tools: sof_perf_analyzer: change module CPC calculation

Previously, module CPC is calculated by multiplying the mean
value of cpu_peak with a margin scalar, however, abnormal peak
cycles are finally proved that it is caused by IRQ, with further
team discussion, an agreement was reached to use module average
multiplied with a margin as new way for CPC calculation.

Signed-off-by: Baofeng Tian <[email protected]>

tools/sof_perf_analyzer.py

@@ -27,8 +27,13 @@
 
 import pandas as pd
 
-# currently, keep CPC = max(module peak) * CPC_MARGIN
-CPC_MARGIN = 1.1
+# CPC_MARGIN set to 1.5 because some modules have not running code, for example
+# volume: volume have two kind of ramp type, linear and log, it only be runned when
+# volume change happens.
+# copier: there is attenuation inside, normally, it will not be runned.
+# So we set a relative high margin for avg cycles.
+# CPC = AVG(module) * CPC_MARGIN
+CPC_MARGIN = 1.5
 
 @dataclass()
 class TraceItem:
@@ -243,7 +248,7 @@ def analyze_perf_info():
     perf_stats.columns = ['CPU_AVG(MIN)', 'CPU_AVG(AVG)', 'CPU_AVG(MAX)',
                           'CPU_PEAK(MIN)', 'CPU_PEAK(AVG)', 'CPU_PEAK(MAX)']
     perf_stats['PEAK(MAX)/AVG(AVG)'] = perf_stats['CPU_PEAK(MAX)'] / perf_stats['CPU_AVG(AVG)']
-    perf_stats['MODULE_CPC'] = perf_info.groupby('COMP_ID')['CPU_PEAK'].max() * CPC_MARGIN
+    perf_stats['MODULE_CPC'] = perf_info.groupby('COMP_ID')['CPU_AVG'].mean() * CPC_MARGIN
     # change data type from float to int
     perf_stats['MODULE_CPC'] = perf_stats['MODULE_CPC'].astype(int)