kernel_name_shortener now called in create_df_kerenel_top_stats and

create_df_pmc

Signed-off-by: Jose Santos <[email protected]>

src/omniperf_analyze/analysis_cli.py

@@ -45,11 +45,15 @@ def pre_processing(self):
                 filter_dispatch_ids=self._runs[d[0]].filter_dispatch_ids,
                 time_unit=self.get_args().time_unit,
                 max_stat_num=self.get_args().max_stat_num,
+                kernel_verbose = self.get_args().kernel_verbose,
             )
             # create 'mega dataframe'
             self._runs[d[0]].raw_pmc = file_io.create_df_pmc(
-                d[0], self.get_args().verbose
+                d[0], self.get_args().kernel_verbose, self.get_args().verbose
             )
+            # demangle and overwrite original 'Kernel_Name'
+            kernel_name_shortener(self._runs[d[0]].raw_pmc, self.get_args().kernel_verbose)
+
             # create the loaded table
             parser.load_table_data(
                 workload=self._runs[d[0]],
@@ -58,6 +62,7 @@ def pre_processing(self):
                 debug=self.get_args().debug,
                 verbose=self.get_args().verbose,
             )
+
 
     @demarcate
     def run_analysis(self):

src/omniperf_analyze/analysis_webui.py

@@ -123,6 +123,7 @@ def generate_from_filter(
                 filter_dispatch_ids=base_data[base_run].filter_dispatch_ids,
                 time_unit=self.get_args().time_unit,
                 max_stat_num=base_data[base_run].filter_top_n,
+                kernel_verbose=self.get_args().kernel_verbose
             )
             # Only display basic metrics if no filters are applied
             if not (disp_filt or kernel_filter or gcd_filter):
@@ -275,10 +276,11 @@ def pre_processing(self):
                 filter_dispatch_ids=self._runs[self.dest_dir].filter_dispatch_ids,
                 time_unit=args.time_unit,
                 max_stat_num=args.max_stat_num,
+                kernel_verbose=self.get_args().kernel_verbose
             )
             # create 'mega dataframe'
             self._runs[self.dest_dir].raw_pmc = file_io.create_df_pmc(
-                self.dest_dir, args.verbose
+                self.dest_dir, self.get_args().kernel_verbose,  args.verbose
             )
             # create the loaded kernel stats
             parser.load_kernel_top(self._runs[self.dest_dir], self.dest_dir)

src/utils/file_io.py

@@ -34,6 +34,7 @@
 from pathlib import Path
 from utils import schema
 from utils.utils import console_debug, console_error
+from utils.kernel_name_shortener import kernel_name_shortener
 import config
 
 # TODO: use pandas chunksize or dask to read really large csv file
@@ -89,6 +90,7 @@ def create_df_kernel_top_stats(
     filter_dispatch_ids,
     time_unit,
     max_stat_num,
+    kernel_verbose,
     sortby="sum",
 ):
     """
@@ -97,6 +99,8 @@ def create_df_kernel_top_stats(
     # NB:
     #   We even don't have to create pmc_kernel_top.csv explictly
     df = pd.read_csv(os.path.join(raw_data_dir, schema.pmc_perf_file_prefix + ".csv"))
+    # Demangle original KernelNames
+    kernel_name_shortener(df, kernel_verbose)
 
     # The logic below for filters are the same as in parser.apply_filters(),
     # which can be merged together if need it.
@@ -154,7 +158,7 @@ def create_df_kernel_top_stats(
         grouped.to_csv(os.path.join(raw_data_dir, "pmc_kernel_top.csv"), index=False)
 
 
-def create_df_pmc(raw_data_dir, verbose):
+def create_df_pmc(raw_data_dir, kernel_verbose, verbose):
     """
     Load all raw pmc counters and join into one df.
     """
@@ -170,6 +174,8 @@ def create_df_pmc(raw_data_dir, verbose):
                 f == schema.pmc_perf_file_prefix + ".csv"
             ):
                 tmp_df = pd.read_csv(os.path.join(root, f))
+                # Demangle original KernelNames
+                kernel_name_shortener(tmp_df, kernel_verbose)
                 dfs.append(tmp_df)
                 coll_levels.append(f[:-4])
     final_df = pd.concat(dfs, keys=coll_levels, axis=1, copy=False)

src/utils/kernel_name_shortener.py

@@ -23,12 +23,12 @@
 ##############################################################################el
 
 import os
-import glob
+import logging
 import re
 import subprocess
 import pandas as pd
 
-from utils.utils import console_error, console_debug, console_log
+from utils.utils import console_error, console_log
 
 cache = dict()
 
@@ -59,6 +59,7 @@ def shorten_file(df, level):
 
                 demangled_name, e = proc.communicate()
                 demangled_name = str(demangled_name, "UTF-8").strip()
+                demangled_name="hello"
 
                 # cache miss, add the shortened name to the dictionary
                 new_name = ""
@@ -109,7 +110,7 @@ def shorten_file(df, level):
                             current_level -= 1
                         curr_index += 1
 
-                cache[original_name] = new_name
+                cache[original_name] = "hello"
                 if new_name == None or new_name == "":
                     cache[original_name] = demangled_name
 
@@ -125,18 +126,13 @@ def shorten_file(df, level):
                 "Could not resolve c++filt in expected directory: %s" % cpp_filt
             )
 
-        for fpath in glob.glob(workload_dir + "/[SQpmc]*.csv"):
-            try:
-                orig_df = pd.read_csv(
-                    fpath,
-                    on_bad_lines="skip",
-                    engine="python",
-                )
-                modified_df = shorten_file(orig_df, level)
+        try:
+                modified_df = shorten_file(df, level)
+                console_log("profiling", "Kernel_Name shortening complete.")
                 return modified_df
-            except pd.errors.EmptyDataError:
-                console_debug(
-                    "profiling", "Skipping shortening on empty csv: %s" % str(fpath)
+        except pd.errors.EmptyDataError:
+                logging.debug("[profiling] Skipping shortening on empty csv"
                 )
 
-        console_log("profiling", "Kernel_Name shortening complete.")
+
+