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 16, 03:15 UTC
[Misc] sort torch profiler table by kernel timing (#11813)
Divakar Verma committed
Jan 7, 2025 at 20:57 UTC
4d29e91be84d27ca313d657eee92c067439a4c23
1 file changed
+1
-1
benchmarks/benchmark_latency.py
+1
-1
@@ -52,7 +52,7 @@ def main(args: argparse.Namespace):
52
llm.generate(dummy_prompts,
53
sampling_params=sampling_params,
54
use_tqdm=False)
55
- print(p.key_averages())
55
+ print(p.key_averages().table(sort_by="self_cuda_time_total"))
56
else:
57
start_time = time.perf_counter()
58
llm.generate(dummy_prompts,