Experimental memory allocator with vendored mimalloc by tiran · Pull Request #29123 · python/cpython

tiran · 2021-10-21T14:28:02Z

EXPERIMENTAL This is an experimental pull request and a proof of concept. mimalloc integration is in a very early stage.

Vendor and hook up mimalloc. The PR only replaces raw memory allocator and Python's internal obmalloc with mimalloc. It does not use any of the advanced features of mimalloc such as heaps.

Includes mimalloc v1.7.1 and additional fixes
--with-mimalloc=yes by default (for testing)
--with-pymalloc=no by default
--with-freelists=yes

mimalloc requires C11 <stdatomic.h> on POSIX platforms to provide atomic operations such as CAS. GCC on RHEL 7 / CentOS 7 is too old. GCC 4.9 and clang 3.6 seem to be minimum versions. Users need install Red Hat Developer Toolset 10.0 from Software Collections. The CentOS 7-based manylinux2014 container image comes with GCC 10.2 pre-installed.

related mimalloc tickets

Initial benchmarks

Initial benchmarks should be taken with a big grain of salt. It seems like mimalloc is in the same ballpark as libc malloc + obmalloc with slightly better performance in some cases. I haven't measured memory consumption yet.

with mimalloc, without pymalloc, without freelists

In general about same performance, with some benchmarks 7% faster to 12% slower, https://gist.github.com/pablogsal/33fce1cbe5f908c1d85cae5d678f50a6

with mimalloc, without pymalloc, with freelists

In general 2% speedup, with some benchmarks 9% faster to 6% slower, https://gist.github.com/pablogsal/8027937b71cd30f17aaaa5ef7c885d3e

debug output

>>> sys._debugmallocstats()
mimalloc (version: 171)
heap stats:     peak      total      freed    current       unit      count  
  reserved:   256.0 mb   256.0 mb       0 b    256.0 mb       1 b              not all freed!
 committed:    20.0 mb    20.0 mb       0 b     20.0 mb       1 b              not all freed!
     reset:     1.0 mb     1.0 mb       0 b      1.0 mb       1 b              not all freed!
   touched:       0 b        0 b    932.3 kb  -932.3 kb       1 b              ok
  segments:       5          5          1          4                           not all freed!
-abandoned:       0          0          0          0                           ok
   -cached:       0          0          0          0                           ok
     pages:      94        123         29         94                           not all freed!
-abandoned:       0          0          0          0                           ok
 -extended:       0   
 -noretire:       0   
     mmaps:       0   
   commits:      69   
   threads:       0          0          0          0                           ok
  searches:     0.0 avg
numa nodes:       1
   elapsed:       5.572 s
   process: user: 0.073 s, system: 0.007 s, faults: 0, rss: 11.9 mb, commit: 20.0 mb
...

bedevere-bot · 2021-10-24T21:44:11Z

🤖 New build scheduled with the buildbot fleet by @tiran for commit a4e71652eaacf491eb682d99fdbb0a521b8cb33f 🤖

If you want to schedule another build, you need to add the ":hammer: test-with-buildbots" label again.

bedevere-bot · 2021-10-25T08:11:27Z

🤖 New build scheduled with the buildbot fleet by @tiran for commit b6f5965e657df821b2345433163293d5345d5bab 🤖

If you want to schedule another build, you need to add the ":hammer: test-with-buildbots" label again.

bedevere-bot · 2021-10-25T09:02:15Z

🤖 New build scheduled with the buildbot fleet by @tiran for commit 11efb79362b5f49e441d3a24da01f3740a2222d5 🤖

If you want to schedule another build, you need to add the ":hammer: test-with-buildbots" label again.

bedevere-bot · 2021-10-25T10:28:28Z

🤖 New build scheduled with the buildbot fleet by @tiran for commit 8d7373a0c3f1f47b1320087f8deafc7bd7c9ad4a 🤖

If you want to schedule another build, you need to add the ":hammer: test-with-buildbots" label again.

insilications · 2021-10-25T20:36:47Z

Saw your PR in microsoft/mimalloc#475. This is great. I'm using mimalloc as a drop-in replacement with the ENV variables:

PYTHONMALLOC=malloc MIMALLOC_PAGE_RESET=0 MIMALLOC_LARGE_OS_PAGES=1 LD_PRELOAD='/usr/lib64/libmimalloc.so'

I expected that this upstream integration would bring other benefits, but it would not be able to use mimalloc's allocation of large OS pages?

bedevere-bot · 2021-10-26T16:13:46Z

🤖 New build scheduled with the buildbot fleet by @tiran for commit 8d7373a0c3f1f47b1320087f8deafc7bd7c9ad4a 🤖

If you want to schedule another build, you need to add the ":hammer: test-with-buildbots" label again.

ifeq is not portable. Original patch by Neil. Co-authored-by: Neil Schemenauer <nas@arctrix.com> Signed-off-by: Christian Heimes <christian@python.org>

See: microsoft/mimalloc#426 Signed-off-by: Christian Heimes <christian@python.org>

Use mimalloc's ``static.c`` to include mimalloc library into obmalloc. This allows us to prefix all symbols with ``_Py_`` prefix. Signed-off-by: Christian Heimes <christian@python.org>

tiran · 2022-04-06T12:43:18Z

This PR has been replaced by #31164.

tiran added type-feature A feature request or enhancement sprint DO-NOT-MERGE labels Oct 21, 2021

bedevere-bot added the awaiting core review label Oct 21, 2021

the-knights-who-say-ni added the CLA signed label Oct 21, 2021

tiran force-pushed the mimalloc branch 3 times, most recently from 3cadf5e to f865363 Compare October 21, 2021 17:35

tiran mentioned this pull request Oct 21, 2021

Add option to set mimalloc symbols to visibility=hidden microsoft/mimalloc#475

Closed

tiran force-pushed the mimalloc branch 3 times, most recently from e1cecad to a4e7165 Compare October 24, 2021 16:46

tiran added the 🔨 test-with-buildbots Test PR w/ buildbots; report in status section label Oct 24, 2021

bedevere-bot removed the 🔨 test-with-buildbots Test PR w/ buildbots; report in status section label Oct 24, 2021

tiran added the 🔨 test-with-buildbots Test PR w/ buildbots; report in status section label Oct 25, 2021

bedevere-bot removed the 🔨 test-with-buildbots Test PR w/ buildbots; report in status section label Oct 25, 2021

tiran added the 🔨 test-with-buildbots Test PR w/ buildbots; report in status section label Oct 25, 2021

bedevere-bot removed the 🔨 test-with-buildbots Test PR w/ buildbots; report in status section label Oct 25, 2021

tiran force-pushed the mimalloc branch 2 times, most recently from d3a1605 to 8d7373a Compare October 25, 2021 10:26

tiran added the 🔨 test-with-buildbots Test PR w/ buildbots; report in status section label Oct 25, 2021

bedevere-bot removed the 🔨 test-with-buildbots Test PR w/ buildbots; report in status section label Oct 25, 2021

tiran added the 🔨 test-with-buildbots Test PR w/ buildbots; report in status section label Oct 26, 2021

bedevere-bot removed the 🔨 test-with-buildbots Test PR w/ buildbots; report in status section label Oct 26, 2021

tiran and others added 26 commits January 13, 2022 11:39

Try mimalloc on Windows

aca86ea

Replace ifeq with AC_SUBST

f363f41

ifeq is not portable. Original patch by Neil. Co-authored-by: Neil Schemenauer <nas@arctrix.com> Signed-off-by: Christian Heimes <christian@python.org>

Only include Include/mimalloc for some files

b3ec03c

Hide symbols when vendoring mimalloc

8a51036

Prefix _os_random_weak with _mi_

9debbe6

Ignore mimalloc symbols in archive file

f9715a2

Skip test_pyobject_freed_is_freed under mimalloc

602d5fa

Attempt to fix mimalloc build on Windows ARM64

c3fde01

See: microsoft/mimalloc#426 Signed-off-by: Christian Heimes <christian@python.org>

Use different approach to hide symbols

4838dbc

Inline entire mimalloc code in obmalloc

30df26c

Use mimalloc's ``static.c`` to include mimalloc library into obmalloc. This allows us to prefix all symbols with ``_Py_`` prefix. Signed-off-by: Christian Heimes <christian@python.org>

Use fputs instead of fprintf

c2e3446

Add sys._malloc_info

b1cec21

mimalloc requires stdatomic.h on POSIX

8fd2df4

Skip test_decimal test_maxcontext_exact_arith on s390x and mimalloc

c5bf41b

Update to mimalloc 2.0.3

37625cf

Use proper mem API for mimalloc

463c87d

Use Python's PYMEM debug values

fa07d9d

Experimental Valgrind integration

4a6baa8

Support pymalloc and mimalloc in same build

9365d32

Fix tests

38e0e47

Add documentation

97a7dc5

malloc_info.with_

7f3955b

Fix docs

9fb2bc7

Enable pymalloc on Windows again

362f1c6

Cleanup

b5bbe3b

Fix Emscripten build and runtime issues

0632796

tiran force-pushed the mimalloc branch from 71df5b4 to 0632796 Compare January 13, 2022 10:40

Merge remote-tracking branch 'upstream/main' into mimalloc

e57472a

tiran closed this Apr 6, 2022

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Experimental memory allocator with vendored mimalloc#29123

Experimental memory allocator with vendored mimalloc#29123
tiran wants to merge 33 commits intopython:mainfrom
tiran:mimalloc

tiran commented Oct 21, 2021 •

edited

Loading

Uh oh!

bedevere-bot commented Oct 24, 2021

Uh oh!

bedevere-bot commented Oct 25, 2021

Uh oh!

bedevere-bot commented Oct 25, 2021

Uh oh!

bedevere-bot commented Oct 25, 2021

Uh oh!

insilications commented Oct 25, 2021

Uh oh!

bedevere-bot commented Oct 26, 2021

Uh oh!

tiran commented Apr 6, 2022

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Uh oh!

Conversation

tiran commented Oct 21, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

related mimalloc tickets

Initial benchmarks

with mimalloc, without pymalloc, without freelists

with mimalloc, without pymalloc, with freelists

debug output

Uh oh!

bedevere-bot commented Oct 24, 2021

Uh oh!

bedevere-bot commented Oct 25, 2021

Uh oh!

bedevere-bot commented Oct 25, 2021

Uh oh!

bedevere-bot commented Oct 25, 2021

Uh oh!

insilications commented Oct 25, 2021

Uh oh!

bedevere-bot commented Oct 26, 2021

Uh oh!

tiran commented Apr 6, 2022

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

tiran commented Oct 21, 2021 •

edited

Loading