User Details
- User Since
- Nov 15 2022, 6:55 AM (46 w, 1 d)
Fri, Sep 15
Rebased and dropped comment change
Thu, Sep 14
@qcolombet I've updated the comment. Do you think we still need to modify X86FastPreTileConfig? In the latest version we disable this pass for GlobalISel because it was intended for the code selected by FastISel. It's unclear whether we need it with GlobalISel in the future or not.
Updated a comment and rebased
Fri, Sep 8
@qcolombet ping
Rebased
Aug 30 2023
Rebased and updated a comment
Change the comment to say that only alive registers are guaranteed to have a regclass.
Aug 18 2023
Aug 17 2023
Just curious, in RFC we have -mavx10.x-256/-mavx10.x-512 but here we refer to -mavx10.x/-mavx10.x,-mavx10-512bit. Is it compliant with GCC, or the revision is just for the illustrative purpose?
Aug 14 2023
Use X86FastPreTileConfigPass only with FastISel
Aug 10 2023
Aug 8 2023
Moved from a ll test to a more illustrative mir test.
Typo
Jul 26 2023
Return nounwind attribute missed during the last rebase.
Rebased
Jul 11 2023
Hi @pengfei @RKSimon, what do you think about such diagnostic improvement? It works perfectly with clang, however llc ignores errors in a handler and continues execution. It seems that we need a combination of diagnostic and report_fatal_error to stop both clang and llc. Or is it expected behavior of llc?
Jun 16 2023
Rebased and removed FIXME
Jun 14 2023
May 15 2023
Addressed the comments.
May 12 2023
Rebase. Spelling.
Rebased
Great! Thanks. LGTM.
May 11 2023
May 10 2023
LGTM. Thanks!
May 8 2023
May 4 2023
Maybe a few more vector tests? Existing tests cover the no-nans and no-signed-zeros case, they were added to avoid bloat of scalarization. E.g. a test with all checks (generic vectorized version) and two test with constant vectors to eliminate a zero check and a NaN check.
LGTM. Thank you!
LGTM. Glibc tests pass (I haven't checked correspondence of qnan/snan though).
May 3 2023
Good catch! Missed that we don't need to check Y on NaN because it is already a result of min therefore we can get rid of NaN loading.
Apr 26 2023
Apr 25 2023
Fix single line if style.
Apr 24 2023
Rebased.
Excluded refactor of half.ll.
Apr 18 2023
Rebased.
Added nounwind attribute to tests.
Apr 14 2023
Rebased.
Uncommented existing tests for the intrinsics.
Addressed to comments.
Apr 7 2023
@RKSimon @goldstein.w.n ping
Rebase
Rebased
Rebased
Apr 3 2023
Support i686 target: can't use integer representation of double -0.0
Mar 31 2023
Addressed comments.
Added comments.
Fixed negative value search in case of undefs.
Updated tests.
Updated tests with undefs.
Use not only splat vector constants.
Mar 30 2023
Addressed the comments.
New tests
Mar 28 2023
Mar 27 2023
Mar 19 2023
@RKSimon @goldstein.w.n ping.
Mar 15 2023
Addressed formatting comments.
Check f16 explicitly even if avx512f16 implies avx512dq for now.
Mar 14 2023
Broke formatting for premerge checks
Rebased.
Moved from combine to lowering.
Supported f16 version.
Added optimization for avx512dq.
Added and updated tests.
Mar 12 2023
Mar 9 2023
Rebased.
Supported cases with nsz and nnan.
Updated tests.
Mar 8 2023
Mar 6 2023
Feb 24 2023
LGTM. I second that masked versions should be handled as well. I thought they have TP equal to perms but I've double checked and it seems that masked shuffles have TP=0.5 comparing to perms.
Dec 21 2022
Rebased.
Dec 18 2022
Dec 14 2022
Rebased and added a bitcast of IsNOT.
Dec 13 2022
Added vector splitting
Dec 6 2022
Dec 5 2022
Propose more generic approach using vector_shuffle
Nov 28 2022
Rebased on precommit tests
Update tests with old codegen
Nov 27 2022
Add peeking through bitcasts for BROADCAST as well.
Nov 25 2022
Address to comments and put tests into https://reviews.llvm.org/D138734