On 05/14/2015 11:18 AM, Armin Rigo wrote:
Hi Stanislav,
Then it looks likely to be a "randomness" issue: details in your
program are JIT-compiled differently when cProfile is enabled. For
example, cProfile makes traces longer and this can change a few
heuristics. It's likely that by tweaking your program a bit randomly
you can also get the same 20% variation.
Thank you for the answer. I have tried to benchmark some older versions
of my program before series of program core modifications, but results
have still the same characteristics, pypy is slower than CPython,
pypy+cProfile is faster than CPython. Hence I am afraid that a random
tweaking does not help me.
On 05/15/2015 08:05 AM, Armin Rigo wrote:
That's idle speculation at this point. I can't say more without
trying to look exactly at what occurs in this example, which I'm
unsure is a good way to spend my time --- but I'd be glad to if
Stanislav says he really wants to know in more details.
I am very curious to discover the root of the problem and I am ready for
a cooperation. At least, I would like to end with an argument as "--jit
MAGIC_VALUE=123" than with enabling cProfile for a better performance:)
But I completely understand that you do not have time to dig into my
program.
Best regards,
Stanislav Bohm
_______________________________________________
pypy-dev mailing list
pypy-dev@python.org
https://mail.python.org/mailman/listinfo/pypy-dev
