[AMD] support preshuffle weight mfma (#806)

Co-authored-by: Jiaxing Ding <jiaxing.ding@bytedance.com>

testing/python/amd/test_tilelang_gemm_mfma_preshuffle.py

+import torch
+import tilelang.testing
+from tilelang import tvm as tvm
+import tilelang.language as T
+from tilelang.intrinsics import make_mfma_swizzle_layout as make_swizzle_layout
+from tilelang.intrinsics.mfma_macro_generator import (
+    MatrixCoreIntrinEmitter,)
+from tilelang.transform import simplify_prim_func
+
+tilelang.testing.set_random_seed(0)
+
+
+@simplify_prim_func
+def tl_matmul(
+    M,
+    N,
+    K,
+    in_dtype,
+    out_dtype,
+    accum_dtype,
+    a_transposed=False,
+    b_transposed=True,
+    k_pack=1,
+    b_preshuffle=False,
+):
+    assert in_dtype in [
+        "float16",
+        "int8",
+    ], "Currently only float16 and int8 are supported"
+    assert out_dtype in [
+        "float16",
+        "float32",
+        "int32",
+    ], "Currently only float16, float32 and int32 are supported"
+
+    micro_size_x = micro_size_y = micro_size_k = 16
+
+    if in_dtype in {"float8_e4m3fnuz", "int8"}:
+        micro_size_k = 32
+
+    block_row_warps = 2
+    block_col_warps = 2
+    warp_row_tiles = 32
+    warp_col_tiles = 32
+
+    # for preshuffle_b, warp_layout = {1, 4}
+    if b_preshuffle:
+        block_row_warps = 1
+        block_col_warps = 4
+        warp_row_tiles = 128
+        warp_col_tiles = 32
+
+    chunk = 32 * k_pack
+
+    pack_size_k = micro_size_k * k_pack
+
+    shared_scope = "shared"
+    cache_write_shared = False
+
+    block_M = block_row_warps * warp_row_tiles
+    block_N = block_col_warps * warp_col_tiles
+    block_K = chunk
+
+    A_shape = (K, M) if a_transposed else (M, K)
+    if b_preshuffle:
+        B_shape = (N // micro_size_y, K // pack_size_k, micro_size_y,
+                   pack_size_k) if b_transposed else (K // pack_size_k, N // micro_size_y,
+                                                      pack_size_k, micro_size_y)
+    else:
+        B_shape = (N, K) if b_transposed else (K, N)
+    A_shared_shape = (block_K, block_M) if a_transposed else (block_M, block_K)
+    if b_preshuffle:
+        B_shared_shape = (block_N // micro_size_y, block_K // pack_size_k, micro_size_y,
+                          pack_size_k) if b_transposed else (block_K // pack_size_k,
+                                                             block_N // micro_size_y, pack_size_k,
+                                                             micro_size_y)
+    else:
+        B_shared_shape = (block_N, block_K) if b_transposed else (block_K, block_N)
+    C_shared_shape = (
+        block_M // micro_size_x,
+        block_N // micro_size_y,
+        micro_size_x,
+        micro_size_y,
+    )
+
+    warp_size = 64
+    threads = warp_size * (block_row_warps * block_col_warps)
+    local_size_a = (k_pack * micro_size_x * micro_size_k) // warp_size
+    local_size_b = (k_pack * micro_size_y * micro_size_k) // warp_size
+    local_size_c = (micro_size_x * micro_size_y) // warp_size
+    warp_rows = warp_row_tiles // micro_size_x
+    warp_cols = warp_col_tiles // micro_size_y
+
+    # MMA Wrapper to Auto Generate Code for MMA
+    mfma_emitter = MatrixCoreIntrinEmitter(
+        a_dtype=in_dtype,
+        b_dtype=in_dtype,
+        accum_dtype=accum_dtype,
+        a_transposed=a_transposed,
+        b_transposed=b_transposed,
+        block_row_warps=block_row_warps,
+        block_col_warps=block_col_warps,
+        warp_row_tiles=warp_row_tiles,
+        warp_col_tiles=warp_col_tiles,
+        chunk=chunk,
+        k_pack=k_pack,
+        b_preshuffle=b_preshuffle,
+    )
+
+    @T.prim_func
+    def main(
+            A: T.Tensor(A_shape, in_dtype),
+            B: T.Tensor(B_shape, in_dtype),
+            C: T.Tensor((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, scope=shared_scope)
+            B_shared = T.alloc_shared(B_shared_shape, in_dtype, scope=shared_scope)
+            C_shared = T.alloc_shared(C_shared_shape, out_dtype, scope=shared_scope)
+            A_local = T.alloc_local((warp_rows * local_size_a), in_dtype)
+            B_local = T.alloc_local((warp_cols * local_size_b), in_dtype)
+            C_local = T.alloc_local((warp_rows * warp_cols * local_size_c), accum_dtype)
+
+            T.annotate_layout({
+                A_shared: make_swizzle_layout(A_shared),
+            })
+
+            # Improve L2 Cache
+            T.use_swizzle(panel_size=10)
+
+            T.clear(C_local)
+
+            for ko in T.Pipelined((K // block_K), num_stages=0):
+
+                # Load A into shared memory
+                if a_transposed:
+                    T.copy(A[ko * block_K, by * block_M], A_shared)
+                else:
+                    T.copy(A[by * block_M, ko * block_K], A_shared)
+
+                # Load B into shared memory
+                if b_preshuffle:
+                    if b_transposed:
+                        for j, k, jj, kk in T.Parallel(block_N // micro_size_y,
+                                                       block_K // pack_size_k, micro_size_y,
+                                                       pack_size_k):
+                            B_shared[j, k, jj, kk] = B[bx * block_N // micro_size_y + j,
+                                                       ko * block_K // pack_size_k + k, jj, kk]
+                    else:
+                        for k, j, kk, jj in T.Parallel(block_K // pack_size_k,
+                                                       block_N // micro_size_y, pack_size_k,
+                                                       micro_size_y):
+                            B_shared[k, j, kk, jj] = B[ko * block_K // pack_size_k + k,
+                                                       bx * block_N // micro_size_y + j, kk, jj]
+                else:
+                    if b_transposed:
+                        T.copy(B[bx * block_N, ko * block_K], B_shared)
+                    else:
+                        T.copy(B[ko * block_K, bx * block_N], B_shared)
+
+                for ki in T.serial(0, (block_K // (k_pack * micro_size_k))):
+
+                    # Load A into fragment
+                    mfma_emitter.ldmatrix_a(
+                        A_local,
+                        A_shared,
+                        ki,
+                    )
+
+                    # Load B into fragment
+                    mfma_emitter.ldmatrix_b(
+                        B_local,
+                        B_shared,
+                        ki,
+                    )
+
+                    # Perform Matrix Multiplication
+                    mfma_emitter.mfma(A_local, B_local, C_local)
+
+            # Perform STMatrix
+            if cache_write_shared:
+                mfma_emitter.stmatrix(
+                    C_local,
+                    C_shared,
+                )
+
+                # Store shared into global
+                for i, j in T.Parallel(block_M, block_N):
+                    C[by * block_M + i, bx * block_N + j] = C_shared[
+                        i // micro_size_x,
+                        j // micro_size_y,
+                        i % micro_size_x,
+                        j % micro_size_y,
+                    ]
+            else:
+                mfma_emitter.stmatrix(
+                    C_local,
+                    C,
+                    pid_m=by,
+                    pid_n=bx,
+                )
+
+    return main
+
+
+def shuffle_weight(
+        x: torch.Tensor,
+        layout=(16, 32),
+        k_pack=1,
+        is_transpose=False,
+) -> torch.Tensor:
+    IN, IK = layout
+    BK = IK * k_pack
+    BN = IN
+
+    N, K = (x.shape[-2], x.shape[-1]) if is_transpose else (x.shape[-1], x.shape[-2])
+    assert N % BN == 0
+    assert K % BK == 0
+
+    x = x.view(N // BN, BN, K // BK, BK) if is_transpose else x.view(K // BK, BK, N // BN, BN)
+    x = x.permute(0, 2, 1, 3)
+    return x.contiguous()
+
+
+def assert_tl_matmul_correctness(M,
+                                 N,
+                                 K,
+                                 in_dtype,
+                                 out_dtype,
+                                 accum_dtype="float32",
+                                 a_transposed=False,
+                                 b_transposed=True,
+                                 k_pack=1,
+                                 b_preshuffle=False):
+    matmul = tl_matmul(M, N, K, in_dtype, out_dtype, accum_dtype, a_transposed, b_transposed,
+                       k_pack, b_preshuffle)
+    print(matmul)
+    kernel = tilelang.compile(matmul)
+    src_code = kernel.get_kernel_source()
+    # src_code is the generated cuda source
+    assert src_code is not None
+    A_shape = (K, M) if a_transposed else (M, K)
+    B_shape = (N, K) if b_transposed else (K, N)
+    if in_dtype == "int8":
+        A = torch.randint(-128, 127, A_shape, device="cuda", dtype=torch.int8)
+        B = torch.randint(-128, 127, B_shape, device="cuda", dtype=torch.int8)
+    else:
+        A = torch.rand(A_shape, device="cuda", dtype=getattr(torch, in_dtype))
+        B = torch.rand(B_shape, device="cuda", dtype=getattr(torch, in_dtype))
+
+    C = torch.zeros(M, N, device="cuda", dtype=getattr(torch, out_dtype))
+
+    B_preshuffle = B
+    if b_preshuffle:
+        B_preshuffle = shuffle_weight(B_preshuffle, k_pack=k_pack, is_transpose=b_transposed)
+        kernel(A, B_preshuffle, C)
+    else:
+        kernel(A, B, C)
+
+    print(kernel.get_kernel_source())
+
+    profiler = kernel.get_profiler()
+
+    latency = profiler.do_bench()
+
+    # Ensure that the latency is not None
+    assert latency is not None
+
+    if a_transposed and b_transposed:
+        # Get Reference Result
+        ref_c = torch.matmul(A.T.to(torch.float32),
+                             B.T.to(torch.float32)).to(getattr(torch, out_dtype))
+    elif a_transposed and not b_transposed:
+        # Get Reference Result
+        ref_c = torch.matmul(A.Tto(torch.float32),
+                             B.to(torch.float32)).to(getattr(torch, out_dtype))
+    elif not a_transposed and b_transposed:
+        # Get Reference Result
+        ref_c = torch.matmul(A.to(torch.float32),
+                             B.T.to(torch.float32)).to(getattr(torch, out_dtype))
+    else:
+        # Get Reference Result
+        ref_c = torch.matmul(A.to(torch.float32), B.to(torch.float32)).to(getattr(torch, out_dtype))
+
+    print(C)
+    print(ref_c)
+    torch.testing.assert_close(C, ref_c, rtol=1e-2, atol=1e-2)
+
+
+@tilelang.testing.requires_rocm
+def test_assert_tl_matmul():
+    assert_tl_matmul_correctness(128, 128, 128, "int8", "int32", accum_dtype="int32")
+    assert_tl_matmul_correctness(128, 256, 256, "int8", "int32", accum_dtype="int32")
+    assert_tl_matmul_correctness(
+        128, 256, 256, "int8", "int32", b_transposed=False, accum_dtype="int32")
+    assert_tl_matmul_correctness(128, 256, 256, "int8", "int32", accum_dtype="int32", k_pack=2)
+
+    assert_tl_matmul_correctness(
+        128, 128, 128, "int8", "int32", accum_dtype="int32", b_preshuffle=True)
+    assert_tl_matmul_correctness(
+        128, 256, 256, "int8", "int32", accum_dtype="int32", b_preshuffle=True)
+    assert_tl_matmul_correctness(
+        128, 256, 256, "int8", "int32", b_transposed=False, accum_dtype="int32", b_preshuffle=True)
+
+    assert_tl_matmul_correctness(
+        128, 256, 256, "int8", "int32", accum_dtype="int32", k_pack=2, b_preshuffle=True)
+    assert_tl_matmul_correctness(
+        128,
+        256,
+        256,
+        "int8",
+        "int32",
+        b_transposed=False,
+        accum_dtype="int32",
+        k_pack=2,
+        b_preshuffle=True)
+
+
+if __name__ == "__main__":
+    tilelang.testing.main()

tilelang/intrinsics/mfma_macro_generator.py

@@ -53,6 +53,7 @@ class MatrixCoreIntrinEmitter(object):
         num_elems_per_byte: int = 1,
         k_pack: Optional[int] = None,
         is_m_first: Optional[bool] = False,
+        b_preshuffle: Optional[bool] = False,
     ):
         self.a_dtype = a_dtype
         self.b_dtype = b_dtype
@@ -72,6 +73,7 @@ class MatrixCoreIntrinEmitter(object):
         self._initialize_micro_size(self.M_DIM, self.N_DIM, self.k_dim)
         self._initialize_k_pack(k_pack)
         self._initialize_is_m_first(is_m_first)
+        self._initialize_b_preshuffle(b_preshuffle)
 
         self.warp_rows = warp_row_tiles // self.micro_size_x
         self.warp_cols = warp_col_tiles // self.micro_size_y
@@ -141,6 +143,10 @@ class MatrixCoreIntrinEmitter(object):
         if is_m_first is not None:
             self.is_m_first = is_m_first
 
+    def _initialize_b_preshuffle(self, b_preshuffle: Optional[bool] = False):
+        if b_preshuffle is not None:
+            self.b_preshuffle = b_preshuffle
+
     def get_ldmatrix_index_map(self, is_b=False):
         from .mfma_layout import (
             shared_16x4_to_local_64x1_layout_A,
@@ -288,6 +294,31 @@ class MatrixCoreIntrinEmitter(object):
         ):
             tx, warp_n, _ = self.extract_thread_binding(thread_binding)
 
+            # 4 dim
+            if self.b_preshuffle:
+                if is_transposed:
+                    for j in T.serial(warp_cols):
+                        for local_id in T.vectorized(k_pack * local_size_b):
+                            row, col = T.meta_var(reverse_index_map(tx, local_id))
+                            l, r = (
+                                warp_n * warp_cols + j,
+                                rk * (chunk // micro_size_k) + ki,
+                            )
+                            B_local_buf[j * k_pack * local_size_b + local_id] = B_shared_buf[l, r,
+                                                                                             row,
+                                                                                             col]
+                else:
+                    for j in T.serial(warp_cols):
+                        for local_id in T.vectorized(k_pack * local_size_b):
+                            row, col = T.meta_var(reverse_index_map(tx, local_id))
+                            l, r = (
+                                rk * (chunk // micro_size_k) + ki,
+                                warp_n * warp_cols + j,
+                            )
+                            B_local_buf[j * k_pack * local_size_b + local_id] = B_shared_buf[l, r,
+                                                                                             row,
+                                                                                             col]
+            else:
                 if is_transposed:
                     for j in T.serial(warp_cols):
                         for local_id in T.vectorized(k_pack * local_size_b):
@@ -296,8 +327,8 @@ class MatrixCoreIntrinEmitter(object):
                                 warp_n * warp_col_tiles + j * micro_size_y,
                                 rk * chunk + ki * (k_pack * micro_size_k),
                             )
-                        B_local_buf[j * k_pack * local_size_b + local_id] = B_shared_buf[l + row,
-                                                                                         r + col]
+                            B_local_buf[j * k_pack * local_size_b +
+                                        local_id] = B_shared_buf[l + row, r + col]
                 else:
                     for j in T.serial(warp_cols):
                         for local_id in T.vectorized(k_pack * local_size_b):
@@ -306,8 +337,8 @@ class MatrixCoreIntrinEmitter(object):
                                 rk * chunk + ki * (k_pack * micro_size_k),
                                 warp_n * warp_col_tiles + j * micro_size_y,
                             )
-                        B_local_buf[j * k_pack * local_size_b + local_id] = B_shared_buf[l + row,
-                                                                                         r + col]
+                            B_local_buf[j * k_pack * local_size_b +
+                                        local_id] = B_shared_buf[l + row, r + col]
 
         return _warp_ldmatrix_b(B_local_buf, B_shared_buf, ki, thread_binding, rk)