mdk/formations
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Working on python-perfs.

Browse Source
This commit is contained in:
Julien Palard 2023-09-22 15:33:14 +02:00
parent 8d923687e2
commit 188e4bd870
Signed by: mdk
GPG Key ID: 0EFC1AC1006886F8
22 changed files with 933 additions and 115 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

44
python-perfs/bin/cache
View File

@ -1,17 +1,37 @@
#!/bin/sh #!/usr/bin/env python
ARGS="$*" import sys
import shlex
from subprocess import run, STDOUT
from time import perf_counter
from textwrap import wrap
args="--ttl 1y --cache-dir .cache" args = sys.argv[1:]
while [ "$1" != "--" ] bkt_args = []
do cmd_args = args
args="$args $1"
shift
done
before="$(date +"%s.%N")" if '--' in args:
bkt $args "$@" 2>&1 at = args.index('--')
after="$(date +"%s.%N")" bkt_args, cmd_args = args[:at], args[at+1:]
printf "%s: %.2fs\n\n" "$ARGS" "$(echo "$after - $before"|bc)" >&2 bkt_args = ["--ttl", "1y", "--cache-dir", ".cache"] + bkt_args
def wrap_bash_command(cmd):
"""Given args as a list of string, wraps
the args to look like a bash cmd."""
cmd = [arg.replace(" ", " ") for arg in cmd]
lines = wrap(shlex.join(cmd), width=60)
lines = [line.replace(" ", " ") for line in lines]
return " \\\n> ".join(lines)
before = perf_counter()
print(f"$ {shlex.join(cmd_args)}", file=sys.stderr)
print(f"$ {wrap_bash_command(cmd_args)}", flush=True)
run(["bkt"] + bkt_args + ["--"] + cmd_args, stderr=STDOUT)
after = perf_counter()
print(f"$ {shlex.join(cmd_args)} # took {after-before:.2f}s\n", file=sys.stderr)

0
python-perfs/include/big.o.py → python-perfs/examples/big.o.py
View File

0
python-perfs/include/collatz_length.py → python-perfs/examples/collatz_length.py
View File

0
python-perfs/include/collatz_length_annotated.py → python-perfs/examples/collatz_length_annotated.py
View File

0
python-perfs/include/collatz_length_nuitka.py → python-perfs/examples/collatz_length_cython.py
View File

2
python-perfs/examples/collatz_length_cython_to_c.pyx Normal file
View File

@ -0,0 +1,2 @@
cdef extern from "my_collatz_length.c":
cpdef int collatz_length(long n)

0
python-perfs/include/collatz_length_mypy.py → python-perfs/examples/collatz_length_mypy.py
View File

7
python-perfs/examples/collatz_length_nuitka.py Normal file
View File

@ -0,0 +1,7 @@
def collatz_length(n):
if n == 1:
return 0
if n % 2 == 0:
return 1 + collatz_length(n // 2)
else:
return 1 + collatz_length(n * 3 + 1)

0
python-perfs/include/collatz_length_numba.py → python-perfs/examples/collatz_length_numba.py
View File

0
python-perfs/include/collatz_length_pythran.py → python-perfs/examples/collatz_length_pythran.py
View File

0
python-perfs/include/find_in_list.py → python-perfs/examples/find_in_list.py
View File

0
python-perfs/include/phi1.py → python-perfs/examples/phi1.py
View File

0
python-perfs/include/phi2.py → python-perfs/examples/phi2.py
View File

0
python-perfs/include/phi3.py → python-perfs/examples/phi3.py
View File

0
python-perfs/include/phi4.py → python-perfs/examples/phi4.py
View File

0
python-perfs/include/phi5.py → python-perfs/examples/phi5.py
View File

0
python-perfs/include/prenom-1.py → python-perfs/examples/prenom-1.py
View File

0
python-perfs/include/prenom-2.py → python-perfs/examples/prenom-2.py
View File

0
python-perfs/include/prenom.py → python-perfs/examples/prenom.py
View File

705
python-perfs/examples/profile.json Normal file
View File

@ -0,0 +1,705 @@
{
"alloc_samples": 0,
"elapsed_time_sec": 35.077271699905396,
"files": {
"phi5.py": {
"functions": [
{
"line": "fib",
"lineno": 5,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 4.2576382717302454,
"n_core_utilization": 0.062425365799925484,
"n_cpu_percent_c": 0.0034262080640758584,
"n_cpu_percent_python": 0.7126821564209772,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0009376888160889245,
"n_usage_fraction": 0
},
{
"line": "approx_phi",
"lineno": 12,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.6998240221201137,
"n_core_utilization": 0.06237139811505196,
"n_cpu_percent_c": 96.09209325974999,
"n_cpu_percent_python": 0.3418254092394039,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.20029089301487063,
"n_usage_fraction": 0
},
{
"line": "approx_phi_up_to",
"lineno": 16,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.06219424868052924,
"n_cpu_percent_c": 2.351736421918283,
"n_cpu_percent_python": 0.2556447055821568,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.012818642838908139,
"n_usage_fraction": 0
}
],
"imports": [
"from decimal import Decimal, localcontext",
"from itertools import count",
"from functools import cache",
"import sys"
],
"leaks": {},
"lines": [
{
"end_region_line": 1,
"line": "from decimal import Decimal, localcontext\n",
"lineno": 1,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.0,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0,
"n_usage_fraction": 0,
"start_region_line": 1
},
{
"end_region_line": 2,
"line": "from itertools import count\n",
"lineno": 2,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.0,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0,
"n_usage_fraction": 0,
"start_region_line": 2
},
{
"end_region_line": 3,
"line": "from functools import cache\n",
"lineno": 3,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.0,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0,
"n_usage_fraction": 0,
"start_region_line": 3
},
{
"end_region_line": 4,
"line": "\n",
"lineno": 4,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.0,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0,
"n_usage_fraction": 0,
"start_region_line": 4
},
{
"end_region_line": 5,
"line": "@cache\n",
"lineno": 5,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.0,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0,
"n_usage_fraction": 0,
"start_region_line": 5
},
{
"end_region_line": 10,
"line": "def fib(n):\n",
"lineno": 6,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.0,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0,
"n_usage_fraction": 0,
"start_region_line": 6
},
{
"end_region_line": 10,
"line": " if n in (0, 1):\n",
"lineno": 7,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.0,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0,
"n_usage_fraction": 0,
"start_region_line": 6
},
{
"end_region_line": 10,
"line": " return 1\n",
"lineno": 8,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.0,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0,
"n_usage_fraction": 0,
"start_region_line": 6
},
{
"end_region_line": 10,
"line": " x = n // 2\n",
"lineno": 9,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.0,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0,
"n_usage_fraction": 0,
"start_region_line": 6
},
{
"end_region_line": 10,
"line": " return fib(x - 1) * fib(n - x - 1) + fib(x) * fib(n - x)\n",
"lineno": 10,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 4.2576382717302454,
"n_core_utilization": 0.062425365799925484,
"n_cpu_percent_c": 0.0034262080640758584,
"n_cpu_percent_python": 0.7126821564209772,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0009376888160889245,
"n_usage_fraction": 0,
"start_region_line": 6
},
{
"end_region_line": 11,
"line": "\n",
"lineno": 11,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.0,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0,
"n_usage_fraction": 0,
"start_region_line": 11
},
{
"end_region_line": 12,
"line": "@cache\n",
"lineno": 12,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.0,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0,
"n_usage_fraction": 0,
"start_region_line": 12
},
{
"end_region_line": 14,
"line": "def approx_phi(n):\n",
"lineno": 13,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.0,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0,
"n_usage_fraction": 0,
"start_region_line": 13
},
{
"end_region_line": 14,
"line": " return Decimal(fib(n + 1)) / Decimal(fib(n))\n",
"lineno": 14,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.6998240221201137,
"n_core_utilization": 0.06237139811505196,
"n_cpu_percent_c": 96.09209325974997,
"n_cpu_percent_python": 0.3418254092394039,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.20029089301488137,
"n_usage_fraction": 0,
"start_region_line": 13
},
{
"end_region_line": 15,
"line": "\n",
"lineno": 15,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.0,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0,
"n_usage_fraction": 0,
"start_region_line": 15
},
{
"end_region_line": 23,
"line": "def approx_phi_up_to(n_digits):\n",
"lineno": 16,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.0,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0,
"n_usage_fraction": 0,
"start_region_line": 16
},
{
"end_region_line": 23,
"line": " with localcontext() as ctx:\n",
"lineno": 17,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.0,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0,
"n_usage_fraction": 0,
"start_region_line": 16
},
{
"end_region_line": 23,
"line": " ctx.prec = n_digits + 1\n",
"lineno": 18,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.0,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0,
"n_usage_fraction": 0,
"start_region_line": 16
},
{
"end_region_line": 23,
"line": " for n in count():\n",
"lineno": 19,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.0,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0,
"n_usage_fraction": 0,
"start_region_line": 19
},
{
"end_region_line": 23,
"line": " step1 = approx_phi(2 ** n)\n",
"lineno": 20,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.0,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0,
"n_usage_fraction": 0,
"start_region_line": 19
},
{
"end_region_line": 23,
"line": " step2 = approx_phi(2 ** (n+1))\n",
"lineno": 21,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.06219424868052924,
"n_cpu_percent_c": 2.351736421918283,
"n_cpu_percent_python": 0.2556447055821568,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.012818642838908139,
"n_usage_fraction": 0,
"start_region_line": 19
},
{
"end_region_line": 23,
"line": " if step1 == step2:\n",
"lineno": 22,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.0,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0,
"n_usage_fraction": 0,
"start_region_line": 19
},
{
"end_region_line": 23,
"line": " return step1\n",
"lineno": 23,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.0,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0,
"n_usage_fraction": 0,
"start_region_line": 19
},
{
"end_region_line": 24,
"line": "\n",
"lineno": 24,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.0,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0,
"n_usage_fraction": 0,
"start_region_line": 24
},
{
"end_region_line": 25,
"line": "import sys\n",
"lineno": 25,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.0,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0,
"n_usage_fraction": 0,
"start_region_line": 25
},
{
"end_region_line": 26,
"line": "\n",
"lineno": 26,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.0,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0,
"n_usage_fraction": 0,
"start_region_line": 26
},
{
"end_region_line": 27,
"line": "if __name__ == \"__main__\":\n",
"lineno": 27,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.0,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.0,
"n_usage_fraction": 0,
"start_region_line": 27
},
{
"end_region_line": 28,
"line": " print(approx_phi_up_to(int(sys.argv[1])))\n",
"lineno": 28,
"memory_samples": [],
"n_avg_mb": 0.0,
"n_copy_mb_s": 0.0,
"n_core_utilization": 0.005697373432005744,
"n_cpu_percent_c": 0.0,
"n_cpu_percent_python": 0.0026020692392709254,
"n_gpu_avg_memory_mb": 0.0,
"n_gpu_peak_memory_mb": 0.0,
"n_gpu_percent": 0,
"n_growth_mb": 0.0,
"n_malloc_mb": 0.0,
"n_mallocs": 0,
"n_peak_mb": 0.0,
"n_python_fraction": 0,
"n_sys_percent": 0.025942545115975865,
"n_usage_fraction": 0,
"start_region_line": 28
}
],
"percent_cpu_time": 100.0
}
},
"gpu": false,
"growth_rate": 0.0,
"max_footprint_fname": null,
"max_footprint_lineno": null,
"max_footprint_mb": 0,
"memory": true,
"program": "phi5.py",
"samples": [],
"stacks": []
}

39
python-perfs/examples/sandpile.py Normal file
View File

@ -0,0 +1,39 @@
import sys
def should_apply_gravity(terrain):
width = len(terrain)
return any(terrain[x][y] >= 4 for x in range(width) for y in range(width))
def show_terrain(terrain):
width = len(terrain)
for x in range(width):
for y in range(width):
print(" ·●⬤"[terrain[x][y]], end="")
print()
def apply_gravity(terrain):
width = len(terrain)
for x in range(width):
for y in range(width):
if terrain[x][y] >= 4:
terrain[x][y] -= 4
terrain[x - 1][y] += 1
terrain[x + 1][y] += 1
terrain[x][y + 1] += 1
terrain[x][y - 1] += 1
def main(height, show=True):
width = int(height ** .5) + 1
terrain = [[0] * width for _ in range(width)]
terrain[width // 2][width // 2] = height
while should_apply_gravity(terrain):
apply_gravity(terrain)
if show:
show_terrain(terrain)
if __name__ == "__main__":
main(int(sys.argv[1]))

251
python-perfs/perfs.md
View File

@ -1,4 +1,4 @@
# Les Performances en Python # Les performances en Python
par par
@ -31,7 +31,7 @@ O(n!) Factorielle
notes: notes:
Il faut les grapher pour s'en rendre compte : cf. include/big.o.py Il faut les grapher pour s'en rendre compte : cf. examples/big.o.py
## Comparaison asymptotique ## Comparaison asymptotique
@ -73,13 +73,13 @@ de temps mais peut-être une opération de plus.
## O(log n) ## O(log n)
```python ```python
#!sed -n '/def index/,/raise ValueError/p' include/find_in_list.py #!sed -n '/def index/,/raise ValueError/p' examples/find_in_list.py
``` ```
## O(n) ## O(n)
```python ```python
#!sed -n '/def dumb_index/,/raise ValueError/p' include/find_in_list.py #!sed -n '/def dumb_index/,/raise ValueError/p' examples/find_in_list.py
``` ```
## O(n log n) ## O(n log n)
@ -92,7 +92,7 @@ Typique d'algorithmes de tris.
## Les mesures de complexité ## Les mesures de complexité
- De temps (CPU consommé). - De temps (CPU consommé).
- D'espace (Mémoire consommée). - D'espace (mémoire consommée).
- Dans le meilleur des cas. - Dans le meilleur des cas.
- Dans le pire des cas. - Dans le pire des cas.
- Dans le cas moyen. - Dans le cas moyen.
@ -130,15 +130,9 @@ Mais retenir par cœur la complexité de quelques structures
## Le cas typique ## Le cas typique
```shell ```shell
$ python -m pyperf timeit \ #!cache python -m pyperf timeit --setup 'container = list(range(10_000_000))' '10_000_001 in container'
> --setup 'container = list(range(10_000_000))' \
> '10_000_001 in container'
#!cache -- python -m pyperf timeit --fast -s 'container = list(range(10_000_000))' '10_000_001 in container'
$ python -m pyperf timeit \ #!cache python -m pyperf timeit --setup 'container = set(range(10_000_000))' '10_000_001 in container'
> --setup 'container = set(range(10_000_000))' \
> '10_000_001 in container'
#!cache -- python -m pyperf timeit --fast -s 'container = set(range(10_000_000))' '10_000_001 in container'
``` ```
Pourquoi une si grande différence !? Pourquoi une si grande différence !?
@ -149,6 +143,19 @@ notes:
C'est l'heure du live coding ! C'est l'heure du live coding !
# À vous !
Simulons un tas de sable.
Ne vous souciez pas des perfs, on s'en occupera.
Vous avez 10mn.
notes:
voir sandpile.py
# Les outils # Les outils
## Les outils en ligne de commande ## Les outils en ligne de commande
@ -156,8 +163,7 @@ C'est l'heure du live coding !
`time`, un outil POSIX, mais aussi une fonction native de bash : `time`, un outil POSIX, mais aussi une fonction native de bash :
```shell ```shell
$ time python -c 'container = set(range(10_000_000))' #!cache time -p python examples/sandpile.py 16
#!cache -- time -p python -c 'container = set(range(10_000_000))'
``` ```
Mais `time` ne teste qu'une fois. Mais `time` ne teste qu'une fois.
@ -173,10 +179,8 @@ sys 0m0.195s # Temps CPU passé dans le kernel
`hyperfine` teste plusieurs fois : `hyperfine` teste plusieurs fois :
```text ```shell
$ hyperfine "python -c 'container = set(range(10_000_000))'" #!cache hyperfine "python examples/sandpile.py 1000"
Benchmark 1: python -c 'container = set(range(10_000_000))'
Time (mean ± σ): 735.6 ms ± 11.2 ms
``` ```
@ -184,10 +188,8 @@ Benchmark 1: python -c 'container = set(range(10_000_000))'
Mais attention, démarrer un processus Python n'est pas gratuit : Mais attention, démarrer un processus Python n'est pas gratuit :
```python ```shell
$ hyperfine "python -c pass" #!cache hyperfine "python -c pass"
Benchmark 1: python -c pass
Time (mean ± σ): 19.4 ms ± 0.6 ms
``` ```
notes: notes:
@ -199,7 +201,7 @@ N'essayez pas de retenir les chiffres, retenez les faits.
Et puis il peut dépendre de la version de Python, des options de compilation, ... : Et puis il peut dépendre de la version de Python, des options de compilation, ... :
```text ```shell
$ hyperfine "~/.local/bin/python3.10 -c pass" # Avec pydebug $ hyperfine "~/.local/bin/python3.10 -c pass" # Avec pydebug
Benchmark 1: ~/.local/bin/python3.10 -c pass Benchmark 1: ~/.local/bin/python3.10 -c pass
Time (mean ± σ): 37.6 ms ± 0.6 ms Time (mean ± σ): 37.6 ms ± 0.6 ms
@ -218,19 +220,19 @@ Leur parler de `--enable-optimizations` (PGO).
Timeit c'est dans la stdlib de Python, ça s'utilise en ligne de commande ou depuis Python. Timeit c'est dans la stdlib de Python, ça s'utilise en ligne de commande ou depuis Python.
```shell
#!cache python -m timeit -s 'import sandpile' 'sandpile.main(1000, show=False)'
```
Mais il ne vous donne que le meilleur temps, ce qui n'est pas toujours représentatif.
## pyperf ## pyperf
C'est l'équivalent d'hyperfine mais exécutant du Python plutôt qu'un programme : Il n'est pas dans la stdlib, mais il replace avantageusement `timeit` :
```shell ```shell
$ ~/.local/bin/python3.10 -m pyperf timeit pass #!cache python -m pyperf timeit --setup 'import sandpile' 'sandpile.main(1000, show=False)'
.....................
Mean +- std dev: 7.33 ns +- 0.18 ns
$ /usr/bin/python3.10 -m pyperf timeit pass
.....................
Mean +- std dev: 6.10 ns +- 0.11 ns
``` ```
notes: notes:
@ -240,17 +242,39 @@ démarrer** puis d'exécuter `pass`, ici on teste combien de temps ça
prend d'exécuter `pass`. prend d'exécuter `pass`.
# cProfile ## pyperf
time, timeit, hyperfine, pyperf c'est bien pour mesurer, comparer. Digression, `pyperf` peut aussi remplacer `hyperfine` :
cProfile nous aider à trouver la fonction coupable. ```shell
#!cache python -m pyperf command python examples/sandpile.py 1000
```
notes:
Avec hyperfine on teste combien de temps ça prend à Python **de
démarrer** puis d'exécuter `pass`, ici on teste combien de temps ça
prend d'exécuter `pass`.
## Les outils — À vous !
Effectuez quelques mesures sur votre implémentation.
Tentez d'en déterminer la complexité en fonction du nombre de grains.
# Profilage
`time`, `timeit`, `hyperfine`, `pyperf` c'est bien pour mesurer, comparer.
`cProfile` peut nous aider à trouver la fonction coupable.
## cProfile, exemple ## cProfile, exemple
```python ```python
#!sed -n '/def fib/,/return approx/p' include/phi1.py #!sed -n '/def fib/,/return approx/p' examples/phi1.py
``` ```
@ -259,12 +283,11 @@ cProfile nous aider à trouver la fonction coupable.
Testons : Testons :
```python ```python
#!sed -n '/import sys/,$p' include/phi1.py #!sed -n '/import sys/,$p' examples/phi1.py
``` ```
```text ```text
$ time python phi1.py 10 #!cache time -p python examples/phi1.py 10
#!cache -- time -p python include/phi1.py 10
``` ```
@ -275,10 +298,10 @@ C'est déjà lent, et pour `20` c'est interminable...
Sortons cProfile : Sortons cProfile :
```text ```shell
$ python -m cProfile --sort cumulative phi1.py 10 $ python -m cProfile --sort cumulative phi1.py 10
... ...
#!cache -- python -m cProfile --sort cumulative include/phi1.py 10 | sed -n '/fib\|function calls/{s/ \+/ /g;s/^ *//;p}' #!cache python -m cProfile --sort cumulative examples/phi1.py 10 | sed -n '/fib\|function calls/{s/ \+/ /g;s/^ *//;p}'
... ...
``` ```
@ -291,16 +314,18 @@ C'est donc `fib` la coupable :
Cachons les résultats de `fib` : Cachons les résultats de `fib` :
```python ```python
#!sed -n '/import cache/,/return fib/p' include/phi2.py #!sed -n '/import cache/,/return fib/p' examples/phi2.py
``` ```
## cProfile, exemple ## cProfile, exemple
Et on repasse dans cProfile ! Et on repasse dans cProfile !
```text ```shell
$ python -m cProfile --sort cumulative phi2.py 10 $ python -m cProfile --sort cumulative phi2.py 10
#!cache -- python -m cProfile --sort cumulative include/phi2.py 10 | sed -n '/fib\|function calls/{s/ \+/ /g;s/^ *//;p}' ...
#!cache python -m cProfile --sort cumulative examples/phi2.py 10 | sed -n '/fib\|function calls/{s/ \+/ /g;s/^ *//;p}'
...
``` ```
C'est mieux ! C'est mieux !
@ -309,9 +334,8 @@ C'est mieux !
On essaye d'aller plus loin ? On essaye d'aller plus loin ?
```text ```shell
$ python -m cProfile --sort cumulative phi2.py 2000 #!cache python -m cProfile --sort cumulative examples/phi2.py 2000 | head -n 3 | sed 's/^ *//g;s/seconds/s/g'
#!cache -- python -m cProfile --sort cumulative include/phi2.py 2000 | head -n 2 | sed 's/^ *//g;s/seconds/s/g'
``` ```
Ça tient, mais peut-on faire mieux ? Ça tient, mais peut-on faire mieux ?
@ -323,22 +347,22 @@ Divisons par 10 le nombre d'appels, on réduira mécaniquement par 10 le
temps d'exécution ? temps d'exécution ?
```python ```python
#!sed -n '/def approx_phi_up_to/,/return step1/p' include/phi3.py #!sed -n '/def approx_phi_up_to/,/return step1/p' examples/phi3.py
``` ```
## cProfile, exemple ## cProfile, exemple
```text ```shell
$ python -m cProfile --sort cumulative phi3.py 2000 #!cache python -m cProfile --sort cumulative examples/phi3.py 2000 | head -n 3 | sed 's/^ *//g;s/seconds/s/g'
#!cache -- python -m cProfile --sort cumulative include/phi3.py 2000 | head -n 2 | sed 's/^ *//g;s/seconds/s/g'
``` ```
## cProfile, exemple ## cProfile, exemple
En cachant `approx_phi` ? En cachant `approx_phi` ?
```python ```python
#!sed -n '10,/return step1/p' include/phi4.py #!sed -n '10,/return step1/p' examples/phi4.py
``` ```
notes: notes:
@ -346,10 +370,11 @@ notes:
Notez l'astuce pour que le `step2` d'un Notez l'astuce pour que le `step2` d'un
tour soit le `step1` du suivant... tour soit le `step1` du suivant...
## cProfile, exemple ## cProfile, exemple
``` ```shell
$ python -m cProfile --sort cumulative phi4.py 2000 $ python -m cProfile --sort cumulative examples/phi4.py 2000
``` ```
`RecursionError` !? En effet, en avançant par si grands pas, le cache `RecursionError` !? En effet, en avançant par si grands pas, le cache
@ -363,15 +388,14 @@ Il est temps de sortir une implémentation de `fib` plus robuste, basée
sur l'algorithme « matrix exponentiation » : sur l'algorithme « matrix exponentiation » :
```python ```python
#!sed -n '/def fib/,/return fib/p' include/phi5.py #!sed -n '/def fib/,/return fib/p' examples/phi5.py
``` ```
## cProfile, exemple ## cProfile, exemple
```text ```text
$ python -m cProfile --sort cumulative phi5.py 2000 #!cache python -m cProfile --sort cumulative examples/phi5.py 2000 | head -n 3 | sed 's/^ *//g;s/seconds/s/g'
#!cache -- python -m cProfile --sort cumulative include/phi5.py 2000 | head -n 2 | sed 's/^ *//g;s/seconds/s/g'
``` ```
notes: notes:
@ -381,14 +405,15 @@ Mieux.
## Snakeviz ## Snakeviz
```text ```shell
python -m pip install snakeviz $ python -m pip install snakeviz
python -m cProfile -o phi5.prof phi5.py 2000 $ python -m cProfile -o phi5.prof phi5.py 2000
#!if [ ! -f .cache/phi5.prof ]; then python -m cProfile -o .cache/phi5.prof include/phi5.py 2000 >/dev/null 2>&1; fi $ python -m snakeviz phi5.prof
python -m snakeviz phi5.prof
#!if [ ! -f output/phi5-snakeviz.png ]; then python -m snakeviz -s .cache/phi5.prof & TOKILL=$!; sleep 1; cutycapt --min-width=1024 --delay=500 --url=http://127.0.0.1:8080/snakeviz/%2Ftmp%2Fphi5.prof --out=output/phi5-snakeviz.png ; kill $TOKILL; fi
``` ```
#!if [ ! -f .cache/phi5.prof ]; then python -m cProfile -o .cache/phi5.prof examples/phi5.py 2000 >/dev/null 2>&1; fi
#!if [ ! -f output/phi5-snakeviz.png ]; then python -m snakeviz -s .cache/phi5.prof & TOKILL=$!; sleep 1; cutycapt --min-width=1024 --delay=500 --url=http://127.0.0.1:8080/snakeviz/%2Ftmp%2Fphi5.prof --out=output/phi5-snakeviz.png ; kill $TOKILL; fi
## Snakeviz ## Snakeviz
@ -400,56 +425,64 @@ python -m snakeviz phi5.prof
```shell ```shell
$ python -m pip install scalene $ python -m pip install scalene
$ scalene phi5.py 100000 $ scalene phi5.py 100000
#!if [ ! -f output/phi5.html ]; then ( cd include; scalene phi5.py 100000 --html --outfile ../output/phi5.html --cli >&2 ); fi
#!if [ ! -f output/phi5-scalene.png ]; then cutycapt --min-width=1024 --delay=100 --url=file://$(pwd)/output/phi5.html --out=output/phi5-scalene.png; fi
``` ```
#!if [ ! -f output/phi5.html ]; then ( cd examples; scalene phi5.py 100000 --html --outfile ../output/phi5.html --cli >&2 ); fi
#!if [ ! -f output/phi5-scalene.png ]; then cutycapt --min-width=1024 --delay=100 --url=file://$(pwd)/output/phi5.html --out=output/phi5-scalene.png; fi
## Scalene ## Scalene
![](phi5-scalene.png) ![](phi5-scalene.png)
## Atelier ## Aussi
Générateur de prénoms français. - https://github.com/gaogaotiantian/viztracer
- https://github.com/joerick/pyinstrument
Notes: voir includes/prenom-*.py - https://github.com/benfred/py-spy
- https://github.com/pyutils/line_profiler
- https://github.com/sumerc/yappi
- https://github.com/vmprof/vmprof-python
- https://github.com/bloomberg/memray
## TODO ## Profilage — À vous !
- vprof
- https://pypi.org/project/pp3yflame/ Profilez votre implémentation et tentez quelques améliorations.
# Cython # Cython
Cython est un dialecte de Python transpilable en C. Cython est un dialecte de Python transpilable en C.
## Cython démo ## Cython démo
```python ```python
#!cat include/collatz_length.py #!cat examples/collatz_length.py
``` ```
## Cython démo ## Cython démo
```shell ```shell
$ python -m pyperf timeit \ $ python -m pyperf timeit \
> -s 'from collatz_length import collatz_length' > -s 'from collatz_length import collatz_length'
> 'collatz_length(837799)' > 'collatz_length(837799)'
#!cache -- python -m pyperf timeit --fast --setup 'from include.collatz_length import collatz_length' 'collatz_length(837799)' #!cache python -m pyperf timeit --fast --setup 'from examples.collatz_length import collatz_length' 'collatz_length(837799)'
``` ```
```shell ```shell
$ cythonize --inplace collatz_length_cython.py $ cythonize --inplace collatz_length_cython.py
#!if ! [ -f include/collatz_length_cython.*.so ] ; then cythonize --inplace include/collatz_length_cython.py; fi #!if ! [ -f examples/collatz_length_cython.*.so ] ; then cythonize --inplace examples/collatz_length_cython.py; fi
``` ```
```shell ```shell
$ python -m pyperf timeit \ $ python -m pyperf timeit \
> -s 'from collatz_length import collatz_length' > -s 'from collatz_length import collatz_length'
> 'collatz_length(837799)' > 'collatz_length(837799)'
#!cache -- python -m pyperf timeit --fast -s 'from include.collatz_length import collatz_length' 'collatz_length(837799)' # faster #!cache python -m pyperf timeit --fast -s 'from examples.collatz_length import collatz_length' 'collatz_length(837799)' # faster
#!# Beware, the cythonized use `-s` while the non cythonized uses `--setup` just to have two cache buckets :D #!# Beware, the cythonized use `-s` while the non cythonized uses `--setup` just to have two cache buckets :D
``` ```
@ -458,8 +491,8 @@ $ python -m pyperf timeit \
```shell ```shell
$ cython -a collatz_length.py $ cython -a collatz_length.py
#!if ! [ -f include/collatz_length.html ] ; then cython -a include/collatz_length.py; fi #!if ! [ -f examples/collatz_length.html ] ; then cython -a examples/collatz_length.py; fi
#!if ! [ -f output/collatz_length.png ] ; then cutycapt --min-width=1024 --delay=500 --url=file://$(pwd)/include/collatz_length.html --out=output/collatz_length.png; fi #!if ! [ -f output/collatz_length.png ] ; then cutycapt --min-width=1024 --delay=500 --url=file://$(pwd)/examples/collatz_length.html --out=output/collatz_length.png; fi
``` ```
![](collatz_length.png) ![](collatz_length.png)
@ -468,96 +501,106 @@ $ cython -a collatz_length.py
## Cython annotated ## Cython annotated
```python ```python
#!cat include/collatz_length_annotated.py #!cat examples/collatz_length_annotated.py
``` ```
```shell ```shell
$ cythonize --inplace collatz_length_annotated.py $ cythonize --inplace collatz_length_annotated.py
``` ```
#!if ! [ -f include/collatz_length_annotated.*.so ] ; then cythonize --inplace include/collatz_length_annotated.py; fi #!if ! [ -f examples/collatz_length_annotated.*.so ] ; then cythonize --inplace examples/collatz_length_annotated.py; fi
```shell ```shell
$ python -m pyperf timeit \ $ python -m pyperf timeit \
> -s 'from collatz_length_annotated import collatz_length' > -s 'from collatz_length_annotated import collatz_length'
> 'collatz_length(837799)' > 'collatz_length(837799)'
#!cache -- python -m pyperf timeit --fast -s 'from include.collatz_length_annotated import collatz_length' 'collatz_length(837799)' #!cache python -m pyperf timeit --fast -s 'from examples.collatz_length_annotated import collatz_length' 'collatz_length(837799)'
``` ```
## Cython annotate again ## Cython annotate again
```shell ```shell
$ cython -a include/collatz_length_annotated.py $ cython -a examples/collatz_length_annotated.py
#!if ! [ -f include/collatz_length_annotated.html ] ; then cython -a include/collatz_length_annotated.py; fi #!if ! [ -f examples/collatz_length_annotated.html ] ; then cython -a examples/collatz_length_annotated.py; fi
#!if ! [ -f output/collatz_length_annotated.png ] ; then cutycapt --min-width=1024 --delay=500 --url=file://$(pwd)/include/collatz_length_annotated.html --out=output/collatz_length_annotated.png; fi #!if ! [ -f output/collatz_length_annotated.png ] ; then cutycapt --min-width=1024 --delay=500 --url=file://$(pwd)/examples/collatz_length_annotated.html --out=output/collatz_length_annotated.png; fi
``` ```
![](collatz_length_annotated.png) ![](collatz_length_annotated.png)
## Cython — À vous !
# Numba # Numba
Numba est un `JIT` : « Just In Time compiler ». Numba est un `JIT` : « Just In Time compiler ».
```python ```python
#!cat include/collatz_length_numba.py #!cat examples/collatz_length_numba.py
``` ```
## Numba démo ## Numba démo
```shell ```shell
$ python -m pyperf timeit \ $ python -m pyperf timeit \
> -s 'from include.collatz_length_numba import collatz_length' \ > -s 'from examples.collatz_length_numba import collatz_length' \
> 'collatz_length(837799)' > 'collatz_length(837799)'
#!cache -- python -m pyperf timeit --fast -s 'from include.collatz_length_numba import collatz_length' 'collatz_length(837799)' #!cache python -m pyperf timeit --fast -s 'from examples.collatz_length_numba import collatz_length' 'collatz_length(837799)'
``` ```
## numba — À vous !
# mypyc # mypyc
mypyc est un compilateur qui s'appuie sur les annotationes de type mypy : mypyc est un compilateur qui s'appuie sur les annotationes de type mypy :
```python ```python
#!cat include/collatz_length_mypy.py #!cat examples/collatz_length_mypy.py
``` ```
## mypyc demo ## mypyc demo
```shell ```shell
$ mypyc include/collatz_length_mypy.py $ mypyc examples/collatz_length_mypy.py
#!if ! [ -f collatz_length_mypy.*.so ] ; then mypyc include/collatz_length_mypy.py; fi #!if ! [ -f collatz_length_mypy.*.so ] ; then mypyc examples/collatz_length_mypy.py; fi
``` ```
```shell ```shell
$ python -m pyperf timeit \ $ python -m pyperf timeit \
> -s 'from collatz_length_mypy import collatz_length' \ > -s 'from collatz_length_mypy import collatz_length' \
> 'collatz_length(837799)' > 'collatz_length(837799)'
#!cache -- python -m pyperf timeit --fast -s 'from collatz_length_mypy import collatz_length' 'collatz_length(837799)' #!cache python -m pyperf timeit --fast -s 'from collatz_length_mypy import collatz_length' 'collatz_length(837799)'
``` ```
## mypyc — À vous !
# Pythran # Pythran
pythran est un compilateur pour du code scientifique : pythran est un compilateur pour du code scientifique :
```python ```python
#!cat include/collatz_length_pythran.py #!cat examples/collatz_length_pythran.py
``` ```
## Pythran demo ## Pythran demo
```shell ```shell
$ pythran include/collatz_length_pythran.py $ pythran examples/collatz_length_pythran.py
#!if ! [ -f collatz_length_pythran.*.so ]; then pythran include/collatz_length_pythran.py; fi #!if ! [ -f collatz_length_pythran.*.so ]; then pythran examples/collatz_length_pythran.py; fi
``` ```
```shell ```shell
$ python -m pyperf timeit \ $ python -m pyperf timeit \
> -s 'from collatz_length_pythran import collatz_length' \ > -s 'from collatz_length_pythran import collatz_length' \
> 'collatz_length(837799)' > 'collatz_length(837799)'
#!cache -- python -m pyperf timeit --fast -s 'from collatz_length_pythran import collatz_length' 'collatz_length(837799)' #!cache python -m pyperf timeit --fast -s 'from collatz_length_pythran import collatz_length' 'collatz_length(837799)'
``` ```
## pythran — À vous !
# Nuitka # Nuitka
@ -565,44 +608,46 @@ Aussi un compilateur, aussi utilisable pour distribuer une application.
```shell ```shell
$ python -m nuitka --module collatz_length_nuitka.py $ python -m nuitka --module collatz_length_nuitka.py
#!if ! [ -f collatz_length_nuitka.*.so ]; then python -m nuitka --module include/collatz_length_nuitka.py >/dev/null; fi #!if ! [ -f collatz_length_nuitka.*.so ]; then python -m nuitka --module examples/collatz_length_nuitka.py >/dev/null; fi
``` ```
```shell ```shell
$ python -m pyperf timeit \ $ python -m pyperf timeit \
> -s 'from collatz_length_nuitka import collatz_length' \ > -s 'from collatz_length_nuitka import collatz_length' \
> 'collatz_length(837799)' > 'collatz_length(837799)'
#!cache -- python -m pyperf timeit --fast -s 'from collatz_length_nuitka import collatz_length' 'collatz_length(837799)' #!cache python -m pyperf timeit --fast -s 'from collatz_length_nuitka import collatz_length' 'collatz_length(837799)'
``` ```
# Et d'autres # Et d'autres…
https://github.com/pfalcon/awesome-python-compilers https://github.com/pfalcon/awesome-python-compilers
# Hand crafted C # Hand crafted C
```c ```c
#!sed -n '/int collatz_length/,/^$/p' include/my_collatz_length.c #!sed -n '/int collatz_length/,/^$/p' examples/my_collatz_length.c
``` ```
Mais comment l'utiliser ? Mais comment l'utiliser ?
## Hand crafted C ## Hand crafted C
### Avec Cython ### Avec Cython
```cpython ```cpython
#!cat include/collatz_length_cython_to_c.pyx #!cat examples/collatz_length_cython_to_c.pyx
``` ```
```shell ```shell
$ cythonize -i include/collatz_length_cython_to_c.pyx $ cythonize -i examples/collatz_length_cython_to_c.pyx
#!if ! [ -f include/collatz_length_cython_to_c.*.so ] ; then cythonize -i include/collatz_length_cython_to_c.pyx; fi #!if ! [ -f examples/collatz_length_cython_to_c.*.so ] ; then cythonize -i examples/collatz_length_cython_to_c.pyx; fi
``` ```
```shell ```shell
$ python -m pyperf timeit \ $ python -m pyperf timeit \
> -s 'from collatz_length_cython_to_c import collatz_length' \ > -s 'from collatz_length_cython_to_c import collatz_length' \
> 'collatz_length(837799)' > 'collatz_length(837799)'
#!cache -- python -m pyperf timeit --fast -s 'from include.collatz_length_cython_to_c import collatz_length' 'collatz_length(837799)' #!cache python -m pyperf timeit --fast -s 'from examples.collatz_length_cython_to_c import collatz_length' 'collatz_length(837799)'
``` ```