[TESTING] Added infrastructure for executing TTGIR program and test for layout conversions (#885)

This commit is contained in:
Philippe Tillet
2022-11-18 07:46:45 +01:00
committed by GitHub
parent 9ea6135eb5
commit dab4855bdf
6 changed files with 243 additions and 67 deletions

View File

@@ -16,7 +16,7 @@ def test_empty_kernel_cubin_compile():
device = torch.cuda.current_device()
kernel = triton.compile(empty_kernel,
"*fp32,i32,i32",
signature="*fp32,i32,i32",
device=device,
constants={"BLOCK": 256})