gh-100239: specialize bitwise logical binary ops on ints #128927

iritkatriel · 2025-01-16T22:55:36Z

This adds specialisations for bitwise |, &, ^ on non-negative ints.

I'm not adding more in the same PR so we can more easily bisect in the future if we need to.

Issue: Specialize long tail of binary operations using a table. #100239

…e int

eendebakpt · 2025-01-16T23:59:59Z

Python/specialize.c

+    return (is_nonnegative_compactlong(lhs) && is_nonnegative_compactlong(rhs));
+}
+
+#define NONNEGATIVE_LONGS_ACTION(NAME, OP) \


Why restrict to nonnegative longs here? If we do restrict, then we can replace the calls _PyLong_CompactValue with direct access to op->long_value.ob_digit[0]

Because negative ints need more work at runtime and I don't think they're common with bitwise logical ops.

The extra work is already done by _PyLong_CompactValue or am missing something? The output of ls OP rhs might not be a compact int, but there are no guards for the output type.

The extra work is already done by _PyLong_CompactValue or am missing something?

No, I think you're right. Good point.

The output of ls OP rhs might not be a compact int, but there are no guards for the output type.

For bitwise logical operators we should expect the results to have the same size as the inputs.

Misc/NEWS.d/next/Core_and_Builtins/2025-01-16-22-54-12.gh-issue-100239.7_HpBU.rst

…e-100239.7_HpBU.rst Co-authored-by: Pieter Eendebak <[email protected]>

iritkatriel · 2025-01-17T09:07:06Z

The stats are weird with this PR, lot of misses: https://github.com/faster-cpython/benchmarking-public/blob/main/results/bm-20250116-3.14.0a4%2B-c57fe46/bm-20250116-azure-x86_64-iritkatriel-binaryops-3.14.0a4%2B-c57fe46-pystats-vs-base.md

Python/specialize.c

chris-eibl

I think non-negative should be dropped in the news entry now?

Misc/NEWS.d/next/Core_and_Builtins/2025-01-16-22-54-12.gh-issue-100239.7_HpBU.rst

chris-eibl · 2025-01-19T09:11:49Z

Python/specialize.c

+    { \
+        Py_ssize_t rhs_val = _PyLong_CompactValue((PyLongObject *)rhs); \
+        Py_ssize_t lhs_val = _PyLong_CompactValue((PyLongObject *)lhs); \
+        return PyLong_FromLong(lhs_val OP rhs_val); \


Regarding the conversion warnings on Windows: they stem from long being 32bit on Windows x86_64, whereas Py_ssize_t is 64bit there. @markshannon forsightfully constructed _PyLong_CompactValue() to return 64bit, so that digits in the future can grow above 30 bits.

Maybe use PyLong_FromSsize_t() here?

P.S: The only disadvantage is, that PyLong_FromSsize_t() misses the fast path using _PyLong_FromMedium() like PyLong_FromLong() (and also PyLong_FromLongLong()) does.

Is this worth an issue? Maybe then implement PyLong_FromSsize_t(), PyLong_FromLong() and PyLong_FromLongLong()) using a macro similar to PYLONG_FROM_UINT to get rid of the repetitive code?

BTW: PYLONG_FROM_UINT is missing the fast path for medium values, too.

Thanks, I've updated. Feel free to create an issue about the fast path.

Done: #129149.
This is my first issue, hopefully I didn't mess up :)

chris-eibl · 2025-01-21T16:20:48Z

I think non-negative should be dropped in the news entry now?

And maybe in the title of this pull request, too?

markshannon

It looks like we are seeing a high proportion of specialization failures for non-compact ints with the & operator.
I don't know why that would be. My guess is that some of the benchmarks are using ints as bit vectors and using more than one digit.

markshannon · 2025-01-22T12:46:15Z

Python/specialize.c

@@ -2556,6 +2609,7 @@ binary_op_extended_specialization(PyObject *lhs, PyObject *rhs, int oparg,

    LOOKUP_SPEC(compactlong_float_specs, oparg);
    LOOKUP_SPEC(float_compactlong_specs, oparg);
+    LOOKUP_SPEC(compactlongs_specs, oparg);


Why three tables, rather than one?

A single table would need to have a list of (guard, action) pairs for each OP. So it's a table of tables. Same thing basically.

chris-eibl · 2025-01-22T16:21:25Z

It looks like we are seeing a high proportion of specialization failures for non-compact ints with the & operator. I don't know why that would be. My guess is that some of the benchmarks are using ints as bit vectors and using more than one digit.

Just a wild guess: could those be from enums, which derive from int? Especially flag enums?

iritkatriel · 2025-01-22T17:31:41Z

Just a wild guess: could those be from enums, which derive from int? Especially flag enums?

I don't think so. We check for int with PyLong_CheckExact.

eendebakpt · 2025-01-22T21:03:17Z

I suspect the misses might be bm_pyflate:

https://github.com/python/pyperformance/blob/1d9261a7da8fcaa642a36181db8e7c4a306a1303/pyperformance/data-files/benchmarks/bm_pyflate/run_benchmark.py#L153-L163

Other python constructions that would causes misses (but are not in pyperformance afaics) are uuid.uuid4 (and variants) and bitwise logical operations on the hash of python objects.

iritkatriel · 2025-01-22T21:40:41Z

I suspect the misses might be bm_pyflate:

https://github.com/python/pyperformance/blob/1d9261a7da8fcaa642a36181db8e7c4a306a1303/pyperformance/data-files/benchmarks/bm_pyflate/run_benchmark.py#L153-L163

Could be.

https://github.com/faster-cpython/benchmarking/blob/main/results/bm-20250121-3.14.0a4%2B-6476205/bm-20250121-azure-x86_64-iritkatriel-binaryops-3.14.0a4%2B-6476205-pystats-pyflate.md

iritkatriel · 2025-01-22T21:41:59Z

(Would be nice if the report was rendered so that it's easy to see which benchmark contributed to a stat.)

pythongh-100239: specialize bitwise logical binary ops on non-negativ…

c6f2deb

…e int

bedevere-app bot mentioned this pull request Jan 16, 2025

Specialize long tail of binary operations using a table. #100239

Open

bedevere-app bot added the awaiting core review label Jan 16, 2025

Merge branch 'main' into binaryops

c57fe46

iritkatriel requested review from mdboom and markshannon January 16, 2025 22:57

eendebakpt reviewed Jan 16, 2025

View reviewed changes

eendebakpt reviewed Jan 17, 2025

View reviewed changes

Misc/NEWS.d/next/Core_and_Builtins/2025-01-16-22-54-12.gh-issue-100239.7_HpBU.rst Outdated Show resolved Hide resolved

the globals thing

862bc67

iritkatriel requested a review from ericsnowcurrently as a code owner January 17, 2025 09:03

Update Misc/NEWS.d/next/Core_and_Builtins/2025-01-16-22-54-12.gh-issu…

69e66f5

…e-100239.7_HpBU.rst Co-authored-by: Pieter Eendebak <[email protected]>

iritkatriel added 4 commits January 17, 2025 18:09

add checks in test

0173413

no need to exclude negative

61d8bb3

EXIT_IF --> DEOPT_IF

1f48300

add specialization failure kind stats

e43ac58

kumaraditya303 reviewed Jan 18, 2025

View reviewed changes

Python/specialize.c Show resolved Hide resolved

chris-eibl reviewed Jan 18, 2025

View reviewed changes

iritkatriel commented Jan 18, 2025

View reviewed changes

Misc/NEWS.d/next/Core_and_Builtins/2025-01-16-22-54-12.gh-issue-100239.7_HpBU.rst Outdated Show resolved Hide resolved

remove non-negative

6916df4

chris-eibl reviewed Jan 19, 2025

View reviewed changes

add inplace ops

6476205

iritkatriel changed the title ~~gh-100239: specialize bitwise logical binary ops on non-negative int~~ gh-100239: specialize bitwise logical binary ops on ints Jan 21, 2025

iritkatriel added 2 commits January 21, 2025 17:10

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

8448b4b

PyLong_FromLong --> PyLong_FromSsize_t

634b388

chris-eibl mentioned this pull request Jan 21, 2025

Missing fast path in PyLong_From*() functions for compact integers #129149

Open

markshannon reviewed Jan 22, 2025

View reviewed changes

chris-eibl mentioned this pull request Jan 22, 2025

gh-129149: Add Missing fast path in PYLONG_FROM_UINT macro for compact integers #129168

Merged

Merge branch 'main' into binaryops

6041025

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

gh-100239: specialize bitwise logical binary ops on ints #128927

gh-100239: specialize bitwise logical binary ops on ints #128927

iritkatriel commented Jan 16, 2025 •

edited

Loading

eendebakpt Jan 16, 2025

iritkatriel Jan 17, 2025

eendebakpt Jan 17, 2025

iritkatriel Jan 17, 2025

iritkatriel commented Jan 17, 2025

chris-eibl left a comment

chris-eibl Jan 19, 2025

iritkatriel Jan 21, 2025

chris-eibl Jan 21, 2025

chris-eibl commented Jan 21, 2025

markshannon left a comment

markshannon Jan 22, 2025

iritkatriel Jan 22, 2025

chris-eibl commented Jan 22, 2025

iritkatriel commented Jan 22, 2025

eendebakpt commented Jan 22, 2025

iritkatriel commented Jan 22, 2025

iritkatriel commented Jan 22, 2025

gh-100239: specialize bitwise logical binary ops on ints #128927

Are you sure you want to change the base?

gh-100239: specialize bitwise logical binary ops on ints #128927

Conversation

iritkatriel commented Jan 16, 2025 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

iritkatriel commented Jan 17, 2025

chris-eibl left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

chris-eibl commented Jan 21, 2025

markshannon left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

chris-eibl commented Jan 22, 2025

iritkatriel commented Jan 22, 2025

eendebakpt commented Jan 22, 2025

iritkatriel commented Jan 22, 2025

iritkatriel commented Jan 22, 2025

iritkatriel commented Jan 16, 2025 •

edited

Loading