mirror of
https://github.com/tile-ai/tilelang.git
synced 2026-10-25 07:14:16 +08:00
Tong WU
2e854b068e
* support cluster launch for tvm_ffi and cython backend * lint * add test * lint * update tvm * Add cluster synchronization operations and related functions - Introduced new built-in operations for cluster synchronization: `cluster_arrive_relaxed`, `cluster_arrive`, `cluster_wait`, `cluster_sync`, and `block_rank_in_cluster`. - Updated the CUDA code generator to handle these new operations. - Added corresponding Python bindings and documentation for the new cluster functions. - Included necessary header files for cluster operations in the CUDA code generation process. * add test for intrinsics * lint * remove legacy code * Add cluster barrier support and related functionality - Introduced `ptx_arrive_cluster_barrier` built-in for arriving at cluster barriers. - Added `alloc_cluster_barrier` function for allocating cluster barrier buffers. - Updated CUDA code generation to handle the new cluster barrier operations. - Enhanced existing functions to support shared and cluster barrier scopes. - Added tests for cluster barrier functionality in the TileLang testing suite. * lint * fix typo * add storage sync for cluster
110 lines
3.5 KiB
Python
110 lines
3.5 KiB
Python
import tilelang
|
|
import tilelang.language as T
|
|
import tilelang.testing
|
|
import torch
|
|
|
|
|
|
def matmul(M, N, K, block_M, block_N, block_K, dtype=T.float16, accum_dtype=T.float32):
|
|
@T.prim_func
|
|
def gemm(
|
|
A: T.Tensor((M, K), dtype),
|
|
B: T.Tensor((K, N), dtype),
|
|
C: T.Tensor((M, N), dtype),
|
|
):
|
|
with T.Kernel(T.ceildiv(N, block_N), T.ceildiv(M, block_M), threads=128, cluster_dims=(2, 1, 1)) as (bx, by):
|
|
A_shared = T.alloc_shared((block_M, block_K), dtype)
|
|
B_shared = T.alloc_shared((block_K, block_N), dtype)
|
|
C_local = T.alloc_fragment((block_M, block_N), accum_dtype)
|
|
|
|
T.clear(C_local)
|
|
for k in T.Pipelined(T.ceildiv(K, block_K), num_stages=3):
|
|
T.copy(A[by * block_M, k * block_K], A_shared)
|
|
T.copy(B[k * block_K, bx * block_N], B_shared)
|
|
T.gemm(A_shared, B_shared, C_local)
|
|
|
|
T.copy(C_local, C[by * block_M, bx * block_N])
|
|
|
|
return gemm
|
|
|
|
|
|
@tilelang.jit(out_idx=-1)
|
|
def get_cta_rank_in_cluster(cluster_size=4):
|
|
assert 128 % cluster_size == 0
|
|
|
|
@T.prim_func
|
|
def main(A: T.Tensor((128), T.int32)):
|
|
with T.Kernel(128, cluster_dims=(cluster_size, 1, 1)) as bx:
|
|
T.cluster_arrive()
|
|
T.cluster_wait()
|
|
cta_rank_in_cluster = T.block_rank_in_cluster()
|
|
if T.get_thread_binding() == 0:
|
|
A[bx] = cta_rank_in_cluster
|
|
T.cluster_sync()
|
|
|
|
return main
|
|
|
|
|
|
@tilelang.jit(out_idx=-1)
|
|
def barrier_kernel():
|
|
@T.prim_func
|
|
def main(A: T.Tensor((128), T.int32)):
|
|
with T.Kernel(128, threads=128, cluster_dims=(4, 1, 1)):
|
|
mbar = T.alloc_cluster_barrier([256])
|
|
T.cluster_sync()
|
|
T.mbarrier_arrive(mbar, 0)
|
|
if T.block_rank_in_cluster() == 0:
|
|
T.mbarrier_wait_parity(mbar, 0)
|
|
T.cluster_sync()
|
|
|
|
return main
|
|
|
|
|
|
def run_cython_cluster_launch():
|
|
kernel = matmul(1024, 1024, 1024, 128, 128, 32)
|
|
mod = tilelang.compile(kernel, execution_backend="cython")
|
|
assert "clusterDim = {2, 1, 1}" in mod.get_host_source()
|
|
|
|
|
|
def run_tvm_ffi_cluster_launch():
|
|
kernel = matmul(1024, 1024, 1024, 128, 128, 32)
|
|
mod = tilelang.compile(kernel, execution_backend="tvm_ffi")
|
|
check_str = r"""
|
|
(((TVMFFIAny*)stack_ffi_any)[3].type_index) = 1;
|
|
(((TVMFFIAny*)stack_ffi_any)[3].zero_padding) = 0;
|
|
(((TVMFFIAny*)stack_ffi_any)[3].v_int64) = ((int64_t)2);
|
|
(((TVMFFIAny*)stack_ffi_any)[4].type_index) = 1;
|
|
(((TVMFFIAny*)stack_ffi_any)[4].zero_padding) = 0;
|
|
(((TVMFFIAny*)stack_ffi_any)[4].v_int64) = ((int64_t)1);
|
|
(((TVMFFIAny*)stack_ffi_any)[5].type_index) = 1;
|
|
(((TVMFFIAny*)stack_ffi_any)[5].zero_padding) = 0;
|
|
(((TVMFFIAny*)stack_ffi_any)[5].v_int64) = ((int64_t)1);
|
|
"""
|
|
assert check_str in mod.get_host_source()
|
|
|
|
|
|
@tilelang.testing.requires_cuda
|
|
@tilelang.testing.requires_cuda_compute_version_ge(9, 0)
|
|
def test_cluster_launch():
|
|
run_cython_cluster_launch()
|
|
run_tvm_ffi_cluster_launch()
|
|
|
|
|
|
@tilelang.testing.requires_cuda
|
|
@tilelang.testing.requires_cuda_compute_version_ge(9, 0)
|
|
def test_cluster_launch_intrinsics(cluster_size=4):
|
|
kernel = get_cta_rank_in_cluster(cluster_size)
|
|
result = kernel()
|
|
ref = torch.arange(128, dtype=torch.int32, device="cuda") % cluster_size
|
|
assert torch.all(result == ref)
|
|
|
|
|
|
@tilelang.testing.requires_cuda
|
|
@tilelang.testing.requires_cuda_compute_version_ge(9, 0)
|
|
def test_cluster_barrier():
|
|
kernel = barrier_kernel()
|
|
kernel()
|
|
|
|
|
|
if __name__ == "__main__":
|
|
tilelang.testing.main()
|