Add spmm bf16 support (#269)

* Add spmm bf16 support and add ut

* Add bf16 support for spspmm

* Disable bf16 test before torch_scatter 2.0.9

* Refactor version compare

* Update test/utils.py

Co-authored-by: Matthias Fey <[email protected]>

Author: yanbing-j

csrc/cpu/spmm_cpu.cpp

@@ -44,7 +44,7 @@ spmm_cpu(torch::Tensor rowptr, torch::Tensor col,
   auto K = mat.size(-1);
   auto B = mat.numel() / (N * K);
 
-  AT_DISPATCH_ALL_TYPES_AND(at::ScalarType::Half, mat.scalar_type(), "_", [&] {
+  AT_DISPATCH_ALL_TYPES_AND2(at::ScalarType::Half, at::ScalarType::BFloat16, mat.scalar_type(), "spmm_cpu", [&] {
     scalar_t *value_data = nullptr;
     auto mat_data = mat.data_ptr<scalar_t>();
     auto out_data = out.data_ptr<scalar_t>();
@@ -123,7 +123,7 @@ torch::Tensor spmm_value_bw_cpu(torch::Tensor row, torch::Tensor rowptr,
   auto row_data = row.data_ptr<int64_t>();
   auto rowptr_data = rowptr.data_ptr<int64_t>();
   auto col_data = col.data_ptr<int64_t>();
-  AT_DISPATCH_ALL_TYPES_AND(at::ScalarType::Half, mat.scalar_type(), "_", [&] {
+  AT_DISPATCH_ALL_TYPES_AND2(at::ScalarType::Half, at::ScalarType::BFloat16, mat.scalar_type(), "spmm_value_bw_cpu", [&] {
     auto mat_data = mat.data_ptr<scalar_t>();
     auto grad_data = grad.data_ptr<scalar_t>();
     auto out_data = out.data_ptr<scalar_t>();

csrc/cpu/spspmm_cpu.cpp

@@ -55,7 +55,7 @@ spspmm_cpu(torch::Tensor rowptrA, torch::Tensor colA,
   torch::Tensor colC;
   torch::optional<torch::Tensor> optional_valueC = torch::nullopt;
 
-  AT_DISPATCH_ALL_TYPES(scalar_type, "spspmm", [&] {
+  AT_DISPATCH_ALL_TYPES_AND(at::ScalarType::BFloat16, scalar_type, "spspmm", [&] {
     AT_DISPATCH_HAS_VALUE(optional_valueA, [&] {
       scalar_t *valA_data = nullptr, *valB_data = nullptr;
       if (HAS_VALUE) {
@@ -77,7 +77,7 @@ spspmm_cpu(torch::Tensor rowptrA, torch::Tensor colA,
             if (HAS_VALUE)
               tmp_vals[cB] += valA_data[eA] * valB_data[eB];
             else
-              tmp_vals[cB]++;
+              tmp_vals[cB] += 1;
           }
         }
 

test/test_spspmm.py

@@ -35,7 +35,7 @@ def test_sparse_tensor_spspmm(dtype, device):
         ], dtype=dtype, device=device),
     )
 
-    expected = torch.eye(10, dtype=dtype, device=device)
+    expected = torch.eye(10, device=device).to(dtype)
 
     out = x @ x.to_dense().t()
     assert torch.allclose(out, expected, atol=1e-2)

test/utils.py

@@ -1,10 +1,16 @@
 import torch
+import torch_scatter
+from packaging import version
 
 reductions = ['sum', 'add', 'mean', 'min', 'max']
 
 dtypes = [torch.half, torch.float, torch.double, torch.int, torch.long]
 grad_dtypes = [torch.half, torch.float, torch.double]
 
+if version.parse(torch_scatter.__version__) > version.parse("2.0.9"):
+    dtypes.append(torch.bfloat16)
+    grad_dtypes.append(torch.bfloat16)
+
 devices = [torch.device('cpu')]
 if torch.cuda.is_available():
     devices += [torch.device(f'cuda:{torch.cuda.current_device()}')]