adjust pass order and follow naming conventions

PaddlePaddle · Jan 3, 2024 · 8d67748 · 8d67748
1 parent d0df7b5
commit 8d67748
Show file tree

Hide file tree

Showing 5 changed files with 20 additions and 20 deletions.
diff --git a/paddle/fluid/framework/ir/CMakeLists.txt b/paddle/fluid/framework/ir/CMakeLists.txt
@@ -273,7 +273,7 @@ if(WITH_XPU)
                ${XPU_PASS_DEPS})
   pass_library(qk_qkv_attention_xpu_fuse_pass inference DIR xpu DEPS
                ${XPU_PASS_DEPS})
-  pass_library(vis_decoder_attention_xpu_fuse_pass inference DIR xpu DEPS
+  pass_library(decoder_attention_xpu_fuse_pass inference DIR xpu DEPS
                ${XPU_PASS_DEPS})
   pass_library(multi_encoder_xpu_fuse_pass inference DIR xpu DEPS
                ${XPU_PASS_DEPS})

diff --git a/...pu/vis_decoder_attention_xpu_fuse_pass.cc → ...ir/xpu/decoder_attention_xpu_fuse_pass.cc b/...pu/vis_decoder_attention_xpu_fuse_pass.cc → ...ir/xpu/decoder_attention_xpu_fuse_pass.cc
@@ -12,7 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "paddle/fluid/framework/ir/xpu/vis_decoder_attention_xpu_fuse_pass.h"
+#include "paddle/fluid/framework/ir/xpu/decoder_attention_xpu_fuse_pass.h"
 
 #include "glog/logging.h"
 
@@ -27,9 +27,9 @@ namespace ir {
 
 namespace patterns {
 
-struct VisDecoderAttentionFusePattern : public PatternBase {
-  VisDecoderAttentionFusePattern(PDPattern* pattern,
-                                 const std::string& name_scope);
+struct DecoderAttentionFusePattern : public PatternBase {
+  DecoderAttentionFusePattern(PDPattern* pattern,
+                              const std::string& name_scope);
 
   // declare operator node's name
   PATTERN_DECL_NODE(reshape2_1);
@@ -63,7 +63,7 @@ struct VisDecoderAttentionFusePattern : public PatternBase {
   PATTERN_DECL_NODE(output);
 };
 
-VisDecoderAttentionFusePattern::VisDecoderAttentionFusePattern(
+DecoderAttentionFusePattern::DecoderAttentionFusePattern(
     PDPattern* pattern, const std::string& name_scope)
     : PatternBase(pattern, name_scope, name_scope) {
   auto* input_q = pattern->NewNode(input_q_repr())
@@ -179,16 +179,16 @@ VisDecoderAttentionFusePattern::VisDecoderAttentionFusePattern(
 
 }  // namespace patterns
 
-void VisDecoderAttentionXPUFusePass::ApplyVisDecoderAttentionXPUFuse(
+void DecoderAttentionXPUFusePass::ApplyDecoderAttentionXPUFuse(
     ir::Graph* graph) const {
   GraphPatternDetector gpd;
-  patterns::VisDecoderAttentionFusePattern pattern(gpd.mutable_pattern(),
-                                                   name_scope_);
+  patterns::DecoderAttentionFusePattern pattern(gpd.mutable_pattern(),
+                                                name_scope_);
   int found_subgraph_count = 0;
 
   auto handler = [&](const GraphPatternDetector::subgraph_t& subgraph,
                      Graph* graph) {
-    VLOG(4) << "handle VisDecoderAttentionXPUFusePass";
+    VLOG(4) << "handle DecoderAttentionXPUFusePass";
 
     // declare operator node's name
     GET_IR_NODE(reshape2_1);
@@ -292,22 +292,22 @@ void VisDecoderAttentionXPUFusePass::ApplyVisDecoderAttentionXPUFuse(
   AddStatis(found_subgraph_count);
 }
 
-void VisDecoderAttentionXPUFusePass::ApplyImpl(ir::Graph* graph) const {
+void DecoderAttentionXPUFusePass::ApplyImpl(ir::Graph* graph) const {
   PADDLE_ENFORCE_NOT_NULL(
       graph, platform::errors::PreconditionNotMet("graph should not be null."));
   Init(name_scope_, graph);
 
-  ApplyVisDecoderAttentionXPUFuse(graph);
+  ApplyDecoderAttentionXPUFuse(graph);
 }
 
 }  // namespace ir
 }  // namespace framework
 }  // namespace paddle
 
-REGISTER_PASS(vis_decoder_attention_xpu_fuse_pass,
-              paddle::framework::ir::VisDecoderAttentionXPUFusePass);
+REGISTER_PASS(decoder_attention_xpu_fuse_pass,
+              paddle::framework::ir::DecoderAttentionXPUFusePass);
 
-REGISTER_PASS_CAPABILITY(vis_decoder_attention_xpu_fuse_pass)
+REGISTER_PASS_CAPABILITY(decoder_attention_xpu_fuse_pass)
     .AddCombination(
         paddle::framework::compatible::OpVersionComparatorCombination().EQ(
             "qkv_attention_xpu", 0));
diff --git a/...xpu/vis_decoder_attention_xpu_fuse_pass.h → .../ir/xpu/decoder_attention_xpu_fuse_pass.h b/...xpu/vis_decoder_attention_xpu_fuse_pass.h → .../ir/xpu/decoder_attention_xpu_fuse_pass.h
@@ -89,12 +89,12 @@ Fused subgraph:
 
 */
 
-class VisDecoderAttentionXPUFusePass : public FusePassBase {
+class DecoderAttentionXPUFusePass : public FusePassBase {
  protected:
   void ApplyImpl(ir::Graph* graph) const override;
 
  private:
-  void ApplyVisDecoderAttentionXPUFuse(ir::Graph* graph) const;
+  void ApplyDecoderAttentionXPUFuse(ir::Graph* graph) const;
 
   const std::string name_scope_{"vis_decoder_attention_xpu_fuse_pass"};
 };

diff --git a/paddle/fluid/inference/api/paddle_pass_builder.cc b/paddle/fluid/inference/api/paddle_pass_builder.cc
@@ -538,12 +538,12 @@ XpuPassStrategy::XpuPassStrategy() : PassStrategy({}) {
       "generate_sequence_xpu_fuse_pass",
       "embedding_with_eltwise_add_xpu_fuse_pass",
       "qk_qkv_attention_xpu_fuse_pass",
-      "vis_decoder_attention_xpu_fuse_pass",
       "multi_encoder_xpu_fuse_pass",
       "multi_encoder_xpu_adaptive_seqlen_fuse_pass",
       "multi_encoder_xpu_slice_fuse_pass",
       "fused_multi_transformer_cachekv_layout_trans_pass",
       "fused_multi_transformer_int8_cachekv_layout_trans_pass",
+      "decoder_attention_xpu_fuse_pass",
       "one_beam_size_fuse_pass",
       "fold_interp_outsize_fuse_pass",
       "fold_two_squeeze2_fuse_pass",

diff --git a/...pu_vis_decoder_attention_xpu_fuse_pass.py → ...st_xpu_decoder_attention_xpu_fuse_pass.py b/...pu_vis_decoder_attention_xpu_fuse_pass.py → ...st_xpu_decoder_attention_xpu_fuse_pass.py
@@ -20,7 +20,7 @@
 from program_config import OpConfig, ProgramConfig, TensorConfig
 
 
-class TestVisDecoderAttentionXPUFusePass(PassAutoScanTest):
+class TestDecoderAttentionXPUFusePass(PassAutoScanTest):
     def sample_predictor_configs(self, program_config):
         config = self.create_inference_config(use_xpu=True)
         yield config, ["qkv_attention_xpu"], (1e-1, 1e-1)
@@ -164,7 +164,7 @@ def test(self):
         self.run_and_statis(
             quant=False,
             max_examples=25,
-            passes=["vis_decoder_attention_xpu_fuse_pass"],
+            passes=["decoder_attention_xpu_fuse_pass"],
         )