[Feat] Introduce new caching mechanism for compiled kernels (#176)

* Add kernel caching mechanism to TileLang

- Implement a new `cached` function in `tilelang/cache/__init__.py` to cache and reuse compiled kernels
- Expose the `cached` function in the main `tilelang/__init__.py`
- Add a test case for cached matrix multiplication in `testing/python/cache/test_tilelang_cache_matmul.py`
- Provide a `clear_cache()` function to reset the kernel cache when needed

* Refactor kernel caching test and implementation

- Simplify the `cached` function in `tilelang/cache/__init__.py`
- Update test script `test_tilelang_cache_matmul.py` to use `tilelang.testing.main()`
- Remove unnecessary whitespace and improve code formatting

* Update import for `cached` function in MHA examples

- Modify import statement in `example_mha_bwd.py` and `test_tilelang_kernel_mha_bwd.py`
- Change import from `tilelang.profiler import cached` to `tilelang import cached`
- Align with recent refactoring of kernel caching mechanism

* Refactor `cached` function signature in kernel caching

- Update function signature to use keyword-only arguments for `target` and `target_host`
- Improve parameter order and readability of the `cached` decorator
- Maintain existing functionality while enhancing function definition

examples/flash_attention/example_mha_bwd.py

@@ -4,7 +4,7 @@
 import torch
 import torch.nn.functional as F
 import tilelang
-from tilelang.profiler import cached
+from tilelang import cached
 from tilelang.autotuner import *
 import tilelang.language as T
 import argparse

testing/python/cache/test_tilelang_cache_matmul.py

+# Copyright (c) Microsoft Corporation.
+# Licensed under the MIT License.
+
+from tilelang import tvm as tvm
+import tilelang.testing
+from tilelang import cached
+
+
+def matmul(
+    M,
+    N,
+    K,
+    block_M,
+    block_N,
+    block_K,
+    trans_A,
+    trans_B,
+    in_dtype,
+    out_dtype,
+    accum_dtype,
+    num_stages,
+    threads,
+):
+    A_shape = (K, M) if trans_A else (M, K)
+    B_shape = (N, K) if trans_B else (K, N)
+    A_shared_shape = (block_K, block_M) if trans_A else (block_M, block_K)
+    B_shared_shape = (block_N, block_K) if trans_B else (block_K, block_N)
+
+    import tilelang.language as T
+
+    @T.prim_func
+    def main(
+            A: T.Buffer(A_shape, in_dtype),
+            B: T.Buffer(B_shape, in_dtype),
+            C: T.Buffer((M, N), out_dtype),
+    ):
+        with T.Kernel(T.ceildiv(N, block_N), T.ceildiv(M, block_M), threads=threads) as (bx, by):
+            A_shared = T.alloc_shared(A_shared_shape, in_dtype)
+            B_shared = T.alloc_shared(B_shared_shape, in_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=num_stages):
+                if trans_A:
+                    T.copy(A[k * block_K, by * block_M], A_shared)
+                else:
+                    T.copy(A[by * block_M, k * block_K], A_shared)
+                if trans_B:
+                    T.copy(B[bx * block_N, k * block_K], B_shared)
+                else:
+                    T.copy(B[k * block_K, bx * block_N], B_shared)
+                T.gemm(A_shared, B_shared, C_local, trans_A, trans_B)
+            T.copy(C_local, C[by * block_M, bx * block_N])
+
+    return main
+
+
+def run_cache_matmul(
+    M,
+    N,
+    K,
+    trans_A,
+    trans_B,
+    in_dtype,
+    out_dtype,
+    dtypeAccum,
+    block_M,
+    block_N,
+    block_K,
+    num_stages=3,
+    num_threads=128,
+):
+    program = matmul(
+        M,
+        N,
+        K,
+        block_M,
+        block_N,
+        block_K,
+        trans_A,
+        trans_B,
+        in_dtype,
+        out_dtype,
+        dtypeAccum,
+        num_stages,
+        num_threads,
+    )
+
+    kernel = cached(program, [2])
+
+    profiler = kernel.get_profiler()
+
+    def ref_program(A, B):
+        import torch
+
+        if trans_A:
+            A = A.T
+        if trans_B:
+            B = B.T
+        C = torch.matmul(A.to(torch.float), B.to(torch.float))
+        C = C.to(torch.__getattribute__(out_dtype))
+        return C
+
+    profiler.assert_allclose(ref_program, atol=1e-2, rtol=1e-2)
+
+
+def test_cache_matmul_f16f16f16_nn():
+    run_cache_matmul(
+        512,
+        1024,
+        768,
+        False,
+        False,
+        "float16",
+        "float16",
+        "float16",
+        128,
+        256,
+        32,
+        2,
+    )
+
+
+if __name__ == "__main__":
+    tilelang.testing.main()

testing/python/kernel/test_tilelang_kernel_mha_bwd.py

@@ -4,7 +4,7 @@
 import torch
 import torch.nn.functional as F
 import tilelang
-from tilelang.profiler import cached
+from tilelang import cached
 import tilelang.language as T
 
 import tilelang.testing

tilelang/__init__.py

@@ -107,6 +107,7 @@ if SKIP_LOADING_TILELANG_SO == "0":
 
 from .jit import jit, JITKernel, compile  # noqa: F401
 from .profiler import Profiler  # noqa: F401
+from .cache import cached  # noqa: F401
 
 from .utils import (
     TensorSupplyType,  # noqa: F401

tilelang/cache/__init__.py

+"""The cache utils"""
+
+from tilelang import compile
+from tilelang.jit import JITKernel
+from typing import Callable, List, Union
+from tvm.target import Target
+from tvm.tir import PrimFunc
+
+# Dictionary to store cached kernels
+_cached = {}
+
+
+def cached(
+    func: Callable,
+    out_idx: List[int] = None,
+    *args,
+    target: Union[str, Target] = "auto",
+    target_host: Union[str, Target] = None,
+) -> JITKernel:
+    """
+    Cache and reuse compiled kernels to avoid redundant compilation.
+    
+    Args:
+        func: Function to be compiled or a PrimFunc that's already prepared
+        out_idx: Indices specifying which outputs to return
+        target: Compilation target platform
+        target_host: Host target for compilation
+        *args: Arguments passed to func when calling it
+        
+    Returns:
+        JITKernel: The compiled kernel, either freshly compiled or from cache
+    """
+    global _cached
+    # Create a unique key based on the function, output indices and arguments
+    key = (func, tuple(out_idx), *args)
+
+    # Return cached kernel if available
+    if key not in _cached:
+        # Handle both PrimFunc objects and callable functions
+        program = func if isinstance(func, PrimFunc) else func(*args)
+
+        # Compile the program to a kernel
+        kernel = compile(program, out_idx=out_idx, target=target, target_host=target_host)
+        # Store in cache for future use
+        _cached[key] = kernel
+
+    return _cached[key]
+
+
+def clear_cache():
+    """
+    Clear the entire kernel cache.
+    
+    This function resets the internal cache dictionary that stores compiled kernels.
+    Use this when you want to free memory or ensure fresh compilation
+    of kernels in a new context.
+    """
+    global _cached
+    _cached = {}

tilelang/profiler/__init__.py

-# Copyright (c) Microsoft Corporation.
-# Licensed under the MIT License.
 """The profiler and convert to torch utils"""
 
 from typing import List, Literal, Optional, Callable
@@ -9,8 +7,6 @@ from contextlib import suppress
 
 import tvm
 from tvm.relay import TensorType
-
-from tilelang.engine import lower
 from tilelang.jit.adapter import TorchDLPackKernelAdapter
 from tilelang.utils.tensor import (
     get_tensor_supply,
@@ -240,17 +236,3 @@ def do_bench(
             ret = ret[0]
         return ret
     return getattr(torch, return_mode)(times).item()
-
-
-_cached = {}
-
-
-def cached(func, result_idx: List[int], *args):
-    global _cached
-    key = (func, tuple(result_idx), *args)
-    if key not in _cached:
-        program = func(*args)
-        mod, params = lower(program)
-        mod = TorchDLPackKernelAdapter(mod, params, result_idx)
-        _cached[key] = mod
-    return _cached[key]