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

10
benchmarks/kernels/benchmark_mrope.py
View File

@@ -135,14 +135,14 @@ def benchmark_mrope(
key.clone(),
)
torch.cuda.synchronize()
torch.accelerator.synchronize()
# Time reference implementation
torch_times = []
for _ in range(benchmark_iter):
query_clone = query.clone()
key_clone = key.clone()
torch.cuda.synchronize()
torch.accelerator.synchronize()
start_time = time.time()
mrope_helper_class.forward_native(
@@ -151,7 +151,7 @@ def benchmark_mrope(
key_clone,
)
torch.cuda.synchronize()
torch.accelerator.synchronize()
torch_times.append(time.time() - start_time)
# Time triton kernel implementation
@@ -159,14 +159,14 @@ def benchmark_mrope(
for _ in range(benchmark_iter):
query_clone = query.clone()
key_clone = key.clone()
torch.cuda.synchronize()
torch.accelerator.synchronize()
start_time = time.time()
mrope_helper_class.forward_cuda(
positions,
query_clone,
key_clone,
)
torch.cuda.synchronize()
torch.accelerator.synchronize()
triton_times.append(time.time() - start_time)
# Calculate statistics