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

6
benchmarks/kernels/benchmark_w8a8_block_fp8.py
View File

@@ -177,18 +177,18 @@ def benchmark_config(
def run():
w8a8_block_matmul(A, B, As, Bs, block_size, config, out_dtype)
torch.cuda.synchronize()
torch.accelerator.synchronize()
# JIT complication & warmup
for _ in range(5):
run()
torch.cuda.synchronize()
torch.accelerator.synchronize()
start_event = torch.Event(enable_timing=True)
end_event = torch.Event(enable_timing=True)
latencies: list[float] = []
for i in range(num_iters):
torch.cuda.synchronize()
torch.accelerator.synchronize()
start_event.record()
run()
end_event.record()