Page MenuHomePhabricator
Feed Advanced Search

Yesterday

samtebbs updated the diff for D87836: [ARM] Fold select_cc(vecreduce_[u|s][min|max], x) into VMINV or VMAXV.

Add type promotion checks and two new checks

Tue, Sep 29, 8:16 AM · Restricted Project

Mon, Sep 28

samtebbs added inline comments to D88419: [RDA] Switch isSafeToMove iterators.
Mon, Sep 28, 8:54 AM · Restricted Project

Thu, Sep 24

samtebbs accepted D87819: [ARM] Find VPT implicitly predicated by VCTP.

Nice change! LGTM

Thu, Sep 24, 7:17 AM · Restricted Project
samtebbs updated the diff for D87836: [ARM] Fold select_cc(vecreduce_[u|s][min|max], x) into VMINV or VMAXV.

Add truncate_sext test

Thu, Sep 24, 6:45 AM · Restricted Project
samtebbs abandoned D88022: [ARM][LowOverheadLoops] Check VCMP operands have same def as the VPT before combining.

Closed since it's no longer needed.

Thu, Sep 24, 6:34 AM · Restricted Project

Wed, Sep 23

samtebbs updated the diff for D87836: [ARM] Fold select_cc(vecreduce_[u|s][min|max], x) into VMINV or VMAXV.

Allow inverted condition codes

Wed, Sep 23, 3:48 AM · Restricted Project

Mon, Sep 21

samtebbs added inline comments to D87836: [ARM] Fold select_cc(vecreduce_[u|s][min|max], x) into VMINV or VMAXV.
Mon, Sep 21, 9:35 AM · Restricted Project
samtebbs updated the diff for D87836: [ARM] Fold select_cc(vecreduce_[u|s][min|max], x) into VMINV or VMAXV.

Be a bit more strict.

Mon, Sep 21, 9:35 AM · Restricted Project
samtebbs added inline comments to D87819: [ARM] Find VPT implicitly predicated by VCTP.
Mon, Sep 21, 7:51 AM · Restricted Project
samtebbs requested review of D88022: [ARM][LowOverheadLoops] Check VCMP operands have same def as the VPT before combining.
Mon, Sep 21, 7:28 AM · Restricted Project
samtebbs added inline comments to D87616: [ARM][LowOverheadLoops] Combine a VCMP and VPST into a VPT.
Mon, Sep 21, 7:27 AM · Restricted Project
samtebbs added inline comments to D87836: [ARM] Fold select_cc(vecreduce_[u|s][min|max], x) into VMINV or VMAXV.
Mon, Sep 21, 6:35 AM · Restricted Project
samtebbs updated the diff for D87836: [ARM] Fold select_cc(vecreduce_[u|s][min|max], x) into VMINV or VMAXV.

Fold using the LHS and check for LHS and RHS opcodes explicitly.

Mon, Sep 21, 6:35 AM · Restricted Project

Fri, Sep 18

samtebbs added inline comments to D87836: [ARM] Fold select_cc(vecreduce_[u|s][min|max], x) into VMINV or VMAXV.
Fri, Sep 18, 5:25 AM · Restricted Project
samtebbs updated the diff for D87836: [ARM] Fold select_cc(vecreduce_[u|s][min|max], x) into VMINV or VMAXV.

Support commuted operands and add tests to cover them.

Fri, Sep 18, 5:25 AM · Restricted Project

Thu, Sep 17

samtebbs requested review of D87836: [ARM] Fold select_cc(vecreduce_[u|s][min|max], x) into VMINV or VMAXV.
Thu, Sep 17, 8:19 AM · Restricted Project

Wed, Sep 16

samtebbs added inline comments to D87751: [RDA] Fix getUniqueReachingDef for self loops.
Wed, Sep 16, 3:06 AM · Restricted Project
samtebbs added a comment to D87616: [ARM][LowOverheadLoops] Combine a VCMP and VPST into a VPT.

Tests updated in rGac2717bfdd0d

Wed, Sep 16, 3:02 AM · Restricted Project
samtebbs committed rGac2717bfdd0d: [ARM][LowOverheadLoops] Fix tests after ef0b9f3 (authored by samtebbs).
[ARM][LowOverheadLoops] Fix tests after ef0b9f3
Wed, Sep 16, 3:01 AM
samtebbs closed D87616: [ARM][LowOverheadLoops] Combine a VCMP and VPST into a VPT.

Closed by rGef0b9f3307a1

Wed, Sep 16, 1:36 AM · Restricted Project
samtebbs committed rGef0b9f3307a1: [ARM][LowOverheadLoops] Combine a VCMP and VPST into a VPT (authored by samtebbs).
[ARM][LowOverheadLoops] Combine a VCMP and VPST into a VPT
Wed, Sep 16, 1:34 AM

Tue, Sep 15

samtebbs added inline comments to D87616: [ARM][LowOverheadLoops] Combine a VCMP and VPST into a VPT.
Tue, Sep 15, 6:36 AM · Restricted Project
samtebbs updated the diff for D87616: [ARM][LowOverheadLoops] Combine a VCMP and VPST into a VPT.

Clean up the formatting a little.

Tue, Sep 15, 6:33 AM · Restricted Project
samtebbs updated the diff for D87616: [ARM][LowOverheadLoops] Combine a VCMP and VPST into a VPT.

Remove -O3 from test and improve VCMP detection.

Tue, Sep 15, 6:28 AM · Restricted Project

Mon, Sep 14

samtebbs requested review of D87616: [ARM][LowOverheadLoops] Combine a VCMP and VPST into a VPT.
Mon, Sep 14, 8:02 AM · Restricted Project

Thu, Sep 10

samtebbs committed rGb81c57d646e4: [ARM][LowOverheadLoops] Allow tail predication on predicated instructions with… (authored by samtebbs).
[ARM][LowOverheadLoops] Allow tail predication on predicated instructions with…
Thu, Sep 10, 2:35 AM
samtebbs closed D87376: [ARM][LowOverheadLoops] Allow tail predication on predicated instructions with unknown lane values.
Thu, Sep 10, 2:35 AM · Restricted Project

Wed, Sep 9

samtebbs updated the diff for D87376: [ARM][LowOverheadLoops] Allow tail predication on predicated instructions with unknown lane values.

Add test

Wed, Sep 9, 6:56 AM · Restricted Project
samtebbs requested review of D87376: [ARM][LowOverheadLoops] Allow tail predication on predicated instructions with unknown lane values.
Wed, Sep 9, 6:16 AM · Restricted Project

Tue, Sep 8

samtebbs closed D87238: [ARM][LowOverheadLoops] Remove modifications to the correct element count register.

Thanks @samparker . Closed by 7aabb6ad7764

Tue, Sep 8, 2:31 AM · Restricted Project
samtebbs committed rG7aabb6ad7764: [ARM][LowOverheadLoops] Remove modifications to the correct element (authored by samtebbs).
[ARM][LowOverheadLoops] Remove modifications to the correct element
Tue, Sep 8, 2:30 AM

Mon, Sep 7

samtebbs updated the summary of D87238: [ARM][LowOverheadLoops] Remove modifications to the correct element count register.
Mon, Sep 7, 7:56 AM · Restricted Project
samtebbs requested review of D87238: [ARM][LowOverheadLoops] Remove modifications to the correct element count register.
Mon, Sep 7, 7:55 AM · Restricted Project

Tue, Sep 1

samtebbs committed rG15e880a04fcf: [DAGCombiner] Fold an AND of a masked load into a zext_masked_load (authored by samtebbs).
[DAGCombiner] Fold an AND of a masked load into a zext_masked_load
Tue, Sep 1, 9:03 AM
samtebbs closed D86789: [DAGCombiner] Fold an AND of a masked load into a zext_masked_load.
Tue, Sep 1, 9:02 AM · Restricted Project
samtebbs updated the diff for D86789: [DAGCombiner] Fold an AND of a masked load into a zext_masked_load.

Show difference to previous codegen.

Tue, Sep 1, 7:49 AM · Restricted Project
samtebbs added inline comments to D86789: [DAGCombiner] Fold an AND of a masked load into a zext_masked_load.
Tue, Sep 1, 7:23 AM · Restricted Project
samtebbs added inline comments to D86789: [DAGCombiner] Fold an AND of a masked load into a zext_masked_load.
Tue, Sep 1, 7:17 AM · Restricted Project
samtebbs added inline comments to D86789: [DAGCombiner] Fold an AND of a masked load into a zext_masked_load.
Tue, Sep 1, 6:55 AM · Restricted Project
samtebbs added inline comments to D86789: [DAGCombiner] Fold an AND of a masked load into a zext_masked_load.
Tue, Sep 1, 6:52 AM · Restricted Project
samtebbs added inline comments to D86789: [DAGCombiner] Fold an AND of a masked load into a zext_masked_load.
Tue, Sep 1, 6:16 AM · Restricted Project
samtebbs updated the diff for D86789: [DAGCombiner] Fold an AND of a masked load into a zext_masked_load.

Use isMask and move hasOneUse checks

Tue, Sep 1, 6:15 AM · Restricted Project
samtebbs added inline comments to D86789: [DAGCombiner] Fold an AND of a masked load into a zext_masked_load.
Tue, Sep 1, 4:13 AM · Restricted Project
samtebbs updated the diff for D86789: [DAGCombiner] Fold an AND of a masked load into a zext_masked_load.

Improved the BuildVec processing and added old checks to test.

Tue, Sep 1, 4:13 AM · Restricted Project

Aug 28 2020

samtebbs requested review of D86789: [DAGCombiner] Fold an AND of a masked load into a zext_masked_load.
Aug 28 2020, 8:25 AM · Restricted Project

Aug 26 2020

samtebbs committed rG85dd852a0d46: [RDA] Don't visit the BB of the instruction in getReachingUniqueMIDef (authored by samtebbs).
[RDA] Don't visit the BB of the instruction in getReachingUniqueMIDef
Aug 26 2020, 4:41 AM
samtebbs closed D86607: [RDA] Don't visit the BB of the instruction in getUniqueReachingMIDef.
Aug 26 2020, 4:41 AM · Restricted Project
samtebbs added a comment to D86607: [RDA] Don't visit the BB of the instruction in getUniqueReachingMIDef.

I was just writing that I understood that creating a test for this one was very difficult? I.e., creating a small test case, was that this case? Looks like Sam has one now......should it not be part of this change? But anyway, it's fine I guess.

Aug 26 2020, 3:51 AM · Restricted Project
samtebbs added a comment to D86607: [RDA] Don't visit the BB of the instruction in getUniqueReachingMIDef.

I have also stumbled across this bug while making a change so I also have this fix downstream! The existing test cases will cover this once I upload another change, so I think this is fine. Thanks!

Aug 26 2020, 3:50 AM · Restricted Project
samtebbs requested review of D86607: [RDA] Don't visit the BB of the instruction in getUniqueReachingMIDef.
Aug 26 2020, 3:26 AM · Restricted Project

Aug 19 2020

samtebbs added inline comments to D86087: [ARM] Use mov operand if the mov cannot be moved while tail predicating.
Aug 19 2020, 1:38 AM · Restricted Project

Aug 18 2020

samtebbs added a comment to D86087: [ARM] Use mov operand if the mov cannot be moved while tail predicating.

Thanks! LGTM

Aug 18 2020, 9:11 AM · Restricted Project
samtebbs committed rG31f02ac60aa8: [ARM] Use mov operand if the mov cannot be moved while tail predicating (authored by samtebbs).
[ARM] Use mov operand if the mov cannot be moved while tail predicating
Aug 18 2020, 9:11 AM
samtebbs closed D86087: [ARM] Use mov operand if the mov cannot be moved while tail predicating.
Aug 18 2020, 9:10 AM · Restricted Project
samtebbs updated the diff for D86087: [ARM] Use mov operand if the mov cannot be moved while tail predicating.

Add MIR test

Aug 18 2020, 7:48 AM · Restricted Project
samtebbs added inline comments to D86087: [ARM] Use mov operand if the mov cannot be moved while tail predicating.
Aug 18 2020, 7:46 AM · Restricted Project
samtebbs added a comment to D86087: [ARM] Use mov operand if the mov cannot be moved while tail predicating.

Sure, there is already a mov-after-dls mir test that I can add to.

Just generating a MIR file from mov-operand.ll would be fine and save you from manually having to alter an existing test.

Aug 18 2020, 7:43 AM · Restricted Project
samtebbs updated the diff for D86087: [ARM] Use mov operand if the mov cannot be moved while tail predicating.

Rename getCount, use is isMovRegOpcode and cache ElemDef->getOperand(1)

Aug 18 2020, 6:28 AM · Restricted Project
samtebbs updated the summary of D86087: [ARM] Use mov operand if the mov cannot be moved while tail predicating.
Aug 18 2020, 3:26 AM · Restricted Project
samtebbs updated the summary of D86087: [ARM] Use mov operand if the mov cannot be moved while tail predicating.
Aug 18 2020, 3:25 AM · Restricted Project
samtebbs added a comment to D86087: [ARM] Use mov operand if the mov cannot be moved while tail predicating.

I think having the IR test is a good idea, but could you also upload it as a MIR too, just because this pass is very far away from the IR input.

Aug 18 2020, 2:53 AM · Restricted Project

Aug 17 2020

samtebbs requested review of D86087: [ARM] Use mov operand if the mov cannot be moved while tail predicating.
Aug 17 2020, 9:14 AM · Restricted Project

Jul 30 2020

samtebbs committed rG276ed5f7e4ee: [DAGCombiner] Fold sext_inreg of a masked load into a sign extended masked load (authored by samtebbs).
[DAGCombiner] Fold sext_inreg of a masked load into a sign extended masked load
Jul 30 2020, 2:34 AM
samtebbs closed D84332: [DAGCombiner] Fold sext_inreg of a masked load into a signed extended masked load.
Jul 30 2020, 2:34 AM · Restricted Project

Jul 28 2020

samtebbs updated the diff for D84332: [DAGCombiner] Fold sext_inreg of a masked load into a signed extended masked load.

Remove -tail-predication option from test and change SEXTLOAD to NON_EXTLOAD in DAGCombiner check.

Jul 28 2020, 3:28 AM · Restricted Project

Jul 27 2020

samtebbs added inline comments to D84332: [DAGCombiner] Fold sext_inreg of a masked load into a signed extended masked load.
Jul 27 2020, 7:38 AM · Restricted Project
samtebbs updated the diff for D84332: [DAGCombiner] Fold sext_inreg of a masked load into a signed extended masked load.

Add more tests and clean up checks in visitSIGN_EXTEND_INREG

Jul 27 2020, 7:38 AM · Restricted Project

Jul 24 2020

samtebbs updated the diff for D84332: [DAGCombiner] Fold sext_inreg of a masked load into a signed extended masked load.

Simplify test and add extra checks to sext masked load generation.

Jul 24 2020, 9:04 AM · Restricted Project

Jul 23 2020

samtebbs added inline comments to D84332: [DAGCombiner] Fold sext_inreg of a masked load into a signed extended masked load.
Jul 23 2020, 7:56 AM · Restricted Project

Jul 22 2020

Herald added a project to D84332: [DAGCombiner] Fold sext_inreg of a masked load into a signed extended masked load: Restricted Project.
Jul 22 2020, 8:49 AM · Restricted Project

Jul 17 2020

samtebbs committed rG6c348e4067b0: [HWLoops] Stop converting to a while loop when it would be unsafe to (authored by samtebbs).
[HWLoops] Stop converting to a while loop when it would be unsafe to
Jul 17 2020, 3:47 AM
samtebbs closed D83953: [HWLoops] Stop converting to a while loop when it would be unsafe to.
Jul 17 2020, 3:47 AM · Restricted Project
samtebbs updated the diff for D83953: [HWLoops] Stop converting to a while loop when it would be unsafe to.

Update tail predication option in test

Jul 17 2020, 2:56 AM · Restricted Project
samtebbs added a comment to D83953: [HWLoops] Stop converting to a while loop when it would be unsafe to.

Nice and tidy fix, thanks!

Jul 17 2020, 2:46 AM · Restricted Project
samtebbs updated the diff for D83953: [HWLoops] Stop converting to a while loop when it would be unsafe to.

Fix formatting

Jul 17 2020, 2:44 AM · Restricted Project

Jul 16 2020

samtebbs updated the diff for D83953: [HWLoops] Stop converting to a while loop when it would be unsafe to.

Add context to diff

Jul 16 2020, 8:14 AM · Restricted Project
Herald added a project to D83953: [HWLoops] Stop converting to a while loop when it would be unsafe to: Restricted Project.
Jul 16 2020, 8:08 AM · Restricted Project

Jun 30 2020

samtebbs updated the diff for D82553: [ARM] Allow rounding intrinsics to be tail predicated.

Add LLVM_FALLTHROUGH

Jun 30 2020, 8:39 AM · Restricted Project
samtebbs added inline comments to D82553: [ARM] Allow rounding intrinsics to be tail predicated.
Jun 30 2020, 8:39 AM · Restricted Project
samtebbs updated the diff for D82570: [ARM] Allow the fabs intrinsic to be tail predicated.
Jun 30 2020, 8:39 AM · Restricted Project
samtebbs updated the diff for D82570: [ARM] Allow the fabs intrinsic to be tail predicated.

Rebase and clean up test

Jun 30 2020, 8:39 AM · Restricted Project
samtebbs updated the diff for D82553: [ARM] Allow rounding intrinsics to be tail predicated.

Fix types of nearbyint intrinsic

Jun 30 2020, 8:06 AM · Restricted Project
samtebbs updated the diff for D82553: [ARM] Allow rounding intrinsics to be tail predicated.

Fix up nearbyint test, make ST a class member and move fma case.

Jun 30 2020, 4:18 AM · Restricted Project

Jun 26 2020

samtebbs added inline comments to D82553: [ARM] Allow rounding intrinsics to be tail predicated.
Jun 26 2020, 3:45 AM · Restricted Project
samtebbs updated the diff for D82553: [ARM] Allow rounding intrinsics to be tail predicated.

Only tail predicate if the subtarget has float ops, add nearbyint test and clean up the test file.

Jun 26 2020, 3:45 AM · Restricted Project

Jun 25 2020

samtebbs created D82571: [ARM] Allow the usub_sat and ssub_sat intrinsics to be tail predicated.
Jun 25 2020, 9:07 AM · Restricted Project
samtebbs updated the diff for D82570: [ARM] Allow the fabs intrinsic to be tail predicated.
Jun 25 2020, 9:07 AM · Restricted Project
samtebbs created D82570: [ARM] Allow the fabs intrinsic to be tail predicated.
Jun 25 2020, 9:07 AM · Restricted Project
samtebbs updated the diff for D82553: [ARM] Allow rounding intrinsics to be tail predicated.

Add missing floor and ceil tests

Jun 25 2020, 8:33 AM · Restricted Project
samtebbs created D82553: [ARM] Allow rounding intrinsics to be tail predicated.
Jun 25 2020, 8:00 AM · Restricted Project
samtebbs committed rG187f627a5057: [ARM] Allow tail predication on sadd_sat and uadd_sat intrinsics (authored by samtebbs).
[ARM] Allow tail predication on sadd_sat and uadd_sat intrinsics
Jun 25 2020, 4:10 AM
samtebbs closed D82377: [ARM] Allow tail predication on sadd_sat and uadd_sat intrinsics.

Closed by 187f627a505

Jun 25 2020, 4:09 AM · Restricted Project

Jun 24 2020

samtebbs updated the diff for D82377: [ARM] Allow tail predication on sadd_sat and uadd_sat intrinsics.

Remove dso_local and local_unnamed from test

Jun 24 2020, 3:12 AM · Restricted Project
samtebbs updated the diff for D82377: [ARM] Allow tail predication on sadd_sat and uadd_sat intrinsics.

Change to switch statement and merge tests

Jun 24 2020, 3:12 AM · Restricted Project

Jun 23 2020

samtebbs added inline comments to D82377: [ARM] Allow tail predication on sadd_sat and uadd_sat intrinsics.
Jun 23 2020, 9:04 AM · Restricted Project
samtebbs created D82377: [ARM] Allow tail predication on sadd_sat and uadd_sat intrinsics.
Jun 23 2020, 6:53 AM · Restricted Project

Sep 13 2019

samtebbs added a comment to D67444: [ARM][MVE] Add invalidForTailPredication to TSFlags.

+1 :)

Sep 13 2019, 2:50 AM · Restricted Project
samtebbs committed rG1572b6850957: [ARM] Add support for MVE vmaxv and vminv (authored by samtebbs).
[ARM] Add support for MVE vmaxv and vminv
Sep 13 2019, 2:13 AM
samtebbs committed rL371827: [ARM] Add support for MVE vmaxv and vminv.
[ARM] Add support for MVE vmaxv and vminv
Sep 13 2019, 2:13 AM
samtebbs closed D66413: [ARM} Add support for MVE vmaxv and vminv.
Sep 13 2019, 2:13 AM · Restricted Project