mirror of
https://github.com/tile-ai/tilelang.git
synced 2026-10-25 07:14:16 +08:00
Lei Wang
3c5190e019
* Removed redundant simplification step in vectorization logic to streamline performance. * Introduced a new test for atomic addition in TileLang, validating functionality with a reference implementation using PyTorch.
47 lines
1.4 KiB
Python
47 lines
1.4 KiB
Python
import tilelang.testing
|
|
import tilelang.language as T
|
|
|
|
|
|
def atomic_add_program(K, M, N, block_M, block_N, dtype="float"):
|
|
|
|
@T.prim_func
|
|
def atomic_add(A: T.Tensor((K, M, N), dtype), B: T.Tensor((M, N), dtype)):
|
|
with T.Kernel(T.ceildiv(M, block_M), T.ceildiv(N, block_N), K, threads=32) as (bx, by, bz):
|
|
A_shared = T.alloc_shared((block_M, block_N), dtype)
|
|
|
|
T.copy(A[bz, bx * block_M:(bx + 1) * block_M, by * block_N:(by + 1) * block_N],
|
|
A_shared)
|
|
|
|
for i, j in T.Parallel(block_M, block_N):
|
|
T.atomic_add(B[bx * block_M + i, by * block_N + j], A_shared[i, j])
|
|
|
|
return atomic_add
|
|
|
|
|
|
def run_atomic_add(K, M, N, block_M, block_N, dtype="float32"):
|
|
program = atomic_add_program(K, M, N, block_M, block_N, dtype=dtype)
|
|
kernel = tilelang.compile(program)
|
|
# print(kernel.get_kernel_source())
|
|
import torch
|
|
|
|
def ref_program(A, B):
|
|
for k in range(K):
|
|
for i in range(M):
|
|
for j in range(N):
|
|
B[i, j] += A[k, i, j]
|
|
|
|
A = torch.randn(K, M, N, dtype=getattr(torch, dtype)).cuda()
|
|
B = torch.zeros(M, N, dtype=getattr(torch, dtype)).cuda()
|
|
ref_B = B.clone()
|
|
ref_program(A, ref_B)
|
|
kernel(A, B)
|
|
torch.testing.assert_close(B, ref_B)
|
|
|
|
|
|
def test_atomic_add():
|
|
run_atomic_add(8, 128, 128, 32, 32)
|
|
|
|
|
|
if __name__ == "__main__":
|
|
tilelang.testing.main()
|