Merge branch 'fused_kernel_cond' into 'main'

conditioning fused kernels

See merge request ADLR/megatron-lm!228

megatron/arguments.py

@@ -202,7 +202,23 @@ def parse_args(extra_args_provider=None, defaults={},
         assert args.checkpoint_activations, \
             'for distribute-checkpointed-activations to work you '\
             'need to enable checkpoint-activations'
-   
+
+    # custom kernel constraints check
+    seq_len = args.seq_length
+    attn_batch_size = \
+        (args.num_attention_heads / args.tensor_model_parallel_size) * \
+        args.micro_batch_size
+
+    # constraints on sequence length and attn_batch_size to enable warp based
+    # optimization and upper triangular optimization (for causal mask)
+    custom_kernel_constraint = seq_len > 16 and seq_len <=2048 and \
+        seq_len % 4 == 0 and attn_batch_size % 4 == 0
+
+    if args.fp16 and custom_kernel_constraint and args.masked_softmax_fusion:
+        print('WARNING: constraints for invoking optimized'
+            ' fused softmax kernel are not met. We default back to unfused'
+            ' kernel invocations.')
+
     # Load scaled_masked_softmax_fusion_kernels
     if args.masked_softmax_fusion:
         fused_kernels.load_scaled_upper_triang_masked_softmax_fusion_kernel()

megatron/model/fused_softmax.py

@@ -113,18 +113,23 @@ class FusedScaleMaskSoftmax(torch.nn.Module):
         assert (
             self.scale is None or softmax_in_fp32
         ), "softmax should be in fp32 when scaled"
-
+ 
     def forward(self, input, mask):
         # [b, np, sq, sk]
+        assert input.dim() == 4
         data_size = input.size()
         query_seq_len = data_size[-2]
         key_seq_len = data_size[-1]
-        assert input.dim() == 4
+        attn_batch_size = data_size[0] * data_size[1]
 
-        # invoke custom kernel
-        if self.input_in_fp16 and key_seq_len <= 2048 and mask is not None and \
-           query_seq_len % 4 == 0 and self.scaled_masked_softmax_fusion:
+        # constraints on various tensor dimensions to enable warp based
+        # optimization and upper triangular optimization (for causal mask)
+        custom_kernel_constraint = key_seq_len > 16 and key_seq_len <= 2048 and \
+            query_seq_len % 4 == 0 and attn_batch_size % 4 == 0
 
+        # invoke custom kernel
+        if self.input_in_fp16 and mask is not None and \
+           custom_kernel_constraint and self.scaled_masked_softmax_fusion:
             scale = self.scale if self.scale is not None else 1.0
 
             if self.attn_mask_type == AttnMaskType.causal: