Add scaled_grouped_mm_v2 and python API (pytorch#165154)

Summary:

* Add `torch._scaled_grouped_mm_v2` with more functionality and
  extensibility for future formats
* Add `torch.nn.functional.scaled_grouped_mm` as public entrypoint
* Test both original and v2 functionality

Test Plan:

```
pytest -svv -k grouped test/test_scaled_matmul_cuda.py
```

Reviewers:

Subscribers:

Tasks:

Tags:
Signed-off-by: Simon Layton <[email protected]>

Pull Request resolved: pytorch#165154
Approved by: https://github.com/drisspg, https://github.com/danielvegamyhre

Author: slayton58

aten/src/ATen/native/cuda/Blas.cpp

@@ -2578,7 +2578,9 @@ _mx8_mx8_bf16_grouped_mm_fbgemm(
         const Tensor& mat_a,
         const Tensor& mat_b,
         const Tensor& scale_a,
+        const SwizzleType& swizzle_a,
         const Tensor& scale_b,
+        const SwizzleType& swizzle_b,
         const std::optional<at::Tensor>& offs,
         Tensor& out) {
     const bool a_is_2d = mat_a.dim() == 2;
@@ -2589,6 +2591,16 @@ _mx8_mx8_bf16_grouped_mm_fbgemm(
     TORCH_CHECK_VALUE(is_2d_2d || is_2d_3d, "MXFP8 grouped GEMM currently only supports 2d-2d and 2d-3d cases");
     TORCH_CHECK_VALUE(offs.has_value(), "MXFP8 2d-2d and 2d-3d grouped GEMMs requires offsets");
     TORCH_CHECK_VALUE(out.scalar_type() == at::kBFloat16, "Only bf16 out_dtype is supported for MXFP8 grouped gemm");
+    // MXFP8 expects float8_e8m0fnu scales.
+    TORCH_CHECK_VALUE(scale_a.scalar_type() == at::kFloat8_e8m0fnu && scale_b.scalar_type() == at::kFloat8_e8m0fnu,
+        "For MXFP8 grouped gemm, both scales must be float8_e8m0fnu tensors.");
+#ifdef USE_ROCM
+    TORCH_CHECK_VALUE(swizzle_a == SwizzleType::NO_SWIZZLE && swizzle_b == SwizzleType::NO_SWIZZLE,
+        "For ROCM MXFP8 grouped gemm, both scale swizzle types must be SWIZZLE_NONE");
+#else
+    TORCH_CHECK_VALUE(swizzle_a == SwizzleType::SWIZZLE_32_4_4 && swizzle_b == SwizzleType::SWIZZLE_32_4_4,
+        "For CUDA MXFP8 grouped gemm, both scale swizzle types must be SWIZZLE_32_4_4");
+#endif
 
 #if defined(USE_FBGEMM_GENAI) and !defined(USE_ROCM)
     fbgemm_gpu::mx8mx8bf16_grouped_mm(
@@ -2673,6 +2685,9 @@ _f8_f8_bf16_rowwise_grouped_mm(
       const std::optional<Tensor>& bias,
       bool use_fast_accum,
       Tensor& out) {
+  // FP8 per-tensor and per-row scaling expect fp32 scales.
+  TORCH_CHECK_VALUE(scale_a.scalar_type() == kFloat && scale_b.scalar_type() == kFloat,
+      "For grouped FP8 rowwise, both scales must be float32 tensors");
 #ifndef USE_ROCM
   return _f8_f8_bf16_rowwise_grouped_mm_cuda(
       mat_a,
@@ -2772,11 +2787,15 @@ _scaled_grouped_mm_cuda(
 #endif
 
   if (is_mx8mx8bf16) {
+    // Note: Passing implied SwizzleType here, correctness of scale previously checked
+    //       in `check_scale` call
     return _mx8_mx8_bf16_grouped_mm_fbgemm(
         mat_a,
         mat_b,
         scale_a,
+        SwizzleType::SWIZZLE_32_4_4,
         scale_b,
+        SwizzleType::SWIZZLE_32_4_4,
         offs.value(),
         out);
   }
@@ -2793,6 +2812,140 @@ _scaled_grouped_mm_cuda(
       out);
 }
 
+namespace {
+
+std::array<std::tuple<std::string, acceptance_fn, ScaledGemmImplementation>, 2> scale_grouped_kernel_dispatch = {{
+  { "rowwise_rowwise", check_rowwise_recipe, ScaledGemmImplementation::ROWWISE_ROWWISE},
+  { "mxfp8_mxfp8", check_mxfp8_recipe, ScaledGemmImplementation::MXFP8_MXFP8}}};
+
+} // anonymous namespace
+
+Tensor
+_scaled_grouped_mm_cuda_v2(
+          const Tensor& mat_a, const Tensor& mat_b,
+          ArrayRef<Tensor> scale_a,
+          IntArrayRef scale_recipe_a,
+          IntArrayRef swizzle_a,
+          ArrayRef<Tensor> scale_b,
+          IntArrayRef scale_recipe_b,
+          IntArrayRef swizzle_b,
+          const std::optional<Tensor>& offs,
+          const std::optional<Tensor>& bias,
+          const std::optional<c10::ScalarType> out_dtype,
+          IntArrayRef contraction_dim,
+          bool use_fast_accum) {
+  bool allowed_device = _scaled_mm_allowed_device(/*sm90_only*/true, /*sm100_only*/true);
+  TORCH_CHECK_VALUE(allowed_device, "torch._scaled_grouped_mm is only supported on CUDA devices with compute capability = [9.0, 10.0], or ROCm MI300+");
+
+  TORCH_CHECK_VALUE(!check_valid_strides_and_return_transposed(mat_a), "Expected mat1 to not be transposed");
+  TORCH_CHECK_VALUE(check_valid_strides_and_return_transposed(mat_b), "Expected mat2 to be transposed");
+  TORCH_CHECK_VALUE(mat_a.dim() == 2 || mat_a.dim() == 3, "mat_a has to be 2 or 3d");
+  TORCH_CHECK_VALUE(mat_b.dim() == 2 || mat_b.dim() == 3, "mat_b has to be 2 or 3d");
+  const bool a_is_2d = mat_a.dim() == 2;
+  const bool b_is_2d = mat_b.dim() == 2;
+
+  // NOTE(slayton): For sub-1B formats want contraction_dim argument?
+  if (!a_is_2d || !b_is_2d) {
+    if (contraction_dim.size() > 0) {
+      const int dim_a = contraction_dim[0], dim_b = mat_b.size(contraction_dim[1]);
+      TORCH_CHECK_VALUE(mat_a.size(dim_a) == mat_b.size(dim_b),
+          "Contraction dimensions (", dim_a, ",", dim_b, ") of mat_a and mat_b must match, got: ", mat_a.size(dim_a), " and ",
+          mat_b.size(dim_b));
+      // Note: only (-1, -2) is currently supported
+      TORCH_CHECK_VALUE(dim_a == -1 && dim_b == -2, "Curently contraction dims must be (-1, -2) only");
+    } else {
+      TORCH_CHECK_VALUE(mat_a.size(-1) == mat_b.size(-2), "contraction dimension of mat_a and mat_b must match");
+    }
+  }
+  TORCH_CHECK_VALUE(
+    mat_a.size(-1) % 16 == 0,
+    "Expected trailing dimension of mat_a to be divisible by 16 ",
+    "but got mat1 shape: (",
+    mat_a.sizes(),
+    ").");
+  TORCH_CHECK_VALUE(mat_b.size(-2) % 16 == 0 && mat_b.size(-1) % 16 == 0,
+    "Expected mat_b shape to be divisible by 16 ",
+    "but got mat_b shape: (",
+    mat_b.sizes(),
+    ").");
+
+  TORCH_CHECK_VALUE(!bias.has_value(), "Bias not supported yet");
+  TORCH_CHECK_VALUE(offs.has_value() ==  (a_is_2d || b_is_2d), "Have to provide offsets if there is a 2d matrix");
+
+  // NOTE: mxfp8 x mxfp8 requires (and asserts later) that offsets is present.
+  //       for rowwise, no offsets implies 3d-3d and is handled by lower-level
+  //       routines
+  if (offs.has_value()) {
+    TORCH_CHECK_VALUE(offs->dim() == 1, "offs has to be 1D");
+    TORCH_CHECK_VALUE(offs->dtype() == at::kInt, "Offsets have to be int32");
+  }
+
+  const auto out_dtype_ = out_dtype.value_or(kBFloat16);
+  TORCH_CHECK_VALUE(out_dtype_ == kBFloat16, "Only bf16 high precision output types are supported for grouped gemm");
+
+  Tensor out = create_grouped_gemm_output_tensor(mat_a, mat_b, offs, out_dtype_);
+
+  // Conversion of implicitly-defined enums to explicit
+  auto scale_recipe_a_enum = convert_int_to_enum<ScalingType>(scale_recipe_a);
+  auto swizzle_a_enum = convert_int_to_enum<SwizzleType>(swizzle_a);
+  auto scale_recipe_b_enum = convert_int_to_enum<ScalingType>(scale_recipe_b);
+  auto swizzle_b_enum = convert_int_to_enum<SwizzleType>(swizzle_b);
+
+  // at this point we can start working out what we want to be doing
+  // Try to do as few steps as possible.
+  // NOTE: support is deliberately sparse, can explicitly enumerate all combinations allowed.
+  // Do this via a list of defined (name, acceptance, concrete_impl) tuples.
+  ScaledGemmImplementation gemm_impl = ScaledGemmImplementation::NONE;
+  for (const auto& fn_entry : scale_grouped_kernel_dispatch) {
+    const auto [name, accept_fn, scaled_gemm_impl] = fn_entry;
+    bool ok = accept_fn(mat_a.scalar_type(),
+                        scale_recipe_a_enum,
+                        scale_a,
+                        mat_b.scalar_type(),
+                        scale_recipe_b_enum,
+                        scale_b);
+    if (ok) {
+      gemm_impl = scaled_gemm_impl;
+      break;
+    }
+  }
+  TORCH_CHECK_VALUE(gemm_impl != ScaledGemmImplementation::NONE,
+      "No gemm implementation was found");
+
+  switch (gemm_impl) {
+    case ScaledGemmImplementation::ROWWISE_ROWWISE: {
+      const int scale_multiplier = (mat_a.dim() == 2 && mat_b.dim() == 2) ? offs->size(0) : 1;
+      _check_scales_fp8_rowwise(mat_a, scale_a[0], 0 /* dim */ , 0 /* arg_idx */, scale_multiplier);
+      _check_scales_fp8_rowwise(mat_b, scale_b[0], 1 /* dim */ , 1 /* arg_idx */, scale_multiplier);
+      return _f8_f8_bf16_rowwise_grouped_mm(
+          mat_a,
+          mat_b,
+          scale_a[0],
+          scale_b[0],
+          offs,
+          bias,
+          use_fast_accum,
+          out);
+    }
+    case ScaledGemmImplementation::MXFP8_MXFP8: {
+      _check_scales_mxfp8(mat_a, scale_a[0], 0 /* dim */, 0 /* arg_idx */);
+      _check_scales_mxfp8(mat_b, scale_b[0], 1 /* dim */, 1 /* arg_idx */);
+      return _mx8_mx8_bf16_grouped_mm_fbgemm(
+          mat_a,
+          mat_b,
+          scale_a[0],
+          swizzle_a_enum[0],
+          scale_b[0],
+          swizzle_b_enum[0],
+          offs.value(),
+          out);
+    }
+    default:
+      TORCH_CHECK_NOT_IMPLEMENTED(false,
+          "_scaled_grouped_mm_cuda_v2 is in an inconsistent state - should never reach here");
+  }
+}
+
 Tensor _grouped_mm_cuda(const Tensor& mat_a, const Tensor& mat_b,
 const std::optional<at::Tensor>& offs,
 const std::optional<at::Tensor>& bias,

aten/src/ATen/native/native_functions.yaml

@@ -7183,6 +7183,12 @@
     CUDA: _scaled_grouped_mm_cuda
   tags: needs_exact_strides
 
+- func: _scaled_grouped_mm_v2(Tensor self, Tensor mat2, Tensor[] scale_a, int[] recipe_a, int[] swizzle_a, Tensor[] scale_b, int[] recipe_b, int[] swizzle_b, Tensor? offs=None, Tensor? bias=None, ScalarType? out_dtype=None, int[] contraction_dim=[], bool use_fast_accum=False) -> Tensor
+  variants: function
+  dispatch:
+    CUDA: _scaled_grouped_mm_cuda_v2
+  tags: needs_exact_strides
+
 - func: _grouped_mm(Tensor self, Tensor mat2, Tensor? offs=None, Tensor? bias=None, ScalarType? out_dtype=None) -> Tensor
   variants: function
   dispatch:

docs/source/nn.functional.rst

@@ -228,3 +228,4 @@ Low-Precision functions
     ScalingType
     SwizzleType
     scaled_mm
+    scaled_grouped_mm

test/expect/HasDecompTest.test_has_decomposition.expect

@@ -524,6 +524,7 @@ aten::_scaled_dot_product_flash_attention_for_cpu_backward
 aten::_scaled_dot_product_fused_attention_overrideable
 aten::_scaled_dot_product_fused_attention_overrideable_backward
 aten::_scaled_grouped_mm
+aten::_scaled_grouped_mm_v2
 aten::_scaled_mm
 aten::_scaled_mm.out
 aten::_scaled_mm_v2