Core Processor
CPU family:                              26
Model:                                   2
Thread(s) per core:                      2
Core(s) per socket:                      64
Socket(s):                               2
Stepping:                                1
Frequency boost:                         enabled
CPU max MHz:                             4409.3750
CPU min MHz:                             1500.0000
BogoMIPS:                                6399.98
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 amd_lbr_v2 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 hw_pstate ssbd mba perfmon_v2 ibrs ibpb stibp ibrs_enhanced vmmcall fsgsbase tsc_adjust 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 user_shstk avx_vnni avx512_bf16 clzero irperf xsaveerptr rdpru wbnoinvd amd_ppin cppc amd_ibpb_ret arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold avic v_vmsave_vmload vgif x2avic v_spec_ctrl vnmi avx512vbmi umip pku ospke avx512_vbmi2 gfni vaes vpclmulqdq avx512_vnni avx512_bitalg avx512_vpopcntdq la57 rdpid bus_lock_detect movdiri movdir64b overflow_recov succor smca fsrm avx512_vp2intersect flush_l1d debug_swap
Virtualization:                          AMD-V
L1d cache:                               6 MiB (128 instances)
L1i cache:                               4 MiB (128 instances)
L2 cache:                                128 MiB (128 instances)
L3 cache:                                512 MiB (16 instances)
NUMA node(s):                            2
NUMA node0 CPU(s):                       0-63,128-191
NUMA node1 CPU(s):                       64-127,192-255
Vulnerability Gather data sampling:      Not affected
Vulnerability Indirect target selection: 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 Reg file data sampling:    Not affected
Vulnerability Retbleed:                  Not affected
Vulnerability Spec rstack overflow:      Not affected
Vulnerability Spec store bypass:         Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1:                Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2:                Mitigation; Enhanced / Automatic IBRS; IBPB conditional; STIBP always-on; PBRSB-eIBRS Not affected; BHI Not affected
Vulnerability Srbds:                     Not affected
Vulnerability Tsa:                       Not affected
Vulnerability Tsx async abort:           Not affected
Vulnerability Vmscape:                   Not affected

==============================
Versions of relevant libraries
==============================
[pip3] conch-triton-kernels==1.2.1
[pip3] numpy==2.1.3
[pip3] onnx==1.19.0
[pip3] onnx-ir==0.2.0
[pip3] onnxscript==0.6.2
[pip3] onnxslim==0.1.86
[pip3] pyzmq==27.1.0
[pip3] torch==2.9.1+git8907517
[pip3] torchaudio==2.9.0+eaa9e4e
[pip3] torchvision==0.24.1+d801a34
[pip3] **transformers==4.57.6**
[pip3] triton==3.4.0
[pip3] triton_kernels==1.0.0
[conda] Could not collect

==============================
         vLLM Info
==============================
ROCM Version                 : 7.0.51831-a3e329ad8
vLLM Version                 : 0.17.1rc1.dev34+g0b4aea4d4.d20260314 (git sha: 0b4aea4d4, date: 20260314)
vLLM Build Flags:
  CUDA Archs: Not Set; ROCm: Disabled
GPU Topology:
  ============================ ROCm System Management Interface ============================
================================ Weight between two GPUs =================================
       GPU0         GPU1         GPU2         GPU3         GPU4         GPU5         GPU6         GPU7         
GPU0   0            15           15           15           15           15           15           15           
GPU1   15           0            15           15           15           15           15           15           
GPU2   15           15           0            15           15           15           15           15           
GPU3   15           15           15           0            15           15           15           15           
GPU4   15           15           15           15           0            15           15           15           
GPU5   15           15           15           15           15           0            15           15           
GPU6   15           15           15           15           15           15           0            15           
GPU7   15           15           15           15           15           15           15           0            

================================= Hops between two GPUs ==================================
       GPU0         GPU1         GPU2         GPU3         GPU4         GPU5         GPU6         GPU7         
GPU0   0            1            1            1            1            1            1            1            
GPU1   1            0            1            1            1            1            1            1            
GPU2   1            1            0            1            1            1            1            1            
GPU3   1            1            1            0            1            1            1            1            
GPU4   1            1            1            1            0            1            1            1            
GPU5   1            1            1            1            1            0            1            1            
GPU6   1            1            1            1            1            1            0            1            
GPU7   1            1            1            1            1            1            1            0            

=============================== Link Type between two GPUs ===============================
       GPU0         GPU1         GPU2         GPU3         GPU4         GPU5         GPU6         GPU7         
GPU0   0            XGMI         XGMI         XGMI         XGMI         XGMI         XGMI         XGMI         
GPU1   XGMI         0            XGMI         XGMI         XGMI         XGMI         XGMI         XGMI         
GPU2   XGMI         XGMI         0            XGMI         XGMI         XGMI         XGMI         XGMI         
GPU3   XGMI         XGMI         XGMI         0            XGMI         XGMI         XGMI         XGMI         
GPU4   XGMI         XGMI         XGMI         XGMI         0            XGMI         XGMI         XGMI         
GPU5   XGMI         XGMI         XGMI         XGMI         XGMI         0            XGMI         XGMI         
GPU6   XGMI         XGMI         XGMI         XGMI         XGMI         XGMI         0            XGMI         
GPU7   XGMI         XGMI         XGMI         XGMI         XGMI         XGMI         XGMI         0            

======================================= Numa Nodes =======================================
GPU[0]		: (Topology) Numa Node: 0
GPU[0]		: (Topology) Numa Affinity: 0
GPU[1]		: (Topology) Numa Node: 0
GPU[1]		: (Topology) Numa Affinity: 0
GPU[2]		: (Topology) Numa Node: 0
GPU[2]		: (Topology) Numa Affinity: 0
GPU[3]		: (Topology) Numa Node: 0
GPU[3]		: (Topology) Numa Affinity: 0
GPU[4]		: (Topology) Numa Node: 1
GPU[4]		: (Topology) Numa Affinity: 1
GPU[5]		: (Topology) Numa Node: 1
GPU[5]		: (Topology) Numa Affinity: 1
GPU[6]		: (Topology) Numa Node: 1
GPU[6]		: (Topology) Numa Affinity: 1
GPU[7]		: (Topology) Numa Node: 1
GPU[7]		: (Topology) Numa Affinity: 1
================================== End of ROCm SMI Log ===================================

==============================
     Environment Variables
==============================
PYTORCH_ROCM_ARCH=gfx90a;gfx942;gfx950;gfx1100;gfx1101;gfx1200;gfx1201;gfx1150;gfx1151
LD_LIBRARY_PATH=/opt/rocm/lib:/usr/local/lib:
PYTORCH_NVML_BASED_CUDA_CHECK=1
TORCHINDUCTOR_COMPILE_THREADS=1

</details>

🐛 Describe the bug

Community user reported GPU hang issue for model amd/MiniMax-M2.1-MXFP4 with TP=1 on MI300. However, using TP>1 (e.g., 2) can successfully make the server launched.

To help users pass "good" serving CLI args, Quark can emit relevant infos or warnings at an early stage before GPU hang happens, particularly for emulation cases,

cc @BowenBao @fxmarty-amd

Before submitting a new issue...

Make sure you already searched for relevant issues, and asked the chatbot living at the bottom right corner of the documentation page, which can answer lots of frequently asked questions./

extent analysis

Fix Plan

To address the GPU hang issue for the model amd/MiniMax-M2.1-MXFP4 on MI300 with TP=1, we need to modify Quark to emit warnings for emulation cases that may lead to heavy memory pressure. Here are the steps:

Step 1: Check for Emulation Cases
- Identify when Quark is using the emulative (QDQ) path for MXFP4 dequantization to BF16.
- Check if the model size is close to the available GPU memory.
Step 2: Emit Warnings
- Add a warning message when Quark detects an emulation case that may lead to heavy memory pressure.
- Suggest using TP>1 (e.g., 2) to reduce memory usage.
Step 3: Modify Quark Code

import warnings

... (rest of the code)

if using_emulation and model_size > 0.8 * gpu_memory: warnings.warn("Emulation case detected. Model size is close to available GPU memory. Consider using TP>1 to reduce memory usage.")

* **Step 4: Test and Verify**
  * Test the modified Quark code with the `amd/MiniMax-M2.1-MXFP4` model on MI300 with TP=1.
  * Verify that the warning message is emitted and the GPU hang issue is resolved.

### Verification
To verify that the fix worked, check the following:

* The warning message is emitted when running the model with TP=1.
* The GPU hang issue is resolved when using TP>1 (e.g., 2).
* The model runs successfully without any memory-related issues.

### Extra Tips
* Consider adding a documentation page or FAQ section to inform users about the potential GPU hang issue and the recommended solution (using TP>1).
* Monitor user feedback and adjust the warning message and suggested solution as needed.

Vote matrix · Quick signals

Works

Did the solution work? Tap to confirm.

Easy Fix

Was it a quick fix?

Time Saver

Did it save you time?

Blocking

Was it severely blocking?

Common Issue

Are others likely hitting this too?

Flaky / Intermittent

Is it intermittent?

Verified / Reproducible

Can you reproduce it reliably?

#api #ssr #installation #tensor shape #environment variable #serialization error #model compatibility #GPU setup

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Data

Security

Network

Code

UI/UX

Text

System

Multimedia

Protocol

API

Engineering

vllm - ✅(Solved) Fix [Bug]: GPU hang on MI325/300 when serving MiniMax-M2.1-MXFP4 with TP=1 [1 pull requests, 3 comments, 3 participants]

Recommended Tools

GitHub issue graph ai analysis

Root Cause

Fix Action

Fix / Workaround

============================== CPU Info

PR fix notes

PR #37408: [ROCm][Quantization] fallback trust_remote_code=True in Quark config for some cases

Description (problem / solution / changelog)

Purpose

Test Plan & Result

Changed files

Code Example

Your current environment

🐛 Describe the bug

Before submitting a new issue...

extent analysis

Fix Plan

... (rest of the code)

Still need to ship something?

TRENDING

vllm - ✅(Solved) Fix [Bug]: GPU hang on MI325/300 when serving MiniMax-M2.1-MXFP4 with TP=1 [1 pull requests, 3 comments, 3 participants]

Recommended Tools

GitHub issue graph ai analysis

Root Cause

Fix Action

Fix / Workaround

============================== CPU Info

PR fix notes

PR #37408: [ROCm][Quantization] fallback trust_remote_code=True in Quark config for some cases

Description (problem / solution / changelog)

Purpose

Test Plan & Result

Changed files

Code Example

Your current environment

🐛 Describe the bug

Before submitting a new issue...

extent analysis

Fix Plan

... (rest of the code)

Still need to ship something?

RELATED_DISCOVERY

TRENDING