I’m profiling a C code with a python automated algorithm using GNU gprof via subprocess.run calls. I’m in trouble to remove a function (let’s say a clean_cache function) from results. I’ve already tried to use -P, -Q and --no-time flags but these only remove the function to be printed in results output. I want to remove a function to be computed at all. I’ll leave my code in the end of this session. Any help will be greatly appreciated!
[some imports here!]
class Gprof:
"""
Flags:
-b : Print data without explanation.
"""
FLAGS = "-b --no-time=clean_cache -Pclean_cache -Qclean_cache"
COMMAND = f"gprof {FLAGS}"
CALL_GRAPH_COLS = [
'index',
'% time',
'self',
'children',
'called',
'name'
]
FLAT_PROFILE_COLS = [
'% time',
'cumulative seconds',
'self seconds',
'calls',
'self ms/call',
'total ms/call',
'name'
]
def main():
[...]
for i in tqdm.tqdm(range(runs)):
# sys.argv[2] is the name of C algorithm file
_ = subprocess.run(f"{sys.argv[2]}", capture_output = True, shell = True)
result = subprocess.run(f"{Gprof.COMMAND} {sys.argv[2]} {gmon_filename}",
capture_output = True, shell = True)
[...]
Example of output:
Flat profile:
Each sample counts as 0.01 seconds.
no time accumulated
% cumulative self self total
time seconds seconds calls Ts/call Ts/call name
0.00 0.00 0.00 12095 0.00 0.00 swap_scalars
0.00 0.00 0.00 2049 0.00 0.00 copy_array
0.00 0.00 0.00 1535 0.00 0.00 heapify
0.00 0.00 0.00 1024 0.00 0.00 random_scalar
0.00 0.00 0.00 1023 0.00 0.00 merge
0.00 0.00 0.00 931 0.00 0.00 median_pivot
0.00 0.00 0.00 3 0.00 0.00 is_sorted
0.00 0.00 0.00 1 0.00 0.00 heap_sort
0.00 0.00 0.00 1 0.00 0.00 merge_sort
0.00 0.00 0.00 1 0.00 0.00 quick_sort
0.00 0.00 0.00 1 0.00 0.00 randomize_array
Call graph
granularity: each sample hit covers 2 byte(s) no time propagated
index % time self children called name
0.00 0.00 1023/12095 heap_sort [9]
0.00 0.00 2755/12095 quick_sort [11]
0.00 0.00 8317/12095 heapify [3]
[1] 0.0 0.00 0.00 12095 swap_scalars [1]
-----------------------------------------------
0.00 0.00 3/2049 main [18]
0.00 0.00 2046/2049 merge_sort [10]
[2] 0.0 0.00 0.00 2049 copy_array [2]
-----------------------------------------------
8317 heapify [3]
0.00 0.00 1535/1535 heap_sort [9]
[3] 0.0 0.00 0.00 1535+8317 heapify [3]
0.00 0.00 8317/12095 swap_scalars [1]
8317 heapify [3]
-----------------------------------------------
0.00 0.00 1024/1024 randomize_array [12]
[4] 0.0 0.00 0.00 1024 random_scalar [4]
-----------------------------------------------
0.00 0.00 1023/1023 merge_sort [10]
[5] 0.0 0.00 0.00 1023 merge [5]
-----------------------------------------------
0.00 0.00 931/931 quick_sort [11]
[6] 0.0 0.00 0.00 931 median_pivot [6]
-----------------------------------------------
0.00 0.00 3/3 main [18]
[8] 0.0 0.00 0.00 3 is_sorted [8]
-----------------------------------------------
0.00 0.00 1/1 main [18]
[9] 0.0 0.00 0.00 1 heap_sort [9]
0.00 0.00 1535/1535 heapify [3]
0.00 0.00 1023/12095 swap_scalars [1]
-----------------------------------------------
2046 merge_sort [10]
0.00 0.00 1/1 main [18]
[10] 0.0 0.00 0.00 1+2046 merge_sort [10]
0.00 0.00 2046/2049 copy_array [2]
0.00 0.00 1023/1023 merge [5]
2046 merge_sort [10]
-----------------------------------------------
930 quick_sort [11]
0.00 0.00 1/1 main [18]
[11] 0.0 0.00 0.00 1+930 quick_sort [11]
0.00 0.00 2755/12095 swap_scalars [1]
0.00 0.00 931/931 median_pivot [6]
930 quick_sort [11]
-----------------------------------------------
0.00 0.00 1/1 main [18]
[12] 0.0 0.00 0.00 1 randomize_array [12]
0.00 0.00 1024/1024 random_scalar [4]
-----------------------------------------------
Index by function name
(7) clean_cache [8] is_sorted [11] quick_sort
[2] copy_array [6] median_pivot [4] random_scalar
[9] heap_sort [5] merge [12] randomize_array
[3] heapify [10] merge_sort [1] swap_scalars
As we can see, all functions have 0% of CPU time. What shows that clean_cache function is only being removed from output.