A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 3, 16:30 UTC
Use hidden_pad and intermediate_pad from vLLM #34301 (#42098)
Signed-off-by: Rebecca Lee <Rebecca.Lee@amd.com>
Rebecca Lee committed
May 13, 2026 at 23:21 UTC
fd7d858c8ac59cd962cb289d059ab29d27ae3a6f
1 file changed
+2
-2
vllm/model_executor/layers/fused_moe/experts/rocm_aiter_moe.py
+2
-2
@@ -356,8 +356,8 @@ def rocm_aiter_fused_experts(
356
doweight_stage1=apply_router_weight_on_input,
357
num_local_tokens=num_local_tokens,
358
output_dtype=output_dtype,
359
- hidden_pad=hidden_pad,
360
- intermediate_pad=intermediate_pad,
359
+ hidden_pad=hidden_pad // 128 * 128,
360
+ intermediate_pad=intermediate_pad // 64 * 64 * 2,
361
bias1=quant_config.w1_bias if quant_config.use_mxfp4_w4a16 else None,
362
bias2=quant_config.w2_bias if quant_config.use_mxfp4_w4a16 else None,
363
)