Addressing review comments - 1

mlir/utils/performance/analysis/quickTuningGen.py

@@ -59,7 +59,7 @@ def replace_section(
         with open(file_path, 'w') as file:
             file.write(content)
 
-    def isAccel(self, arch, datatype):
+    def is_accel(self, arch, datatype):
         instruction_type = self.get_instruction_type(arch, datatype)
         if instruction_type == "XDL" or instruction_type == "Wmma":
             return True
@@ -97,33 +97,33 @@ def init_inc_file(self, file_path):
             file.write("// THIS IS AN AUTOGENERATED FILE.\n")
             file.write("// DO NOT EDIT THIS FILE DIRECTLY!\n\n")
             file.write("// clang-format off\n")
-            for instrction_type, datatypes in instruction_types_to_datatypes.items():
-                file.write(f"#ifdef {instrction_type}_DEFINITIONS_GEN\n\n")
+            for instruction_type, datatypes in instruction_types_to_datatypes.items():
+                file.write(f"#ifdef {instruction_type}_DEFINITIONS_GEN\n\n")
                 for datatype in datatypes:
                     for marker in markers:
                         file.write(
-                            f"{marker}_{instrction_type}_{datatype}_DEFS\n\n")
+                            f"{marker}_{instruction_type}_{datatype}_DEFS\n\n")
                 file.write(f"#endif\n\n")
 
-                file.write(f"#ifdef {instrction_type}_DECLARATIONS_GEN\n\n")
+                file.write(f"#ifdef {instruction_type}_DECLARATIONS_GEN\n\n")
                 for datatype in datatypes:
                     for marker in markers:
                         file.write(
-                            f"{marker}_{instrction_type}_{datatype}_DECS\n\n")
+                            f"{marker}_{instruction_type}_{datatype}_DECS\n\n")
                 file.write(f"#endif\n\n")
 
     def get_init_params_definitions(self, arch, dtype, op):  
         """  
         Generates initialization parameter definitions for a given data type and operation. 
         """ 
-        accel_type = 'Accel' if self.isAccel(arch, dtype) else 'NonAccel'  
+        accel_type = 'Accel' if self.is_accel(arch, dtype) else 'NonAccel'
         instruction_type = self.get_instruction_type(arch, dtype)  
         op_cap = op.capitalize()  
 
         if dtype == 'f32':  
             init_params = f"initParameters{op_cap}"  
-            n_init_params = f"nInitParameters{op_cap}"  
-            if not self.isAccel(arch, dtype):  
+            n_init_params = f"nInitParameters{op_cap}"
+            if not self.is_accel(arch, dtype):
                 instruction_type = ''
         elif dtype == 'f16':  
             init_params = f"initParametersFp16{op_cap}"  
@@ -141,7 +141,7 @@ def get_init_params_declaration(self, arch, dtype, op):
         Generates initialization parameter declarations for a given data type and operation. 
         """ 
         op_cap = op.capitalize()  
-        accel_type = 'Accel' if self.isAccel(arch, dtype) else 'NonAccel'
+        accel_type = 'Accel' if self.is_accel(arch, dtype) else 'NonAccel'
 
         if dtype == 'f32':  
             init_params = f"initParameters{op_cap}"  
@@ -238,7 +238,7 @@ def get_unique_perfconfigs_list(self, problems_to_perfconfigs):
 
     def get_top_n_perfconfigs_per_problems(self, df, targetColumns):
         """
-        Identifies the top perfcofnigs for each problem based on a threshold
+        Identifies the top perfconfigs for each problem based on a threshold
         """
         grouped = df.groupby(targetColumns)
         problem_df = {}
@@ -253,7 +253,7 @@ def find(self):
         """
         Finds the minimal set of perfconfigs that cover all
         problems using set cover optimizaiton.
-        Returns : A dictionary containing data types as keys and thier
+        Returns : A dictionary containing data types as keys and their
                   corresponding selected perfconfigs.
         """
         result = {}
@@ -291,26 +291,20 @@ def find(self):
 
             n = len(problems)
             m = len(perfconfigs)
-            problem_to_index = {
-                problem: idx for idx,
-                problem in enumerate(problems)}
-            perfconfig_to_index = {
-                perfconfig: idx for idx,
-                perfconfig in enumerate(perfconfigs)}
 
             # Create coverage matrix
             A = np.zeros((n, m), dtype=int)
-            for problem, perfconfig_list in problems_to_perfconfigs.items():
-                i = problem_to_index[problem]
-                for perfconfig in perfconfig_list:
-                    j = perfconfig_to_index[perfconfig]
+            for i, (problem, perfconfig_list) in enumerate(problems_to_perfconfigs.items()):
+                for j, perfconfig in enumerate(perfconfig_list):
                     A[i][j] = 1
 
             # Linear programming model to minimize the number of perfconfigs
             prob = pulp.LpProblem("SetCoverProblems", pulp.LpMinimize)
             x = pulp.LpVariable.dicts("x", range(m), cat='Binary')
+            # Set up objective function to minimize the sum of selected perfconfigs
             prob += pulp.lpSum([x[j]] for j in range(m))
             for i in range(n):
+            # Add a constraint for each problem ensuring at least one perfconfig is selected
                 prob += pulp.lpSum([A[i][j] * x[j]
                                    for j in range(m)]) >= 1, f"Cover_problem_{i}"