biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Hardware] Replace torch.cuda.synchronize() api with torch.accelerator.synchronize (#36085)

Browse Source 
Signed-off-by: Kunshang Ji <kunshang.ji@intel.com>
This commit is contained in:
Kunshang Ji
2026-03-05 18:36:39 +08:00
committed by GitHub
parent 0bfa229bf1
commit 66a2209645
59 changed files with 158 additions and 161 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
benchmarks/kernels/benchmark_moe.py
View File

@@ -304,19 +304,19 @@ def benchmark_config(
# JIT compilation & warmup
run()
torch.cuda.synchronize()
torch.accelerator.synchronize()
# Capture 10 invocations with CUDA graph
graph = torch.cuda.CUDAGraph()
with torch.cuda.graph(graph):
for _ in range(10):
run()
torch.cuda.synchronize()
torch.accelerator.synchronize()
# Warmup
for _ in range(5):
graph.replay()
torch.cuda.synchronize()
torch.accelerator.synchronize()
start_event = torch.Event(enable_timing=True)
end_event = torch.Event(enable_timing=True)
@@ -324,7 +324,7 @@ def benchmark_config(
latencies: list[float] = []
for i in range(num_iters):
prepare(i)
torch.cuda.synchronize()
torch.accelerator.synchronize()
start_event.record()
graph.replay()