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

Signed-off-by: Kunshang Ji <kunshang.ji@intel.com>
2026-03-05 18:36:39 +08:00
parent 0bfa229bf1
commit 66a2209645
59 changed files with 158 additions and 161 deletions
--- a/benchmarks/kernels/benchmark_mla_k_concat.py
+++ b/benchmarks/kernels/benchmark_mla_k_concat.py
@@ -47,13 +47,13 @@ def benchmark_method(
    # Warmup
    for _ in range(num_warmup):
        _ = method(k_nope, k_pe)
-    torch.cuda.synchronize()
+    torch.accelerator.synchronize()

    # Benchmark
    start = time.perf_counter()
    for _ in range(num_iters):
        _ = method(k_nope, k_pe)
-    torch.cuda.synchronize()
+    torch.accelerator.synchronize()
    end = time.perf_counter()

    return (end - start) / num_iters * 1000  # Convert to ms