vllm - ✅(Solved) Fix [Bug][ROCm] MI355 + AITER MXFP4 MOE: `Unsupported kernel config for moe heuristic dispatch` [2 pull requests, 3 comments, 2 participants]

vllm2026-04-16 11:59:53

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#40008•Fetched 2026-04-17 08:27:40

View on GitHub

Comments

Participants

Timeline

Reactions

Author

fxmarty-amd

Participants

fxmarty-amd

github-actions[bot]

Timeline (top)

mentioned ×4subscribed ×4commented ×3labeled ×2

Error Message

(EngineCore pid=14719) File "/felmarty/repos/vllm/vllm/model_executor/layers/quantization/quark/quark_moe.py", line 1438, in apply (EngineCore pid=14719) return rocm_aiter_fused_experts( (EngineCore pid=14719) ^^^^^^^^^^^^^^^^^^^^^^^^^ (EngineCore pid=14719) File "/felmarty/repos/vllm/vllm/model_executor/layers/fused_moe/rocm_aiter_fused_moe.py", line 292, in rocm_aiter_fused_experts (EngineCore pid=14719) return rocm_aiter_ops.fused_moe( (EngineCore pid=14719) ^^^^^^^^^^^^^^^^^^^^^^^^^ (EngineCore pid=14719) File "/felmarty/repos/vllm/vllm/_aiter_ops.py", line 1624, in fused_moe (EngineCore pid=14719) return torch.ops.vllm.rocm_aiter_fused_moe( (EngineCore pid=14719) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ (EngineCore pid=14719) File "/usr/local/lib/python3.12/dist-packages/torch/_ops.py", line 1209, in call (EngineCore pid=14719) return self._op(*args, **kwargs) (EngineCore pid=14719) ^^^^^^^^^^^^^^^^^^^^^^^^^ (EngineCore pid=14719) File "/felmarty/repos/vllm/vllm/_aiter_ops.py", line 125, in rocm_aiter_fused_moe_impl (EngineCore pid=14719) return fused_moe( (EngineCore pid=14719) ^^^^^^^^^^ (EngineCore pid=14719) File "/usr/local/lib/python3.12/dist-packages/aiter/fused_moe.py", line 117, in fused_moe (EngineCore pid=14719) return fused_moe( (EngineCore pid=14719) ^^^^^^^^^^^ (EngineCore pid=14719) File "/usr/local/lib/python3.12/dist-packages/aiter/jit/utils/torch_guard.py", line 278, in wrapper_custom (EngineCore pid=14719) getattr(torch.ops.aiter, f"{loadName}")(*args, **kwargs) (EngineCore pid=14719) File "/usr/local/lib/python3.12/dist-packages/torch/_ops.py", line 1209, in call (EngineCore pid=14719) return self.op(*args, **kwargs) (EngineCore pid=14719) ^^^^^^^^^^^^^^^^^^^^^^^^^ (EngineCore pid=14719) File "/usr/local/lib/python3.12/dist-packages/aiter/jit/utils/torch_guard.py", line 301, in outer_wrapper (EngineCore pid=14719) wrapper(*args, **kwargs) (EngineCore pid=14719) File "/usr/local/lib/python3.12/dist-packages/aiter/jit/utils/torch_guard.py", line 196, in wrapper (EngineCore pid=14719) return func(*args, **kwargs) (EngineCore pid=14719) ^^^^^^^^^^^^^^^^^^^^^ (EngineCore pid=14719) File "/usr/local/lib/python3.12/dist-packages/aiter/fused_moe.py", line 300, in fused_moe (EngineCore pid=14719) return fused_moe_2stages( (EngineCore pid=14719) ^^^^^^^^^^^^^^^^^^ (EngineCore pid=14719) File "/usr/local/lib/python3.12/dist-packages/aiter/fused_moe.py", line 1029, in fused_moe_2stages (EngineCore pid=14719) a2 = metadata.stage1( (EngineCore pid=14719) ^^^^^^^^^^^^^^^^ (EngineCore pid=14719) File "/usr/local/lib/python3.12/dist-packages/aiter/fused_moe.py", line 1494, in ck_moe_stage1 (EngineCore pid=14719) aiter.ck_moe_stage1_fwd( (EngineCore pid=14719) File "/usr/local/lib/python3.12/dist-packages/aiter/ops/moe_op.py", line 542, in ck_moe_stage1_fwd (EngineCore pid=14719) ck_moe_stage1( (EngineCore pid=14719) File "/usr/local/lib/python3.12/dist-packages/aiter/jit/utils/torch_guard.py", line 278, in wrapper_custom (EngineCore pid=14719) getattr(torch.ops.aiter, f"{loadName}")(*args, **kwargs) (EngineCore pid=14719) File "/usr/local/lib/python3.12/dist-packages/torch/_ops.py", line 1209, in call (EngineCore pid=14719) return self._op(*args, **kwargs) (EngineCore pid=14719) ^^^^^^^^^^^^^^^^^^^^^^^^^ (EngineCore pid=14719) File "/usr/local/lib/python3.12/dist-packages/aiter/jit/utils/torch_guard.py", line 301, in outer_wrapper (EngineCore pid=14719) wrapper(*args, **kwargs) (EngineCore pid=14719) File "/usr/local/lib/python3.12/dist-packages/aiter/jit/utils/torch_guard.py", line 196, in wrapper (EngineCore pid=14719) return func(*args, **kwargs) (EngineCore pid=14719) ^^^^^^^^^^^^^^^^^^^^^ (EngineCore pid=14719) File "/usr/local/lib/python3.12/dist-packages/aiter/jit/core.py", line 970, in custom_wrapper (EngineCore pid=14719) return wrapper(*args, **kwargs) (EngineCore pid=14719) ^^^^^^^^^^^^^^^^^^^^^^^^ (EngineCore pid=14719) File "/usr/local/lib/python3.12/dist-packages/aiter/jit/core.py", line 966, in wrapper (EngineCore pid=14719) return op(*args, **kwargs) (EngineCore pid=14719) ^^^^^^^^^^^^^^^^^^^ (EngineCore pid=14719) RuntimeError: Unsupported kernel config for moe heuristic dispatch

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): 384 On-line CPU(s) list: 0-383 Vendor ID: AuthenticAMD Model name: AMD EPYC 9965 192-Core Processor CPU family: 26 Model: 17 Thread(s) per core: 1 Core(s) per socket: 192 Socket(s): 2 Stepping: 0 Frequency boost: enabled CPU max MHz: 3700.1951 CPU min MHz: 1500.0000 BogoMIPS: 4499.86 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 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 avx_vnni 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 bus_lock_detect movdiri movdir64b overflow_recov succor smca fsrm avx512_vp2intersect flush_l1d Virtualization: AMD-V L1d cache: 18 MiB (384 instances) L1i cache: 12 MiB (384 instances) L2 cache: 384 MiB (384 instances) L3 cache: 768 MiB (24 instances) NUMA node(s): 2 NUMA node0 CPU(s): 0-191 NUMA node1 CPU(s): 192-383 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 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 disabled, RSB filling, PBRSB-eIBRS Not affected Vulnerability Srbds: Not affected Vulnerability Tsx async abort: Not affected

(EngineCore pid=14719)   File "/felmarty/repos/vllm/vllm/model_executor/layers/quantization/quark/quark_moe.py", line 1438, in apply
(EngineCore pid=14719)     return rocm_aiter_fused_experts(
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/felmarty/repos/vllm/vllm/model_executor/layers/fused_moe/rocm_aiter_fused_moe.py", line 292, in rocm_aiter_fused_experts
(EngineCore pid=14719)     return rocm_aiter_ops.fused_moe(
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/felmarty/repos/vllm/vllm/_aiter_ops.py", line 1624, in fused_moe
(EngineCore pid=14719)     return torch.ops.vllm.rocm_aiter_fused_moe(
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/torch/_ops.py", line 1209, in __call__
(EngineCore pid=14719)     return self._op(*args, **kwargs)
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/felmarty/repos/vllm/vllm/_aiter_ops.py", line 125, in _rocm_aiter_fused_moe_impl
(EngineCore pid=14719)     return fused_moe(
(EngineCore pid=14719)            ^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/fused_moe.py", line 117, in fused_moe
(EngineCore pid=14719)     return fused_moe_(
(EngineCore pid=14719)            ^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/jit/utils/torch_guard.py", line 278, in wrapper_custom
(EngineCore pid=14719)     getattr(torch.ops.aiter, f"{loadName}")(*args, **kwargs)
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/torch/_ops.py", line 1209, in __call__
(EngineCore pid=14719)     return self._op(*args, **kwargs)
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/jit/utils/torch_guard.py", line 301, in outer_wrapper
(EngineCore pid=14719)     wrapper(*args, **kwargs)
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/jit/utils/torch_guard.py", line 196, in wrapper
(EngineCore pid=14719)     return func(*args, **kwargs)
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/fused_moe.py", line 300, in fused_moe_
(EngineCore pid=14719)     return fused_moe_2stages(
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/fused_moe.py", line 1029, in fused_moe_2stages
(EngineCore pid=14719)     a2 = metadata.stage1(
(EngineCore pid=14719)          ^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/fused_moe.py", line 1494, in ck_moe_stage1
(EngineCore pid=14719)     aiter.ck_moe_stage1_fwd(
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/ops/moe_op.py", line 542, in ck_moe_stage1_fwd
(EngineCore pid=14719)     ck_moe_stage1(
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/jit/utils/torch_guard.py", line 278, in wrapper_custom
(EngineCore pid=14719)     getattr(torch.ops.aiter, f"{loadName}")(*args, **kwargs)
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/torch/_ops.py", line 1209, in __call__
(EngineCore pid=14719)     return self._op(*args, **kwargs)
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/jit/utils/torch_guard.py", line 301, in outer_wrapper
(EngineCore pid=14719)     wrapper(*args, **kwargs)
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/jit/utils/torch_guard.py", line 196, in wrapper
(EngineCore pid=14719)     return func(*args, **kwargs)
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/jit/core.py", line 970, in custom_wrapper
(EngineCore pid=14719)     return wrapper(*args, **kwargs)
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/jit/core.py", line 966, in wrapper
(EngineCore pid=14719)     return op(*args, **kwargs)
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719) RuntimeError: Unsupported kernel config for moe heuristic dispatch

PR fix notes

PR #39688: [fix][MOE] Fix MOE experts `intermediate_size` dimension not being narrowed before weight loading

Repository: vllm-project/vllm
Author: fxmarty-amd
State: closed | merged: True
Link: https://github.com/vllm-project/vllm/pull/39688

Description (problem / solution / changelog)

Purpose

The PR https://github.com/vllm-project/vllm/pull/37010 modified _load_per_channel_weight_scale, _load_w13, _load_w2 logic to use a method _narrow_expert_data_for_padding to narrow the instantiated padded nn.Parameter before copying weights into there.

The introduced method handles the hidden size dimension only, whereas the previous logic correctly narrowed both the hidden size + intermediate size dimensions, that may both be padded:

https://github.com/vllm-project/vllm/blob/1b19bd758936496751432eccabf8adb7b5d8936a/vllm/model_executor/layers/fused_moe/oracle/mxfp4.py#L376-L400

The old logic correctly handled this:

        # Handle padding: if loaded_weight is smaller than expert_data (can happen
        # on last TP shard with padding), copy to top-left corner
        if expert_data.shape != loaded_weight.shape:
            expert_data = expert_data[
                : loaded_weight.shape[0], : loaded_weight.shape[1]
            ]

This PR adds narrowing on the intermediate size dimension, as it used to be before https://github.com/vllm-project/vllm/pull/37010 was merged.

Test Plan

CUDA_VISIBLE_DEVICES=7 pytest tests/quantization/test_quark.py -s -vvvvv -k "test_ocp_mx_wikitext_correctness"

on main:

(EngineCore pid=7164)   File "/felmarty/repos/vllm/vllm/model_executor/models/qwen2_moe.py", line 495, in load_weights
(EngineCore pid=7164)     weight_loader(
(EngineCore pid=7164)   File "/felmarty/repos/vllm/vllm/model_executor/layers/fused_moe/layer.py", line 1338, in weight_loader
(EngineCore pid=7164)     self._load_model_weight_or_group_weight_scale(
(EngineCore pid=7164)   File "/felmarty/repos/vllm/vllm/model_executor/layers/fused_moe/layer.py", line 817, in _load_model_weight_or_group_weight_scale
(EngineCore pid=7164)     self._load_w2(
(EngineCore pid=7164)   File "/felmarty/repos/vllm/vllm/model_executor/layers/fused_moe/layer.py", line 991, in _load_w2
(EngineCore pid=7164)     expert_data.copy_(loaded_weight)
(EngineCore pid=7164) RuntimeError: The size of tensor a (768) must match the size of tensor b (704) at non-singleton dimension 1

on this PR: passes.

Changed files

tests/kernels/moe/test_moe_weight_loading_padded.py (modified, +35/-0)
vllm/model_executor/layers/fused_moe/layer.py (modified, +28/-15)

Code Example

Collecting environment information...
==============================
        System Info
==============================
OS                           : Ubuntu 22.04.5 LTS (x86_64)
GCC version                  : (Ubuntu 11.4.0-1ubuntu1~22.04.3) 11.4.0
Clang version                : 22.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-7.2.1 26084 f58b06dce1f9c15707c5f808fd002e18c2accf7e)
CMake version                : version 3.31.10
Libc version                 : glibc-2.35

==============================
       PyTorch Info
==============================
PyTorch version              : 2.10.0+git8514f05
Is debug build               : False
CUDA used to build PyTorch   : N/A
ROCM used to build PyTorch   : 7.2.53211
XPU used to build PyTorch    : N/A

==============================
      Python Environment
==============================
Python version               : 3.12.13 (main, Mar  4 2026, 09:23:07) [GCC 11.4.0] (64-bit runtime)
Python platform              : Linux-5.15.0-70-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 :  (gfx950:sramecc+:xnack-)
Nvidia driver version        : Could not collect
cuDNN version                : Could not collect
HIP runtime version          : 7.2.53211
MIOpen runtime version       : 3.5.1
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):                          384
On-line CPU(s) list:             0-383
Vendor ID:                       AuthenticAMD
Model name:                      AMD EPYC 9965 192-Core Processor
CPU family:                      26
Model:                           17
Thread(s) per core:              1
Core(s) per socket:              192
Socket(s):                       2
Stepping:                        0
Frequency boost:                 enabled
CPU max MHz:                     3700.1951
CPU min MHz:                     1500.0000
BogoMIPS:                        4499.86
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 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 avx_vnni 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 bus_lock_detect movdiri movdir64b overflow_recov succor smca fsrm avx512_vp2intersect flush_l1d
Virtualization:                  AMD-V
L1d cache:                       18 MiB (384 instances)
L1i cache:                       12 MiB (384 instances)
L2 cache:                        384 MiB (384 instances)
L3 cache:                        768 MiB (24 instances)
NUMA node(s):                    2
NUMA node0 CPU(s):               0-191
NUMA node1 CPU(s):               192-383
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 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 disabled, RSB filling, PBRSB-eIBRS Not affected
Vulnerability Srbds:             Not affected
Vulnerability Tsx async abort:   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==0.1.91
[pip3] pyzmq==27.1.0
[pip3] torch==2.10.0+git8514f05
[pip3] torchaudio==2.9.0+eaa9e4e
[pip3] torchvision==0.24.1+d801a34
[pip3] transformers==5.5.3
[pip3] triton==3.6.0
[pip3] triton_kernels==1.0.0
[conda] Could not collect

==============================
         vLLM Info
==============================
ROCM Version                 : 7.2.53211-e1a6bc5663
vLLM Version                 : 0.19.1rc1.dev357+gbfc4f90eb (git sha: bfc4f90eb)
vLLM Build Flags:
  CUDA Archs: Not Set; ROCm: Disabled; XPU: 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_TUNABLEOP_ENABLED=0
PYTORCH_ROCM_ARCH=gfx950
LD_LIBRARY_PATH=/opt/rocm/lib:/usr/local/lib:
PYTORCH_NVML_BASED_CUDA_CHECK=1
TORCHINDUCTOR_COMPILE_THREADS=1
TORCHINDUCTOR_CACHE_DIR=/tmp/torchinductor_root

---

CUDA_VISIBLE_DEVICES="6,7" pytest tests/quantization/test_quark.py -s -vvvvv -k "test_ocp_mx_wikitext_correctness" --exitfirst

---

(EngineCore pid=14719)   File "/felmarty/repos/vllm/vllm/model_executor/layers/quantization/quark/quark_moe.py", line 1438, in apply
(EngineCore pid=14719)     return rocm_aiter_fused_experts(
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/felmarty/repos/vllm/vllm/model_executor/layers/fused_moe/rocm_aiter_fused_moe.py", line 292, in rocm_aiter_fused_experts
(EngineCore pid=14719)     return rocm_aiter_ops.fused_moe(
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/felmarty/repos/vllm/vllm/_aiter_ops.py", line 1624, in fused_moe
(EngineCore pid=14719)     return torch.ops.vllm.rocm_aiter_fused_moe(
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/torch/_ops.py", line 1209, in __call__
(EngineCore pid=14719)     return self._op(*args, **kwargs)
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/felmarty/repos/vllm/vllm/_aiter_ops.py", line 125, in _rocm_aiter_fused_moe_impl
(EngineCore pid=14719)     return fused_moe(
(EngineCore pid=14719)            ^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/fused_moe.py", line 117, in fused_moe
(EngineCore pid=14719)     return fused_moe_(
(EngineCore pid=14719)            ^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/jit/utils/torch_guard.py", line 278, in wrapper_custom
(EngineCore pid=14719)     getattr(torch.ops.aiter, f"{loadName}")(*args, **kwargs)
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/torch/_ops.py", line 1209, in __call__
(EngineCore pid=14719)     return self._op(*args, **kwargs)
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/jit/utils/torch_guard.py", line 301, in outer_wrapper
(EngineCore pid=14719)     wrapper(*args, **kwargs)
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/jit/utils/torch_guard.py", line 196, in wrapper
(EngineCore pid=14719)     return func(*args, **kwargs)
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/fused_moe.py", line 300, in fused_moe_
(EngineCore pid=14719)     return fused_moe_2stages(
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/fused_moe.py", line 1029, in fused_moe_2stages
(EngineCore pid=14719)     a2 = metadata.stage1(
(EngineCore pid=14719)          ^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/fused_moe.py", line 1494, in ck_moe_stage1
(EngineCore pid=14719)     aiter.ck_moe_stage1_fwd(
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/ops/moe_op.py", line 542, in ck_moe_stage1_fwd
(EngineCore pid=14719)     ck_moe_stage1(
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/jit/utils/torch_guard.py", line 278, in wrapper_custom
(EngineCore pid=14719)     getattr(torch.ops.aiter, f"{loadName}")(*args, **kwargs)
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/torch/_ops.py", line 1209, in __call__
(EngineCore pid=14719)     return self._op(*args, **kwargs)
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/jit/utils/torch_guard.py", line 301, in outer_wrapper
(EngineCore pid=14719)     wrapper(*args, **kwargs)
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/jit/utils/torch_guard.py", line 196, in wrapper
(EngineCore pid=14719)     return func(*args, **kwargs)
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/jit/core.py", line 970, in custom_wrapper
(EngineCore pid=14719)     return wrapper(*args, **kwargs)
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/jit/core.py", line 966, in wrapper
(EngineCore pid=14719)     return op(*args, **kwargs)
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719) RuntimeError: Unsupported kernel config for moe heuristic dispatch

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 11.4.0-1ubuntu1~22.04.3) 11.4.0
Clang version                : 22.0.0git (https://github.com/RadeonOpenCompute/llvm-project roc-7.2.1 26084 f58b06dce1f9c15707c5f808fd002e18c2accf7e)
CMake version                : version 3.31.10
Libc version                 : glibc-2.35

==============================
       PyTorch Info
==============================
PyTorch version              : 2.10.0+git8514f05
Is debug build               : False
CUDA used to build PyTorch   : N/A
ROCM used to build PyTorch   : 7.2.53211
XPU used to build PyTorch    : N/A

==============================
      Python Environment
==============================
Python version               : 3.12.13 (main, Mar  4 2026, 09:23:07) [GCC 11.4.0] (64-bit runtime)
Python platform              : Linux-5.15.0-70-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 :  (gfx950:sramecc+:xnack-)
Nvidia driver version        : Could not collect
cuDNN version                : Could not collect
HIP runtime version          : 7.2.53211
MIOpen runtime version       : 3.5.1
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):                          384
On-line CPU(s) list:             0-383
Vendor ID:                       AuthenticAMD
Model name:                      AMD EPYC 9965 192-Core Processor
CPU family:                      26
Model:                           17
Thread(s) per core:              1
Core(s) per socket:              192
Socket(s):                       2
Stepping:                        0
Frequency boost:                 enabled
CPU max MHz:                     3700.1951
CPU min MHz:                     1500.0000
BogoMIPS:                        4499.86
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 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 avx_vnni 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 bus_lock_detect movdiri movdir64b overflow_recov succor smca fsrm avx512_vp2intersect flush_l1d
Virtualization:                  AMD-V
L1d cache:                       18 MiB (384 instances)
L1i cache:                       12 MiB (384 instances)
L2 cache:                        384 MiB (384 instances)
L3 cache:                        768 MiB (24 instances)
NUMA node(s):                    2
NUMA node0 CPU(s):               0-191
NUMA node1 CPU(s):               192-383
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 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 disabled, RSB filling, PBRSB-eIBRS Not affected
Vulnerability Srbds:             Not affected
Vulnerability Tsx async abort:   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==0.1.91
[pip3] pyzmq==27.1.0
[pip3] torch==2.10.0+git8514f05
[pip3] torchaudio==2.9.0+eaa9e4e
[pip3] torchvision==0.24.1+d801a34
[pip3] transformers==5.5.3
[pip3] triton==3.6.0
[pip3] triton_kernels==1.0.0
[conda] Could not collect

==============================
         vLLM Info
==============================
ROCM Version                 : 7.2.53211-e1a6bc5663
vLLM Version                 : 0.19.1rc1.dev357+gbfc4f90eb (git sha: bfc4f90eb)
vLLM Build Flags:
  CUDA Archs: Not Set; ROCm: Disabled; XPU: 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_TUNABLEOP_ENABLED=0
PYTORCH_ROCM_ARCH=gfx950
LD_LIBRARY_PATH=/opt/rocm/lib:/usr/local/lib:
PYTORCH_NVML_BASED_CUDA_CHECK=1
TORCHINDUCTOR_COMPILE_THREADS=1
TORCHINDUCTOR_CACHE_DIR=/tmp/torchinductor_root

</details>

and vllm @ 10e49d263854daf6cf63472b9cd2039196022a59

🐛 Describe the bug

CUDA_VISIBLE_DEVICES="6,7" pytest tests/quantization/test_quark.py -s -vvvvv -k "test_ocp_mx_wikitext_correctness" --exitfirst

crashes with:

(EngineCore pid=14719)   File "/felmarty/repos/vllm/vllm/model_executor/layers/quantization/quark/quark_moe.py", line 1438, in apply
(EngineCore pid=14719)     return rocm_aiter_fused_experts(
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/felmarty/repos/vllm/vllm/model_executor/layers/fused_moe/rocm_aiter_fused_moe.py", line 292, in rocm_aiter_fused_experts
(EngineCore pid=14719)     return rocm_aiter_ops.fused_moe(
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/felmarty/repos/vllm/vllm/_aiter_ops.py", line 1624, in fused_moe
(EngineCore pid=14719)     return torch.ops.vllm.rocm_aiter_fused_moe(
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/torch/_ops.py", line 1209, in __call__
(EngineCore pid=14719)     return self._op(*args, **kwargs)
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/felmarty/repos/vllm/vllm/_aiter_ops.py", line 125, in _rocm_aiter_fused_moe_impl
(EngineCore pid=14719)     return fused_moe(
(EngineCore pid=14719)            ^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/fused_moe.py", line 117, in fused_moe
(EngineCore pid=14719)     return fused_moe_(
(EngineCore pid=14719)            ^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/jit/utils/torch_guard.py", line 278, in wrapper_custom
(EngineCore pid=14719)     getattr(torch.ops.aiter, f"{loadName}")(*args, **kwargs)
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/torch/_ops.py", line 1209, in __call__
(EngineCore pid=14719)     return self._op(*args, **kwargs)
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/jit/utils/torch_guard.py", line 301, in outer_wrapper
(EngineCore pid=14719)     wrapper(*args, **kwargs)
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/jit/utils/torch_guard.py", line 196, in wrapper
(EngineCore pid=14719)     return func(*args, **kwargs)
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/fused_moe.py", line 300, in fused_moe_
(EngineCore pid=14719)     return fused_moe_2stages(
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/fused_moe.py", line 1029, in fused_moe_2stages
(EngineCore pid=14719)     a2 = metadata.stage1(
(EngineCore pid=14719)          ^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/fused_moe.py", line 1494, in ck_moe_stage1
(EngineCore pid=14719)     aiter.ck_moe_stage1_fwd(
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/ops/moe_op.py", line 542, in ck_moe_stage1_fwd
(EngineCore pid=14719)     ck_moe_stage1(
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/jit/utils/torch_guard.py", line 278, in wrapper_custom
(EngineCore pid=14719)     getattr(torch.ops.aiter, f"{loadName}")(*args, **kwargs)
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/torch/_ops.py", line 1209, in __call__
(EngineCore pid=14719)     return self._op(*args, **kwargs)
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/jit/utils/torch_guard.py", line 301, in outer_wrapper
(EngineCore pid=14719)     wrapper(*args, **kwargs)
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/jit/utils/torch_guard.py", line 196, in wrapper
(EngineCore pid=14719)     return func(*args, **kwargs)
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/jit/core.py", line 970, in custom_wrapper
(EngineCore pid=14719)     return wrapper(*args, **kwargs)
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719)   File "/usr/local/lib/python3.12/dist-packages/aiter/jit/core.py", line 966, in wrapper
(EngineCore pid=14719)     return op(*args, **kwargs)
(EngineCore pid=14719)            ^^^^^^^^^^^^^^^^^^^
(EngineCore pid=14719) RuntimeError: Unsupported kernel config for moe heuristic dispatch

This test:

should be parameterized over both the AITER & emulation backends
AITER backend should not crash
this test should be running in the AMD CI, but given this failure and the two weeks long failure at https://github.com/vllm-project/vllm/pull/39688, it is visibly not (for some reason)

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

TL;DR

The test test_ocp_mx_wikitext_correctness crashes with a RuntimeError: Unsupported kernel config for moe heuristic dispatch when run with the AITER backend on AMD CI, indicating a potential issue with the kernel configuration or the moe heuristic dispatch.

Guidance

Verify that the AITER backend is properly configured and installed, and that the necessary dependencies are met.
Check the kernel configuration and ensure that it is compatible with the moe heuristic dispatch.
Investigate the rocm_aiter_fused_moe function and the fused_moe function to see if there are any issues with the implementation or the input parameters.
Consider parameterizing the test over both the AITER and emulation backends to ensure that the test is running correctly in the AMD CI.

Example

No code example is provided as the issue is related to a specific test and backend configuration.

Notes

The issue seems to be related to the AITER backend and the moe heuristic dispatch, and it's not clear if this is a problem with the test itself or with the backend configuration. Further investigation is needed to determine the root cause of the issue.

Recommendation

Apply a workaround by modifying the test to use a different backend or kernel configuration, or by updating the AITER backend to support the necessary kernel configuration.

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 #batch processing #GPU compatibility #latency issue #environment variable

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][ROCm] MI355 + AITER MXFP4 MOE: `Unsupported kernel config for moe heuristic dispatch` [2 pull requests, 3 comments, 2 participants]

Recommended Tools

GitHub issue graph ai analysis

Error Message

Fix Action

Fix / Workaround

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

PR fix notes

PR #39688: [fix][MOE] Fix MOE experts `intermediate_size` dimension not being narrowed before weight loading

Description (problem / solution / changelog)

Purpose

Test Plan

Changed files

Code Example

Your current environment

🐛 Describe the bug

Before submitting a new issue...

extent analysis

TL;DR

Guidance

Example

Notes

Recommendation

Still need to ship something?

TRENDING

vllm - ✅(Solved) Fix [Bug][ROCm] MI355 + AITER MXFP4 MOE: `Unsupported kernel config for moe heuristic dispatch` [2 pull requests, 3 comments, 2 participants]

Recommended Tools

GitHub issue graph ai analysis

Error Message

Fix Action

Fix / Workaround

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

PR fix notes

PR #39688: [fix][MOE] Fix MOE experts intermediate_size dimension not being narrowed before weight loading

Description (problem / solution / changelog)

Purpose

Test Plan

Changed files

Code Example

Your current environment

🐛 Describe the bug

Before submitting a new issue...

extent analysis

TL;DR

Guidance

Example

Notes

Recommendation

Still need to ship something?

RELATED_DISCOVERY

TRENDING

PR #39688: [fix][MOE] Fix MOE experts `intermediate_size` dimension not being narrowed before weight loading