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 1, 18:45 UTC
[Bugfix] Fix benchmark_moe.py after inplace mechanism removal (#44041)
Signed-off-by: Qiuyang Yue <yueqiuyang1389@gmail.com>
Qiuyang Yue committed
Jun 6, 2026 at 21:32 UTC
9c7f7741d45154ca11455caa9875beb94e2e66b7
1 file changed
-3
benchmarks/kernels/benchmark_moe.py
-3
@@ -271,7 +271,6 @@ def benchmark_config(
271
moe_config=moe_config,
272
quant_config=quant_config,
273
),
274
- inplace=not disable_inplace(),
274
)
275
276
with override_config(config):
@@ -279,7 +278,6 @@ def benchmark_config(
278
x, input_gating, topk, renormalize=not use_deep_gemm
279
)
280
282
- inplace = not disable_inplace()
281
if use_deep_gemm:
282
return deep_gemm_experts.apply(
283
x,
@@ -298,7 +296,6 @@ def benchmark_config(
296
w2,
297
topk_weights,
298
topk_ids,
301
- inplace=inplace,
299
quant_config=quant_config,
300
)
301