Merge branch 'faster_mul' of https://github.com/DeepLink-org/deeplink.framework.dev into faster_mul

Author: Wrench-Git

.github/workflows/main.yml

@@ -125,7 +125,7 @@ jobs:
     needs: [Build-Camb]
     runs-on: github-poc-ci
     env:
-      MLU_REQUESTS: 1
+      MLU_REQUESTS: 4 
     steps:
       - name: Run-test
         run: |
@@ -210,7 +210,7 @@ jobs:
     needs: [Build-Camb-Pt211]
     runs-on: github-poc-ci
     env:
-      MLU_REQUESTS: 1
+      MLU_REQUESTS: 4
     steps:
       - name: Run-test
         run: |

dicp/dicp/vendor/AscendGraph/codegen/ascend.py

@@ -61,7 +61,6 @@ def __init__(self, graph, aten_graph=None, folder=None, graph_key=None):
         self.py_output_names = []
         self.graph_output_names = []
         self.build_options = []
-        self.output_nodes = []
 
         self.folder = folder
         self.graph_key = graph_key
@@ -195,26 +194,8 @@ def parse_outputs(self):
                 self.py_output_names.append(str(node))
         self.output_args = real_output_args
 
-        if len(self.sym_in_args) > 0 or len(self.sym_to_inputs) > 0:
-            for output in self.output_args:
-                info = {}
-                info['format'] = 'ND'
-                if hasattr(output, 'meta'):
-                    output = output.meta['val']
-                if isinstance(output, torch.SymInt):
-                    info['data_type'] = 'INT32'
-                elif isinstance(output, torch.SymBool):
-                    info['data_type'] = 'BOOL'
-                info['data_type'] = get_ascend_dtype(output.dtype)
-                self.output_nodes.append(info)
         if len(self.assign_args) > 0:
             self.graph_output_names.extend(list(zip(*self.assign_args))[0])
-            for item in self.assign_args:
-                index = item[1]
-                info = {}
-                info['format'] = self.data_nodes[index]['format']
-                info['data_type'] = self.data_nodes[index]['data_type']
-                self.output_nodes.append(info)
 
     def gen_import_code(self):
         self.import_code.splice(
@@ -225,7 +206,7 @@ def gen_import_code(self):
                 import random
                 from torch import empty_strided, as_strided, device
                 from dicp.dynamo_bridge.compile import AsyncCompileKernel
-                from dicp.vendor.AscendGraph.compile_job import AscendGECompileAclRunJob, AscendGECompileGERunJob
+                from dicp.vendor.AscendGraph.compile_job import AscendCompileJob
 
                 aten = torch.ops.aten
                 assert_size_stride = torch._C._dynamo.guards.assert_size_stride
@@ -480,19 +461,16 @@ def gen_graph_json(self):
             "build_options": self.build_options,
             "data_nodes": self.data_nodes,
             "common_nodes": self.common_nodes,
-            "output_nodes": self.output_nodes,
         }
         self.remove_symint(graph)
         return json.dumps(graph)
 
     def gen_compile_graph_code(self):
         compile_graph_code = IndentedBuffer()
         graph_json = self.gen_graph_json()
-        compile_job_type = os.environ.get("DICP_ASCEND_COMPILE_JOB_TYPE", "AscendGECompileGERunJob")
-        assert compile_job_type in ["AscendGECompileGERunJob", "AscendGECompileAclRunJob"]
         compile_graph_code.splice(
             f"""
-                ascend_compile_job = {compile_job_type}('''{graph_json}''')
+                ascend_compile_job = AscendCompileJob('''{graph_json}''')
                 async_compile = AsyncCompileKernel()
                 kernel_cpp_0 = async_compile.compile_kernel(ascend_compile_job)
             """, strip=True