[issue26110] Speedup method calls 1.2x

[Ken Jin]

Change by Ken Jin :


--
pull_requests: +24793
pull_request: https://github.com/python/cpython/pull/26159

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Ken Jin]

Change by Ken Jin :


--
nosy: +kj
nosy_count: 21.0 -> 22.0
pull_requests: +24665
pull_request: https://github.com/python/cpython/pull/26014

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Jeroen Demeyer]

Jeroen Demeyer  added the comment:

Note that this idea has been generalized by PEP 590: any type can support this 
optimization by setting the Py_TPFLAGS_METHOD_DESCRIPTOR flag.

--
nosy: +jdemeyer

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Jesús Cea Avión]

Changes by Jesús Cea Avión :


--
nosy: +jcea

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Donald Stufft]

Changes by Donald Stufft :


--
pull_requests: +1065

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[INADA Naoki]

Changes by INADA Naoki :


--
status: open -> closed

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Roundup Robot]

Roundup Robot added the comment:

New changeset a6241b2073c6 by INADA Naoki in branch 'default':
Issue #26110: Add document for LOAD_METHOD and CALL_METHOD opcode.
https://hg.python.org/cpython/rev/a6241b2073c6

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[INADA Naoki]

Changes by INADA Naoki :


Added file: http://bugs.python.org/file46294/call-method-doc5.patch

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[INADA Naoki]

INADA Naoki added the comment:

Yury, could you review this?

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[STINNER Victor]

STINNER Victor added the comment:

I just created the issue #29263: "Implement LOAD_METHOD/CALL_METHOD for C 
functions".

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[INADA Naoki]

Changes by INADA Naoki :


Added file: http://bugs.python.org/file45963/call-method-doc4.patch

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[INADA Naoki]

Changes by INADA Naoki :


Added file: http://bugs.python.org/file45896/callmethod-doc3.patch

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[INADA Naoki]

Changes by INADA Naoki :


Added file: http://bugs.python.org/file45891/callmethod-doc2.patch

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[INADA Naoki]

Changes by INADA Naoki :


Removed file: http://bugs.python.org/file45890/callmethod-doc2.patch

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[INADA Naoki]

Changes by INADA Naoki :


Added file: http://bugs.python.org/file45890/callmethod-doc2.patch

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[STINNER Victor]

STINNER Victor added the comment:

INADA Naoki: "My current idea is adding new `tp_fastcall` slot which has same 
signature to _PyFunction_FastCallDict or _PyCFunction_FastCallDict."

FYI I'm working on a solution to avoid tuple and dict to pass parameters to 
tp_new, tp_init and tp_call. I have PoC implementations (yeah, more than one). 
Please come to me directly if you are interested, this issue is not the right 
place to discuss.

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[INADA Naoki]

INADA Naoki added the comment:

This patch modify stack layout slightly and adds document in Doc/library/dis.rst

--
Added file: http://bugs.python.org/file45889/callmethod-doc.patch

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[INADA Naoki]

INADA Naoki added the comment:

PGO benchmark result
https://gist.github.com/methane/64ab60c38324423ef27af70cc6db1dfd

--
Added file: http://bugs.python.org/file45887/default.json.gz

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[INADA Naoki]

Changes by INADA Naoki :


Added file: http://bugs.python.org/file45888/patched.json.gz

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[INADA Naoki]

INADA Naoki added the comment:

I haven't noticed the patch is committed already.

Changing stack layout slightly is for easy to document, not for performance.
Please don't close this issue until adding document in Doc/library/dis.rst

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[INADA Naoki]

INADA Naoki added the comment:

I'm working on changing stack layout slightly

current patch: callable | NULL | arg1 | ...argN
next patch will: NULL | callable | arg1 | ...argN

After benchmark with PGO build, I'll post it.

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Yury Selivanov]

Yury Selivanov added the comment:

> Inada-san, when I tested the patch last time, I think there was a regression 
> somewhere, related to the descriptor protocol.  Have you fixed that one?

Seems to be either fixed, or maybe those bugs were related to my opcode cache 
patch.  Anyways, I decided to commit the patch to 3.7, otherwise it might miss 
the commit window as it did for 3.6.  Let's fix any regressions right in the 
repo now.

> My current idea is adding new `tp_fastcall` slot which has same signature
to _PyFunction_FastCallDict or _PyCFunction_FastCallDict.

I like this idea!

Thanks Inada-san for pushing this patch through, and thanks to Serhiy for 
reviewing it.

--
resolution:  -> fixed
stage: patch review -> commit review

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Roundup Robot]

Roundup Robot added the comment:

New changeset 64afd5cab40a by Yury Selivanov in branch 'default':
Issue #26110: Add LOAD_METHOD/CALL_METHOD opcodes.
https://hg.python.org/cpython/rev/64afd5cab40a

--
nosy: +python-dev

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Yury Selivanov]

Yury Selivanov added the comment:

Inada-san, when I tested the patch last time, I think there was a regression 
somewhere, related to the descriptor protocol.  Have you fixed that one?

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Guido van Rossum]

Changes by Guido van Rossum :


--
nosy:  -gvanrossum

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Serhiy Storchaka]

Serhiy Storchaka added the comment:

> But I think it should be separated issue.

Agreed if that so hard.

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[INADA Naoki]

INADA Naoki added the comment:

> And would be nice to extend the optimization to C functions.

I tried it but skipping creating PyCFunction seems impossible for now.

My current idea is adding new `tp_fastcall` slot which has same signature
to _PyFunction_FastCallDict or _PyCFunction_FastCallDict.

If MethodDescrObject implement it, we can skip temporary PyCFunction object and
argument tuple.

But I think it should be separated issue. Patch is large enough already.

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[INADA Naoki]

INADA Naoki added the comment:

> Technically the patch LGTM. But we should find the cause of the regression in 
> some benchmarks.

The benchmark is on Sandy Bridge (Core i5 2400) and I didn't use PGO build.
perf_event reported branch-miss rate increase at cpickle's save function.

I'll rerun benchmark with PGO build. I hope PGO is friendly with CPU branch
prediction, like L1/L2 cache.

Anyway, recent amd64 CPUs have more large branch history.

> And would be nice to extend the optimization to C functions.
> In any case this optimization is worth mentioning in What's New.

I'll do them.

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Serhiy Storchaka]

Serhiy Storchaka added the comment:

Technically the patch LGTM. But we should find the cause of the regression in 
some benchmarks.

And would be nice to extend the optimization to C functions.

In any case this optimization is worth mentioning in What's New.

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[INADA Naoki]

Changes by INADA Naoki :


Added file: http://bugs.python.org/file45866/call_method_5.patch

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Yury Selivanov]

Yury Selivanov added the comment:

Please don't merge this without my review.

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[INADA Naoki]

INADA Naoki added the comment:

$ ./python-default -m perf compare_to default.json callmethod4.json -G
Slower (7):
- pickle_dict: 66.0 us +- 4.6 us -> 77.0 us +- 5.9 us: 1.17x slower
- json_loads: 63.7 us +- 0.7 us -> 68.4 us +- 1.4 us: 1.07x slower
- unpack_sequence: 120 ns +- 2 ns -> 125 ns +- 3 ns: 1.04x slower
- scimark_lu: 499 ms +- 12 ms -> 514 ms +- 24 ms: 1.03x slower
- scimark_monte_carlo: 272 ms +- 10 ms -> 278 ms +- 9 ms: 1.02x slower
- scimark_sor: 517 ms +- 9 ms -> 526 ms +- 10 ms: 1.02x slower
- regex_effbot: 5.25 ms +- 0.15 ms -> 5.27 ms +- 0.17 ms: 1.00x slower

Faster (52):
- call_method: 18.1 ms +- 0.1 ms -> 14.8 ms +- 0.1 ms: 1.22x faster
- call_method_slots: 17.9 ms +- 0.2 ms -> 14.7 ms +- 0.2 ms: 1.22x faster
- call_method_unknown: 20.0 ms +- 0.9 ms -> 16.5 ms +- 0.1 ms: 1.21x faster
- xml_etree_parse: 304 ms +- 13 ms -> 283 ms +- 6 ms: 1.07x faster
- float: 310 ms +- 4 ms -> 292 ms +- 4 ms: 1.06x faster
- call_simple: 14.2 ms +- 0.6 ms -> 13.4 ms +- 0.3 ms: 1.06x faster
- pickle_pure_python: 1.34 ms +- 0.02 ms -> 1.27 ms +- 0.02 ms: 1.05x faster
- richards: 189 ms +- 3 ms -> 180 ms +- 7 ms: 1.05x faster
- logging_silent: 793 ns +- 14 ns -> 754 ns +- 9 ns: 1.05x faster
- deltablue: 18.4 ms +- 0.2 ms -> 17.6 ms +- 0.1 ms: 1.05x faster
- go: 623 ms +- 5 ms -> 594 ms +- 5 ms: 1.05x faster
- genshi_xml: 202 ms +- 2 ms -> 193 ms +- 3 ms: 1.04x faster
- fannkuch: 1.10 sec +- 0.02 sec -> 1.05 sec +- 0.02 sec: 1.04x faster
- raytrace: 1.38 sec +- 0.02 sec -> 1.32 sec +- 0.00 sec: 1.04x faster
- django_template: 419 ms +- 5 ms -> 402 ms +- 4 ms: 1.04x faster
- hexiom: 24.2 ms +- 0.2 ms -> 23.3 ms +- 0.2 ms: 1.04x faster
- xml_etree_iterparse: 225 ms +- 4 ms -> 216 ms +- 5 ms: 1.04x faster
- spectral_norm: 282 ms +- 14 ms -> 271 ms +- 4 ms: 1.04x faster
- python_startup_no_site: 9.83 ms +- 0.02 ms -> 9.46 ms +- 0.03 ms: 1.04x faster
- python_startup: 16.4 ms +- 0.0 ms -> 15.7 ms +- 0.0 ms: 1.04x faster
- unpickle_pure_python: 918 us +- 24 us -> 884 us +- 24 us: 1.04x faster
- unpickle: 33.0 us +- 0.4 us -> 31.8 us +- 0.4 us: 1.04x faster
- dulwich_log: 158 ms +- 1 ms -> 153 ms +- 1 ms: 1.04x faster
- html5lib: 234 ms +- 6 ms -> 225 ms +- 6 ms: 1.04x faster
- scimark_fft: 735 ms +- 19 ms -> 709 ms +- 22 ms: 1.04x faster
- sqlalchemy_declarative: 325 ms +- 4 ms -> 314 ms +- 3 ms: 1.04x faster
- mako: 45.4 ms +- 0.3 ms -> 43.8 ms +- 0.7 ms: 1.04x faster
- genshi_text: 90.1 ms +- 0.9 ms -> 87.0 ms +- 0.9 ms: 1.04x faster
- tornado_http: 388 ms +- 4 ms -> 375 ms +- 4 ms: 1.03x faster
- 2to3: 758 ms +- 4 ms -> 734 ms +- 4 ms: 1.03x faster
- scimark_sparse_mat_mult: 8.75 ms +- 0.25 ms -> 8.48 ms +- 0.39 ms: 1.03x 
faster
- json_dumps: 29.9 ms +- 0.4 ms -> 29.0 ms +- 0.3 ms: 1.03x faster
- sqlite_synth: 10.1 us +- 0.2 us -> 9.80 us +- 0.19 us: 1.03x faster
- regex_dna: 286 ms +- 6 ms -> 278 ms +- 4 ms: 1.03x faster
- crypto_pyaes: 235 ms +- 2 ms -> 228 ms +- 2 ms: 1.03x faster
- pidigits: 320 ms +- 0 ms -> 311 ms +- 1 ms: 1.03x faster
- logging_format: 38.1 us +- 0.5 us -> 37.1 us +- 0.4 us: 1.03x faster
- meteor_contest: 206 ms +- 3 ms -> 201 ms +- 1 ms: 1.03x faster
- xml_etree_process: 248 ms +- 4 ms -> 241 ms +- 4 ms: 1.03x faster
- chameleon: 30.9 ms +- 0.4 ms -> 30.1 ms +- 0.4 ms: 1.03x faster
- logging_simple: 32.8 us +- 0.4 us -> 32.0 us +- 0.3 us: 1.03x faster
- sqlalchemy_imperative: 61.3 ms +- 1.6 ms -> 59.9 ms +- 1.6 ms: 1.02x faster
- chaos: 308 ms +- 2 ms -> 301 ms +- 2 ms: 1.02x faster
- unpickle_list: 7.90 us +- 0.04 us -> 7.73 us +- 0.10 us: 1.02x faster
- sympy_integrate: 47.9 ms +- 0.3 ms -> 46.9 ms +- 0.4 ms: 1.02x faster
- sympy_sum: 216 ms +- 6 ms -> 212 ms +- 6 ms: 1.02x faster
- sympy_str: 481 ms +- 5 ms -> 473 ms +- 2 ms: 1.02x faster
- sympy_expand: 1.08 sec +- 0.01 sec -> 1.06 sec +- 0.01 sec: 1.02x faster
- regex_compile: 449 ms +- 3 ms -> 442 ms +- 5 ms: 1.02x faster
- nqueens: 275 ms +- 3 ms -> 271 ms +- 3 ms: 1.02x faster
- pathlib: 51.6 ms +- 0.6 ms -> 50.9 ms +- 0.7 ms: 1.01x faster
- pickle_list: 8.85 us +- 0.09 us -> 8.79 us +- 0.11 us: 1.01x faster

Benchmark hidden because not significant (5): nbody, pickle, regex_v8, telco, 
xml_etree_generate

--
Added file: http://bugs.python.org/file45787/default.json.gz

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[INADA Naoki]

Changes by INADA Naoki :


Added file: http://bugs.python.org/file45788/callmethod4.json.gz

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Serhiy Storchaka]

Serhiy Storchaka added the comment:

Added comments on Rietveld. Please document new bytecodes in the dis module 
documentation and What's New.

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Serhiy Storchaka]

Serhiy Storchaka added the comment:

Please increase the magic number by 10. We need to reserve few numbers for the 
case of bytecode bug fixes in 3.6.

--
nosy: +serhiy.storchaka

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[INADA Naoki]

Changes by INADA Naoki :


--
versions: +Python 3.7 -Python 3.6
Added file: http://bugs.python.org/file45786/call_method_4.patch

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Ryan May]

Changes by Ryan May :


--
nosy: +Ryan May

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[INADA Naoki]

INADA Naoki added the comment:

Oops, previous patch doesn't update magic number in PC/launcher.c
Should I update it?  Or don't touch it to avoid additional conflicts?

--
Added file: http://bugs.python.org/file43601/call_method_3.patch

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[INADA Naoki]

INADA Naoki added the comment:

Updated, based on 102241:908b801f8a62

--
nosy: +naoki
Added file: http://bugs.python.org/file43599/call_method_2.patch

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Jakub Stasiak]

Changes by Jakub Stasiak :


--
nosy: +jstasiak

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Josh Rosenberg]

Changes by Josh Rosenberg :


--
nosy: +josh.r

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Guido van Rossum]

Guido van Rossum added the comment:

This patch doesn't apply cleanly any more. Is it easy to update?

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Leonardo Santagada]

Changes by Leonardo Santagada :


--
nosy: +santagada

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Maciej Szulik]

Changes by Maciej Szulik :


--
nosy: +maciej.szulik

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Florin Papa]

Changes by Florin Papa :


--
nosy: +florin.papa

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Emanuel Barry]

Changes by Emanuel Barry :


--
nosy: +ebarry

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Yury Selivanov]

Yury Selivanov added the comment:

For those interested in reviewing this patch at some point: please wait until I 
upload a new version.  The current patch is somewhat outdated.

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Alecsandru Patrascu]

Changes by Alecsandru Patrascu :


--
nosy: +alecsandru.patrascu

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Eric Fahlgren]

Changes by Eric Fahlgren :


--
nosy: +eric.fahlgren

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Alecsandru Patrascu]

Alecsandru Patrascu added the comment:

Yury, thank you for the heads up! Here at Intel, in the Dynamic Scripting 
Languages Optimization Team, we can help the community with reviewing and 
measuring this patch in our quiet and stable environment, the same one that we 
use to provide public CPython daily measurements. We will wait for your update.

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Yury Selivanov]

Yury Selivanov added the comment:

> If you prefer to limit the number of opcodes, you can pass a flag in 
> arguments. For example, use 2 bytes for 2 arguments instead of only 1?

I tried two approaches:

1. Push one more leading NULL to the stack in LOAD_METHOD (which makes it push 
3 vals).  Then, in CALL_FUNCTION, detect when the callable is NULL and to the 
stuff that CALL_METHOD does.

2. Use a bitflag on oparg for CALL_FUNCTION and CALL_* opcodes.  This adds 
EXTENDED_ARG opcode overhead.

Long story short, the only option would be to add dedicated opcodes to work 
with LOAD_METHOD.  However, I'd prefer to first merge this patch, as it's 
relatively small and easy to review, and then focus on improving other things 
(keyword/*arg calls, C methods, etc).  This is just a first step.

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[STINNER Victor]

STINNER Victor added the comment:

> For keyword and var-arg calls we have three more opcodes -- 
> CALL_FUNCTION_VAR, CALL_FUNCTION_KW, and CALL_FUNCTION_VAR_KW.  I suspect 
> that making them work with LOAD_METHOD would slow them down too, which will 
> probably require us to add three (!) more opcodes for LOAD_METHOD.

I don't think that it's an issue to add 3 more opcodes for performance. If you 
prefer to limit the number of opcodes, you can pass a flag in arguments. For 
example, use 2 bytes for 2 arguments instead of only 1?

See also the (now old) WPython project which proposed kind of CISC instructions:
https://code.google.com/p/wpython/

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Yury Selivanov]

New submission from Yury Selivanov:

This issue supersedes issue #6033.  I decided to open a new one, since the 
patch is about Python 3.6 (not 2.7) and is written from scratch.

The idea is to add new opcodes to avoid instantiation of BoundMethods.  The 
patch only affects method calls of Python functions with positional arguments.

I'm working on the attached patch in this repo: 
https://github.com/1st1/cpython/tree/call_meth2

If the patch gets accepted, I'll update it with the docs etc.


Performance Improvements


Method calls in micro-benchmarks are 1.2x faster:

### call_method ###
Avg: 0.330371 -> 0.281452: 1.17x faster

### call_method_slots ###
Avg: 0.333489 -> 0.280612: 1.19x faster

### call_method_unknown ###
Avg: 0.304949 -> 0.251623: 1.21x faster

Improvements in mini-benchmarks, such as Richards are less impressive, I'd say 
it's 3-7% improvement.  The full output of benchmarks/perf.py is here: 
https://gist.github.com/1st1/e00f11586329f68fd490

When the full benchmarks suite is run, some of them report that they were slow. 
 When I ran them separately several times, they all show no real slowdowns.  
It's just some of them (such as nbody) are highly unstable.

It's actually possible to improve the performance another 1-3% if we fuse 
__PyObject_GetMethod with ceval/LOAD_METHOD code.  I've tried this here: 
https://github.com/1st1/cpython/tree/call_meth4, however I don't like to have 
so many details of object.c into ceval.c.


Changes in the Core
---

Two new opcodes are added -- LOAD_METHOD and CALL_METHOD.  Whenever compiler 
sees a method call "obj.method(..)" with positional arguments it compiles it as 
follows:

  LOAD_FAST(obj)
  LOAD_METHOD(method)
  {call arguments}
  CALL_METHOD

LOAD_METHOD implementation in ceval looks up "method" on obj's type, and checks 
that it wasn't overridden in obj.__dict__.  Apparently, even with the __dict__ 
check this is still faster then creating a BoundMethod instance etc.  

If the method is found and not overridden, LOAD_METHOD pushes the resolved 
method object, and 'obj'.  If the method was overridden, the resolved method 
object and NULL are pushed to the stack.

CALL_METHOD then looks at the two stack values after call arguments.  If the 
first one isn't NULL, it means that we have a method call.


Why CALL_METHOD?


It's actually possible to hack CALL_FUNCTION to support LOAD_METHOD.  I've 
tried this approach in https://github.com/1st1/cpython/tree/call_meth3.  It 
looks like that adding extra checks in CALL_FUNCTION have negative impact on 
many benchmarks.  It's really easier to add another opcode.


Why only pure-Python methods?
-

LOAD_METHOD atm works only with methods defined in pure Python.  C methods, 
such as `list.append` are still wrapped into a descriptor, that creates a 
PyCFunction object on each attribute access.  I've tried to do that in 
https://github.com/1st1/cpython/tree/call_cmeth.  It does impact C method calls 
in a positive way, although my implementation is very hacky.  It still uses 
LOAD_METHOD and CALL_METHOD opcodes, so my idea is to consider merging this 
patch first, and then introduce the necessary refactoring of PyCFunction and 
MethodDesctiptors in a separate issue.


Why only calls with positional arguments?
-

As showed in "Why CALL_METHOD?", making CALL_FUNCTION to work with LOAD_METHOD 
slows it down.  For keyword and var-arg calls we have three more opcodes -- 
CALL_FUNCTION_VAR, CALL_FUNCTION_KW, and CALL_FUNCTION_VAR_KW.  I suspect that 
making them work with LOAD_METHOD would slow them down too, which will probably 
require us to add three (!) more opcodes for LOAD_METHOD.

And these kind of calls require much more overhead anyways, I don't expect them 
to be as optimizable as positional arg calls.

--
assignee: yselivanov
components: Interpreter Core
files: call_method_1.patch
keywords: patch
messages: 258204
nosy: benjamin.peterson, brett.cannon, gvanrossum, haypo, ncoghlan, yselivanov
priority: normal
severity: normal
stage: patch review
status: open
title: Speedup method calls 1.2x
type: performance
versions: Python 3.6
Added file: http://bugs.python.org/file41614/call_method_1.patch

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Guido van Rossum]

Guido van Rossum added the comment:

I like this idea! I like the limitations to positional-only calls. I do think 
that it would be nice if we could speed up C calls too -- today, 
s.startswith('abc') is slower than s[:3] == 'abc' precisely because of the 
lookup. But I'm all for doing this one step at a time, so we can be sure it is 
solid before taking the next step(s).

--

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Brett Cannon]

Changes by Brett Cannon :


--
nosy: +twouters

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com

[issue26110] Speedup method calls 1.2x

[Brett Cannon]

Changes by Brett Cannon :


--
nosy: +dino.viehland

___
Python tracker 

___
___
Python-bugs-list mailing list
Unsubscribe: 
https://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com