Add fuse_single_source_parallel_gemm (#977)

Summary:
Pull Request resolved: #977

Add a fusion for single source parallel gemm.
This is an implementation of fuse_parallel_linear in DPER pass.

Reviewed By: khabinov, chenyang78

Differential Revision: D52087558

fbshipit-source-id: e1a0f0c0b2597adf96c02cad2f229cea78c10268

python/aitemplate/compiler/transform/fuse_parallel_gemms.py

@@ -23,6 +23,10 @@
 from aitemplate.compiler.ops.gemm_universal.gemm_common import default_align_ab
 from aitemplate.compiler.tensor_accessor import TensorAccessor
 from aitemplate.compiler.transform import transform_utils
+from aitemplate.compiler.transform.fuse_mm_elementwise_patterns import (
+    get_gemm_rcr_bias_patterns,
+)
+from aitemplate.compiler.transform.fuse_utils import transform_simple_fusion_patterns
 from aitemplate.compiler.transform.toposort import toposort
 from aitemplate.compiler.transform.transform_strided_ops import _is_supported_op
 
@@ -465,3 +469,110 @@ def fuse_parallel_gemms(
     for func in funcs:
         sorted_graph = func(sorted_graph)
     return sorted_graph
+
+
+def _fuse_single_source_parallel_gemms(
+    sorted_graph: List[Tensor],
+) -> Tuple[bool, List[Tensor]]:
+    _fusing_ops = {"gemm_rcr", "gemm_rcr_bias"}
+
+    for tensor in sorted_graph:
+        fusion_groups = {}
+        for dst in tensor.dst_ops():
+            op_type = dst._attrs["op"]
+            if op_type in _fusing_ops:
+                if dst._attrs["outputs"][0]._attrs["is_output"]:
+                    # Skip for outputs.
+                    continue
+
+                if (
+                    tensor == dst._attrs["inputs"][1]
+                    or dst._attrs["inputs"][1].src_ops()
+                    or dst._attrs["inputs"][1]._attrs["is_input"]
+                ):
+                    # Skip if weight or non-const
+                    continue
+                elif len(dst._attrs["inputs"]) > 2 and (
+                    tensor == dst._attrs["inputs"][2]
+                    or dst._attrs["inputs"][2].src_ops()
+                    or dst._attrs["inputs"][2]._attrs["is_input"]
+                ):
+                    # Skip if bias or non-const
+                    continue
+
+                if op_type in fusion_groups:
+                    fusion_groups[op_type].append(dst)
+                else:
+                    fusion_groups[op_type] = [dst]
+
+        for op_type, fusion_group in fusion_groups.items():
+            if len(fusion_group) < 2:
+                continue
+
+            bias = "bias" in op_type
+            W = []  # This stores all weights
+            B = []  # This stores all biases
+            N = []  # This stores all n from (m x k) x (n x k) of gemm_rcr
+            for gemm_op in fusion_group:
+                w = gemm_op._attrs["inputs"][1]
+                W.append(w)
+                if bias:
+                    B.append(gemm_op._attrs["inputs"][2])
+                N.append(w.shape()[0].value())
+            W_concat = ops.concatenate()(W, dim=0)
+            if bias:
+                B_concat = ops.concatenate()(B)
+                fused_gemm = ops.gemm_rcr_bias()(tensor, W_concat, B_concat)
+            else:
+                fused_gemm = ops.gemm_rcr()(tensor, W_concat)
+
+            split_result = ops.split()(fused_gemm, N, dim=-1)
+            for old_op, new_tensor in zip(fusion_group, split_result):
+                transform_utils.replace_tensor(old_op._attrs["outputs"][0], new_tensor)
+
+            sorted_graph = toposort(sorted_graph)
+            return True, transform_utils.sanitize_sorted_graph(sorted_graph)
+
+    return False, sorted_graph
+
+
+def fuse_single_source_parallel_gemms(
+    sorted_graph: List[Tensor], workdir: str = None
+) -> List[Tensor]:
+    """This pass fuses patterns like
+    # x: [m, k], w_i: [n_i, k], b_i: [n_i]
+    y1 = gemm_rcr_bias()(x, w1, b1)
+    y2 = gemm_rcr_bias()(x, w2, b2)
+    ...
+
+    into:
+    # x: [m, k], w: [sum(n_i), k], b: [sum(n_i)]
+    w = concatenate()([w1, w2], dim=0)
+    b = concatenate()([b1, b2], dim=0)
+    y = gemm_rcr_bias()(x, w, b)
+    y1, y2 = split()(y, n)
+
+    For w and b, we rely on constant folding to preprocess them.
+    y1 and y2 would be written directly from y's op.
+    It is required that all the gemm ops have the same layouts.
+
+    On graph pass ordering, we need to make sure this pass runs before
+    any other pass that modifies gemm and concat input/output TensorAccessors.
+
+    Args:
+        sorted_graph (List[Tensor]): a sorted list of tensors
+
+    Returns:
+        List[Tensor]: the transformed graph with all ops sorted
+    """
+
+    # Extract gemm_rcr_bias pattern first.
+    sorted_graph = transform_simple_fusion_patterns(
+        sorted_graph, get_gemm_rcr_bias_patterns()
+    )
+
+    applied = True
+    while applied:
+        applied, sorted_graph = _fuse_single_source_parallel_gemms(sorted_graph)
+
+    return sorted_graph

python/aitemplate/compiler/transform/optimize_graph.py

@@ -36,7 +36,10 @@
     fuse_ops,
     process_singleton_elementwise,
 )
-from aitemplate.compiler.transform.fuse_parallel_gemms import fuse_parallel_gemms
+from aitemplate.compiler.transform.fuse_parallel_gemms import (
+    fuse_parallel_gemms,
+    fuse_single_source_parallel_gemms,
+)
 from aitemplate.compiler.transform.fuse_permute_bmm_and_gemm import (
     fuse_permute_bmm_and_gemm,
 )
@@ -104,6 +107,7 @@ def optimize_graph(
         fuse_expand_bmm,
         transform_odd_alignment,
         fuse_conv_elementwise,
+        fuse_single_source_parallel_gemms,
         fuse_mm_elementwise,
         fuse_mm_reshape_permute,
         # make sure we run move_view_op_before_concat before transform_memory_ops