mirror of
https://github.com/tile-ai/tilelang.git
synced 2026-10-24 07:04:00 +08:00
Lei Wang
f2e99180d5
* remove llvm build * [Refactor] Update kernel compilation and profiling in examples - Replaced `tilelang.lower` with `tilelang.compile` in multiple example scripts to streamline kernel compilation. - Updated profiling calls to utilize the new `get_profiler` method, enhancing performance measurement consistency. - Adjusted assertions and benchmarking methods to align with the new profiling structure across various examples, ensuring correctness and clarity in performance evaluations. * lint fix * License Update * [Refactor] Improve code formatting and documentation in CUDA header and HIP runtime files - Adjusted formatting in `cuda.h` for better readability, including alignment of comments and struct fields. - Cleaned up whitespace and improved comment clarity in `rt_mod_hip.cc` to enhance code maintainability. * [Refactor] Enhance formatting and clarity in CUDA header and HIP runtime files - Improved comment alignment and readability in `cuda.h`. - Cleaned up whitespace and formatting in `rt_mod_hip.cc` to enhance maintainability. * lint fix * lint fix * lint fix * lint fix * fix * License update * [Enhancement] Update JITKernel to use artifact for kernel source - Assigned the generated artifact to `self.artifact` for better management. - Updated kernel source references to use `artifact.kernel_source` for consistency in execution backend handling. * lint fix * Add @tilelang.testing.requires_llvm decorator to vectorization tests * Enhance setup.py and env.py for library management - Added functionality to remove original files after copying in CMakeBuild. - Updated TVM_LIBRARY_PATH in env.py to include the PyPI build library path for better integration. * Refactor TVM_LIBRARY_PATH assignment for improved readability in env.py * Refactor CMakeBuild file handling in setup.py - Added a check to ensure the target library directory exists before copying .so files. - Improved the logic for creating the target directory and copying files to enhance robustness. * bugfix * Rename BuildTLDebug to BuildTileLangCUDAWithoutCompile and update registration. Add @tilelang.testing.requires_llvm decorator to multiple tests for LLVM requirement. * lint fix * Enhance TileLang code generation by adding support for device code generation without compilation. Updated `host_codegen` and `device_codegen` functions to include new transformations and registration for `tilelang_hip_without_compile`. Refactored JIT kernel adapters to accommodate host and device modules, improving overall integration and flexibility. * lint fix * Add support for C target in device code generation - Updated `device_codegen_without_compile` to include handling for the C target by registering the `tilelang_cpp` function. * [Enhancement] Implement auto-clear cache feature based on environment variable * Added TILELANG_CLEAR_CACHE environment variable to control cache clearing. * Updated CI workflow to set TILELANG_CLEAR_CACHE during testing. * Modified cache initialization to clear cache if TILELANG_CLEAR_CACHE is set to true. * [Refactor] Update kernel invocation and import paths in tests and cache * Changed kernel invocation in `test_tilelang_kernel_dequantize_gemm.py` to return the result. * Updated import statements in `test_tilelang_kernel_int4_gemm_mma.py` to use `bitblas` instead of `tilelang`. * Refactored paths for artifact and parameters in `kernel_cache.py` for better maintainability. * [Refactor] Clean up whitespace and improve code formatting in kernel_cache.py * Removed unnecessary blank lines and adjusted spacing for better readability in the KernelCache class. * Enhanced overall code formatting to align with project standards. * [Enhancement] Add bfloat16 test case and improve kernel caching logic * Introduced a new test case for bfloat16 matrix multiplication in `test_tilelang_kernel_gemm_mma_intrinsic.py`. * Updated `KernelCache` to handle multiple kernel source files and improve error handling during saving and loading. * Refactored `JITKernel` to support instantiation from a database, enhancing flexibility in kernel management. * Adjusted `CtypesKernelAdapter` and `CythonKernelAdapter` to utilize the new kernel loading mechanism from the database. * Improved code formatting and readability across several files. * lint fix * Update bfloat16 matrix multiplication test case to use larger dimensions for improved coverage
118 lines
4.1 KiB
Python
118 lines
4.1 KiB
Python
import tilelang
|
|
import tilelang.testing
|
|
from tilelang import tvm as tvm
|
|
import tilelang.language as T
|
|
import torch
|
|
|
|
|
|
def matmul(M, N, K, block_M, block_N, block_K, dtype="float16", accum_dtype="float"):
|
|
num_stages = 0
|
|
|
|
@T.prim_func
|
|
def matmul(
|
|
A: T.Buffer((M, K), dtype),
|
|
B: T.Buffer((K, N), dtype),
|
|
C: T.Buffer((M, N), dtype),
|
|
):
|
|
with T.Kernel(T.ceildiv(N, block_N), T.ceildiv(M, block_M), is_cpu=True) as (bx, by):
|
|
A_local = T.alloc_local((block_M, block_K), dtype)
|
|
B_local = T.alloc_local((block_K, block_N), dtype)
|
|
C_local = T.alloc_local((block_M, block_N), accum_dtype)
|
|
|
|
T.clear(C_local)
|
|
|
|
# Apply layout optimizations or define your own layout
|
|
# (Optional).
|
|
# T.annotate_layout(
|
|
# {
|
|
# A_local: make_swizzle_layout(A_local),
|
|
# B_local: make_swizzle_layout(B_local),
|
|
# }
|
|
# )
|
|
|
|
for ko in T.Pipelined(K // block_K, num_stages=num_stages):
|
|
|
|
T.copy(A[by * block_M, ko * block_K], A_local)
|
|
|
|
# Or Copy with Parallel
|
|
for k, j in T.Parallel(block_K, block_N):
|
|
B_local[k, j] = B[ko * block_K + k, by * block_N + j]
|
|
|
|
for i, j, k in T.grid(block_M, block_N, block_K):
|
|
C_local[i, j] += A_local[i, k] * B_local[k, j]
|
|
|
|
T.copy(C_local, C[by * block_M, bx * block_N])
|
|
|
|
return matmul
|
|
|
|
|
|
def assert_matmul_codegen(M=1024, N=1024, K=1024, block_M=128, block_N=128, block_K=32):
|
|
func = matmul(M, N, K, block_M, block_N, block_K)
|
|
|
|
artifact = tilelang.lower(func, target="c")
|
|
|
|
code = artifact.kernel_source
|
|
|
|
assert code is not None, "Code generation failed"
|
|
|
|
|
|
def test_matmul_codegen():
|
|
assert_matmul_codegen(M=1024, N=1024, K=1024, block_M=128, block_N=128, block_K=32)
|
|
|
|
|
|
def test_matmul_compile():
|
|
|
|
def matmul_jit_test(M, N, K, block_M, block_N, block_K, dtype="float16", accum_dtype="float"):
|
|
# a simple kernel just for jit test
|
|
@T.prim_func
|
|
def matmul(
|
|
A: T.Buffer((M, K), dtype),
|
|
B: T.Buffer((K, N), dtype),
|
|
C: T.Buffer((M, N), dtype),
|
|
):
|
|
with T.Kernel(T.ceildiv(N, block_N), T.ceildiv(M, block_M), is_cpu=True) as (bx, by):
|
|
A_local = T.alloc_local((block_M, block_K), dtype)
|
|
B_local = T.alloc_local((block_K, block_N), dtype)
|
|
C_local = T.alloc_local((block_M, block_N), accum_dtype)
|
|
|
|
for p in T.serial(block_M):
|
|
for w in T.serial(block_N):
|
|
C_local[p, w] = 0
|
|
for ko in T.serial(K // block_K):
|
|
for i in T.serial(block_M):
|
|
for k in T.serial(block_K):
|
|
A_local[i, k] = A[by * block_M + i, ko * block_K + k]
|
|
|
|
for k in T.serial(block_K):
|
|
for j in T.serial(block_N):
|
|
B_local[k, j] = B[ko * block_K + k, bx * block_N + j]
|
|
|
|
for i in T.serial(block_M):
|
|
for j in T.serial(block_N):
|
|
for k in T.serial(block_K):
|
|
C_local[i, j] += A_local[i, k] * B_local[k, j]
|
|
|
|
for i in T.serial(block_M):
|
|
for j in T.serial(block_N):
|
|
C[by * block_M + i, bx * block_N + j] = C_local[i, j]
|
|
|
|
return matmul
|
|
|
|
M, N, K = 1024, 512, 512
|
|
block_M, block_N, block_K = M // 4, N // 4, K // 4
|
|
cpu_func = matmul_jit_test(M, N, K, block_M, block_N, block_K)
|
|
complied_fun = tilelang.compile(cpu_func, -1, execution_backend="ctypes", target="c")
|
|
|
|
in_dtype = "float16"
|
|
A = torch.randn(M, K, dtype=torch.__getattribute__(in_dtype))
|
|
B = torch.randn(K, N, dtype=torch.__getattribute__(in_dtype))
|
|
|
|
C = complied_fun(A, B)
|
|
C_torch = torch.matmul(A, B)
|
|
|
|
tilelang.testing.torch_assert_close(C, C_torch, atol=1e-2, rtol=1e-2, max_mismatched_ratio=0.05)
|
|
|
|
|
|
if __name__ == "__main__":
|
|
tilelang.testing.main()
|