julia
Improve `reinterpret` performance for padded types, with minimal harm to compilation time
#60415
Open

Improve `reinterpret` performance for padded types, with minimal harm to compilation time #60415

NHDaly wants to merge 41 commits into master from nhd/reinterpret-padded-struct-performance
NHDaly
NHDaly Import fast_reinterpret directly from RAI
7cab3027
NHDaly disable fallbacks that maybe aren't needed?
5854df18
NHDaly First attempt at recursive padded-padded src=>dst
c8915e6a
NHDaly Constant-folding version, which only uses tuple-unrolling
fa0edf8a
NHDaly Trying out performance tweaks; nothing really helped
32f75970
NHDaly Revert "Trying out performance tweaks; nothing really helped"
4d0ff614
NHDaly FINALLY fixed the terribly compilation time.
089cb42a
NHDaly Simplify the iteration in match_packed_regions
d0d57dfd
NHDaly Way faster compilation. :) Use depth-first iteration instead of recur…
bb11f4ed
NHDaly Reuse existing Base.padding() for even faster compilation!!!
e39093f3
NHDaly Optimization for matching struct padding
8f77bcd0
NHDaly Fix correctness, keep perf - inspired by Base.padding.
3c1fefbb
NHDaly Throw the right kind of error
99981941
NHDaly Switch to depth-first search which is faster than recursion at compil…
7b86b5b6
NHDaly Very minor compiler perf optimization: reuse the same work instead of
9dbd35bb
NHDaly Simplify code: The new impl generalizes all the others.
9015f870
NHDaly remove benchmark only code;
70035110
NHDaly comments
063c737a
NHDaly Start replacing Base's _reinterpret with our implementation.
dbf0507d
NHDaly Don't mix Base.padding w/ new computations. This saves some compilati…
e5dc76ad
NHDaly Small perf improvement to ispacked
f7964eef
NHDaly Merge branch 'master' into nhd/reinterpret-padded-struct-performance
f3dd0b49
NHDaly Comments: _packedsize, remove lint comment
c56e139d
NHDaly Use the existing ArgumentErrors from master.
5b246605
NHDaly Remove unnecessary macro
60942b4e
NHDaly Add more tests for reinterpret
5cb44c3b
NHDaly fixup merge conflict
53e69766
NHDaly NHDaly added performance
NHDaly NHDaly marked this pull request as ready for review 61 days ago
NHDaly whitespace
34bd84ee
KristofferC
NHDaly Add broken tests
16b20a81
NHDaly NHDaly force pushed from cbcfcefa to 16b20a81 61 days ago
NHDaly
NHDaly Fix broken tests for nested structs
1da2607c
NHDaly
NHDaly Perf improvement on the _packed_regions
6a0bd80f
NHDaly NHDaly requested a review from copilot-pull-request-reviewer copilot-pull-request-reviewer 61 days ago
copilot-pull-request-reviewer
copilot-pull-request-reviewer commented on 2025-12-19
NHDaly Simplify the comment on the byte-cast optimization
28caa396
NHDaly remove todo comment
b5e2e733
NHDaly Make :foldable assertions tighter.
a3c7869f
NHDaly cleanup from copilot
c5b9c19d
NHDaly improve assert message
0ea5b495
NHDaly Add broken test for missed bug;
64a93ecf
NHDaly Fix bug for types that match prefix, but differ in trailing padding
92ad4a55
NHDaly
NHDaly commented on 2025-12-23
NHDaly Update base/reinterpret.jl
7af9dab9
NHDaly Fix reinterpret tests for 32-bit.
e303c350
NHDaly NHDaly requested a review from BioTurboNick BioTurboNick 57 days ago
BioTurboNick
BioTurboNick commented on 2025-12-25
NHDaly
NHDaly commented on 2026-01-03
NHDaly Update base/reinterpret.jl
f213d55f

Login to write a write a comment.

Login via GitHub

Assignees
No one assigned
Labels
Milestone