Page MenuHomePhabricator

bsmith (Bradley Smith)
User

Projects

User does not belong to any projects.

User Details

User Since
Sep 12 2013, 6:57 AM (396 w, 6 d)

Recent Activity

Today

bsmith abandoned D100858: [AArch64][SVE] Allow generation of MOVPRFX for intrinsic nodes.

The fact that the DestructiveBinary case in this needs to insert extra MOV's to make up for what is essentially poor register allocation choices, is less than ideal. I'm abandoning this in favour of a proper solution at a later date that will depend on D88595.

Wed, Apr 21, 10:05 AM · Restricted Project

Yesterday

bsmith requested review of D100858: [AArch64][SVE] Allow generation of MOVPRFX for intrinsic nodes.
Tue, Apr 20, 7:48 AM · Restricted Project
bsmith committed rGb8b075d8d744: [AArch64][SVE] Lower MULHU/MULHS nodes to umulh/smulh instructions (authored by bsmith).
[AArch64][SVE] Lower MULHU/MULHS nodes to umulh/smulh instructions
Tue, Apr 20, 7:18 AM
bsmith closed D100487: [AArch64][SVE] Lower MULHU/MULHS nodes to umulh/smulh instructions.
Tue, Apr 20, 7:18 AM · Restricted Project
bsmith added inline comments to D100487: [AArch64][SVE] Lower MULHU/MULHS nodes to umulh/smulh instructions.
Tue, Apr 20, 5:42 AM · Restricted Project

Fri, Apr 16

bsmith updated the diff for D100487: [AArch64][SVE] Lower MULHU/MULHS nodes to umulh/smulh instructions.
  • Rename ISD nodes to match common names
  • Preserve some alphabetical ordering
  • Fix 2048 vector width tests
  • Update fixed width tests to test vector widths smaller than the given value
  • Allow generation of SVE mulh instructions for neon sized i64 vectors
  • Add test to check divide expansion that can now happen
  • Add minsize attribute to some SVE divide tests to prevent divide expansion from happening
Fri, Apr 16, 5:56 AM · Restricted Project

Thu, Apr 15

bsmith committed rG22c017f0f902: [AArch64][NEON] Match (or (and -a b) (and (a+1) b)) => bit select (authored by bsmith).
[AArch64][NEON] Match (or (and -a b) (and (a+1) b)) => bit select
Thu, Apr 15, 5:53 AM
bsmith closed D100304: [AArch64][NEON] Match (or (and -a b) (and (a+1) b)) => bit select.
Thu, Apr 15, 5:53 AM · Restricted Project

Wed, Apr 14

bsmith requested review of D100487: [AArch64][SVE] Lower MULHU/MULHS nodes to umulh/smulh instructions.
Wed, Apr 14, 8:28 AM · Restricted Project
bsmith updated the diff for D100304: [AArch64][NEON] Match (or (and -a b) (and (a+1) b)) => bit select.
  • Use attribute in tests in favour of -mattr on run line
Wed, Apr 14, 4:11 AM · Restricted Project

Tue, Apr 13

bsmith added inline comments to D100304: [AArch64][NEON] Match (or (and -a b) (and (a+1) b)) => bit select.
Tue, Apr 13, 9:18 AM · Restricted Project
bsmith updated the diff for D100304: [AArch64][NEON] Match (or (and -a b) (and (a+1) b)) => bit select.
  • Add missing test
  • Tidy up comment
Tue, Apr 13, 3:43 AM · Restricted Project

Mon, Apr 12

bsmith requested review of D100304: [AArch64][NEON] Match (or (and -a b) (and (a+1) b)) => bit select.
Mon, Apr 12, 7:00 AM · Restricted Project
bsmith committed rGf2593a0bd1b2: [AArch64][SVE] Remove redundant PTEST of MATCH/NMATCH results (authored by bsmith).
[AArch64][SVE] Remove redundant PTEST of MATCH/NMATCH results
Mon, Apr 12, 4:56 AM
bsmith closed D99584: [AArch64][SVE] Remove redundant PTEST of MATCH/NMATCH results.
Mon, Apr 12, 4:56 AM · Restricted Project

Thu, Apr 1

bsmith committed rG2f45e632c002: [AArch64][SVE] Improve codegen for select nodes with fixed types (authored by bsmith).
[AArch64][SVE] Improve codegen for select nodes with fixed types
Thu, Apr 1, 8:00 AM
bsmith closed D99418: [AArch64][SVE] Improve codegen for select nodes with fixed types.
Thu, Apr 1, 8:00 AM · Restricted Project
bsmith committed rG0934fa4f5d5b: [AArch64][SVE] SVE functions should use the SVE calling convention for fast… (authored by bsmith).
[AArch64][SVE] SVE functions should use the SVE calling convention for fast…
Thu, Apr 1, 7:52 AM
bsmith closed D99657: [AArch64][SVE] SVE functions should use the SVE calling convention for fast calls.
Thu, Apr 1, 7:52 AM · Restricted Project

Wed, Mar 31

bsmith committed rG4e52daa25404: [AArch64][SVE] Add tests for UREM/SREM using fixed SVE types (authored by bsmith).
[AArch64][SVE] Add tests for UREM/SREM using fixed SVE types
Wed, Mar 31, 8:23 AM
bsmith closed D99265: [AArch64][SVE] Add tests for UREM/SREM using fixed SVE types.
Wed, Mar 31, 8:23 AM · Restricted Project
bsmith updated the diff for D99657: [AArch64][SVE] SVE functions should use the SVE calling convention for fast calls.
  • Add extra tests that explicitly check the calling convention used
Wed, Mar 31, 8:07 AM · Restricted Project
bsmith updated the diff for D99418: [AArch64][SVE] Improve codegen for select nodes with fixed types.
  • Use sext for predicate mask extension, rather than anyext
Wed, Mar 31, 7:53 AM · Restricted Project
bsmith requested review of D99657: [AArch64][SVE] SVE functions should use the SVE calling convention for fast calls.
Wed, Mar 31, 5:28 AM · Restricted Project

Tue, Mar 30

bsmith updated the diff for D99418: [AArch64][SVE] Improve codegen for select nodes with fixed types.
  • Fixed incorrect parameter in tests
Tue, Mar 30, 6:38 AM · Restricted Project
bsmith requested review of D99584: [AArch64][SVE] Remove redundant PTEST of MATCH/NMATCH results.
Tue, Mar 30, 5:55 AM · Restricted Project
bsmith updated the diff for D99418: [AArch64][SVE] Improve codegen for select nodes with fixed types.
  • Remove superfluous call to LowerFixedLengthVectorSelectToSVE
Tue, Mar 30, 5:02 AM · Restricted Project

Mon, Mar 29

bsmith committed rG9745dce8c3dc: [SelectionDAG][AArch64][SVE] Perform SETCC condition legalization in… (authored by bsmith).
[SelectionDAG][AArch64][SVE] Perform SETCC condition legalization in…
Mon, Mar 29, 7:43 AM
bsmith closed D98939: [SelectionDAG][AArch64][SVE] Perform SETCC condition legalization in LegalizeVectorOps.
Mon, Mar 29, 7:42 AM · Restricted Project

Fri, Mar 26

bsmith requested review of D99418: [AArch64][SVE] Improve codegen for select nodes with fixed types.
Fri, Mar 26, 8:01 AM · Restricted Project

Thu, Mar 25

bsmith added inline comments to D98939: [SelectionDAG][AArch64][SVE] Perform SETCC condition legalization in LegalizeVectorOps.
Thu, Mar 25, 5:40 AM · Restricted Project
bsmith updated the diff for D98939: [SelectionDAG][AArch64][SVE] Perform SETCC condition legalization in LegalizeVectorOps.
  • Remove unfinished case for SELECT_CC, this isn't needed currently for SVE, can be added later if/when needed.
  • Tidy up code in VectorLegalizer::ExpandSETCC.
  • Fix bogus test names.
Thu, Mar 25, 5:40 AM · Restricted Project

Wed, Mar 24

bsmith requested review of D99265: [AArch64][SVE] Add tests for UREM/SREM using fixed SVE types.
Wed, Mar 24, 7:23 AM · Restricted Project

Mar 22 2021

bsmith updated the diff for D98939: [SelectionDAG][AArch64][SVE] Perform SETCC condition legalization in LegalizeVectorOps.
  • Fixed failures in PPC/RISCV tests due to slightly different codegen.
  • Use getBoolConstant instead of manually constructing true/false values.
Mar 22 2021, 10:02 AM · Restricted Project
bsmith committed rG48f5a392cb73: [IR] Add vscale_range IR function attribute (authored by bsmith).
[IR] Add vscale_range IR function attribute
Mar 22 2021, 5:05 AM
bsmith closed D98030: [IR] Add vscale_range IR function attribute.
Mar 22 2021, 5:05 AM · Restricted Project, Restricted Project

Mar 19 2021

bsmith requested review of D98939: [SelectionDAG][AArch64][SVE] Perform SETCC condition legalization in LegalizeVectorOps.
Mar 19 2021, 4:31 AM · Restricted Project

Mar 17 2021

bsmith committed rGcf0da91ba5e1: [AArch64][SVE/NEON] Add support for FROUNDEVEN for both NEON and fixed length… (authored by bsmith).
[AArch64][SVE/NEON] Add support for FROUNDEVEN for both NEON and fixed length…
Mar 17 2021, 4:42 AM
bsmith closed D98487: [AArch64][SVE/NEON] Add support for FROUNDEVEN for both NEON and fixed length SVE.
Mar 17 2021, 4:41 AM · Restricted Project, Restricted Project

Mar 16 2021

bsmith updated the diff for D98487: [AArch64][SVE/NEON] Add support for FROUNDEVEN for both NEON and fixed length SVE.
  • Remove SDTFPRoundEvenOp as it's not a correct mirror of SDTFPRoundOp since that is not for ISD::FROUND.
  • Fix comments in include/llvm/Target/TargetSelectionDAG.td for SDTFPRoundOp and SDTFPExtendOp.
Mar 16 2021, 5:25 AM · Restricted Project, Restricted Project

Mar 15 2021

bsmith updated the diff for D98030: [IR] Add vscale_range IR function attribute.
  • Prevent vscale_range(0,0) from crashing and instead don't add the attribute
  • Improve CHECK lines in arm-sve-vector-bits-vscale-range.c test
  • Test vscale_range(0,0) case and move some tests around
Mar 15 2021, 9:30 AM · Restricted Project, Restricted Project
bsmith added inline comments to D98487: [AArch64][SVE/NEON] Add support for FROUNDEVEN for both NEON and fixed length SVE.
Mar 15 2021, 8:37 AM · Restricted Project, Restricted Project
bsmith updated the diff for D98487: [AArch64][SVE/NEON] Add support for FROUNDEVEN for both NEON and fixed length SVE.
  • Add AutoUpgrade code to convert aarch64.neon.frintn to roundeven
  • Add test for above AutoUpgrade
Mar 15 2021, 6:02 AM · Restricted Project, Restricted Project
bsmith committed rGd09ae9328f67: [AArch64][SVE] Add unpredicated ld1/st1 patterns for reg+reg addressing modes (authored by bsmith).
[AArch64][SVE] Add unpredicated ld1/st1 patterns for reg+reg addressing modes
Mar 15 2021, 5:37 AM
bsmith closed D95677: [AArch64][SVE] Add unpredicated ld1/st1 patterns for reg+reg addressing modes.
Mar 15 2021, 5:36 AM · Restricted Project
bsmith updated the diff for D95677: [AArch64][SVE] Add unpredicated ld1/st1 patterns for reg+reg addressing modes.
  • Fix test failure in named-vector-shuffles-sve.ll
Mar 15 2021, 3:48 AM · Restricted Project
bsmith added a comment to D98487: [AArch64][SVE/NEON] Add support for FROUNDEVEN for both NEON and fixed length SVE.

Why is this patch only changing int_aarch64_neon_frintn and not int_aarch64_sve_frintn?
Is there a particular reason to do so?

Mar 15 2021, 3:33 AM · Restricted Project, Restricted Project

Mar 12 2021

bsmith added inline comments to D98030: [IR] Add vscale_range IR function attribute.
Mar 12 2021, 4:10 AM · Restricted Project, Restricted Project
bsmith updated the diff for D98030: [IR] Add vscale_range IR function attribute.
  • State what lack of vscale_range attribute means in LangRef
  • Minor formatting change
Mar 12 2021, 4:10 AM · Restricted Project, Restricted Project
bsmith accepted D98434: [llvm][AArch64][SVE] Lower fixed length vector fabs.

LGTM

Mar 12 2021, 4:01 AM · Restricted Project
bsmith requested review of D98487: [AArch64][SVE/NEON] Add support for FROUNDEVEN for both NEON and fixed length SVE.
Mar 12 2021, 3:58 AM · Restricted Project, Restricted Project

Mar 11 2021

bsmith committed rG860ae9d50c92: [AArch64][SVE] Add fixed/scalable lowering of FMAXIMUM/FMINIMUM ISD nodes (authored by bsmith).
[AArch64][SVE] Add fixed/scalable lowering of FMAXIMUM/FMINIMUM ISD nodes
Mar 11 2021, 5:41 AM
bsmith closed D98348: [AArch64][SVE] Add fixed/scalable lowering of FMAXIMUM/FMINIMUM ISD nodes.
Mar 11 2021, 5:40 AM · Restricted Project
bsmith committed rGea834c8365ca: Revert "[AArch64][SVE] Allow accesses to SVE stack objects to use frame pointer" (authored by bsmith).
Revert "[AArch64][SVE] Allow accesses to SVE stack objects to use frame pointer"
Mar 11 2021, 5:34 AM
bsmith added a reverting change for rG42635856ed3c: [AArch64][SVE] Allow accesses to SVE stack objects to use frame pointer: rGea834c8365ca: Revert "[AArch64][SVE] Allow accesses to SVE stack objects to use frame pointer".
Mar 11 2021, 5:34 AM
bsmith closed D98350: Revert "[AArch64][SVE] Allow accesses to SVE stack objects to use frame pointer".
Mar 11 2021, 5:33 AM · Restricted Project
bsmith updated the diff for D95677: [AArch64][SVE] Add unpredicated ld1/st1 patterns for reg+reg addressing modes.
  • Split reg+reg tests from reg+imm tests
  • Add more reg+reg tests to cover all patterns
Mar 11 2021, 4:13 AM · Restricted Project

Mar 10 2021

bsmith added inline comments to D95677: [AArch64][SVE] Add unpredicated ld1/st1 patterns for reg+reg addressing modes.
Mar 10 2021, 7:26 AM · Restricted Project
bsmith updated the diff for D95677: [AArch64][SVE] Add unpredicated ld1/st1 patterns for reg+reg addressing modes.
  • Simplify new addressing mode tests
Mar 10 2021, 7:25 AM · Restricted Project
bsmith added a reverting change for rG42635856ed3c: [AArch64][SVE] Allow accesses to SVE stack objects to use frame pointer: D98350: Revert "[AArch64][SVE] Allow accesses to SVE stack objects to use frame pointer".
Mar 10 2021, 7:10 AM
bsmith requested review of D98350: Revert "[AArch64][SVE] Allow accesses to SVE stack objects to use frame pointer".
Mar 10 2021, 7:10 AM · Restricted Project
bsmith abandoned D97193: [AArch64][SVE] Ensure hasFP has a consistent return value.

Given this approach has other downsides (namely reserving x29 all the time for SVE), I think instead of doing this I'll revert the previous patch this fixes, and wait until we have a better approach for this.

Mar 10 2021, 6:59 AM · Restricted Project
bsmith requested review of D98348: [AArch64][SVE] Add fixed/scalable lowering of FMAXIMUM/FMINIMUM ISD nodes.
Mar 10 2021, 6:55 AM · Restricted Project

Mar 5 2021

bsmith requested review of D98030: [IR] Add vscale_range IR function attribute.
Mar 5 2021, 3:26 AM · Restricted Project, Restricted Project

Mar 3 2021

bsmith updated the diff for D95677: [AArch64][SVE] Add unpredicated ld1/st1 patterns for reg+reg addressing modes.
  • Remove LICM vscale intrinsic changes
  • Update tests to disable LSR to check vscale folding.
Mar 3 2021, 6:50 AM · Restricted Project

Mar 2 2021

bsmith updated the diff for D97193: [AArch64][SVE] Ensure hasFP has a consistent return value.
  • Fix new test that appeared after rebase.
Mar 2 2021, 5:26 AM · Restricted Project

Mar 1 2021

bsmith added inline comments to D97193: [AArch64][SVE] Ensure hasFP has a consistent return value.
Mar 1 2021, 9:05 AM · Restricted Project
bsmith updated the diff for D97193: [AArch64][SVE] Ensure hasFP has a consistent return value.
  • Always explicitly reserve x29 when building for SVE
  • Add comments to indicate the fact that this is a temporary change
  • Add early bailout when not building for SVE
Mar 1 2021, 9:04 AM · Restricted Project

Feb 22 2021

bsmith added a reviewer for D97193: [AArch64][SVE] Ensure hasFP has a consistent return value: ggouaillardet.
Feb 22 2021, 7:16 AM · Restricted Project
bsmith requested review of D97193: [AArch64][SVE] Ensure hasFP has a consistent return value.
Feb 22 2021, 7:15 AM · Restricted Project

Feb 18 2021

bsmith added a comment to D96424: [AArch64][SVE] Add patterns to generate FMLA/FMLS/FNMLA/FNMLS/FMAD.

Given the extensive testing within sve-fp-combine.ll I do wonder why sve-fma-dagcombine.ll is required, but otherwise the patch looks good.

Feb 18 2021, 8:56 AM · Restricted Project
bsmith committed rG8bad8a43c339: [AArch64][SVE] Add patterns to generate FMLA/FMLS/FNMLA/FNMLS/FMAD (authored by bsmith).
[AArch64][SVE] Add patterns to generate FMLA/FMLS/FNMLA/FNMLS/FMAD
Feb 18 2021, 8:56 AM
bsmith closed D96424: [AArch64][SVE] Add patterns to generate FMLA/FMLS/FNMLA/FNMLS/FMAD.
Feb 18 2021, 8:55 AM · Restricted Project
bsmith committed rG5b094bfeb3cc: [AArch64] Allow folding FMUL/FADD into FMA for FP16 types (authored by bsmith).
[AArch64] Allow folding FMUL/FADD into FMA for FP16 types
Feb 18 2021, 8:51 AM
bsmith closed D96599: [AArch64] Allow folding FMUL/FADD into FMA for FP16 types.
Feb 18 2021, 8:51 AM · Restricted Project

Feb 15 2021

bsmith updated the diff for D96424: [AArch64][SVE] Add patterns to generate FMLA/FMLS/FNMLA/FNMLS/FMAD.
  • Reverse order of dependant patches, pulling half tests from D96599
  • Add missing half tests
Feb 15 2021, 8:28 AM · Restricted Project
bsmith updated the diff for D96599: [AArch64] Allow folding FMUL/FADD into FMA for FP16 types.
  • Reverse order of dependant patches, moving all SVE tests into D96424
Feb 15 2021, 8:23 AM · Restricted Project
bsmith updated the diff for D96599: [AArch64] Allow folding FMUL/FADD into FMA for FP16 types.
  • Update tests in line with changes to D96424
Feb 15 2021, 7:48 AM · Restricted Project
bsmith updated the diff for D96424: [AArch64][SVE] Add patterns to generate FMLA/FMLS/FNMLA/FNMLS/FMAD.
  • Regenerate tests with update_llc_test_checks.py
  • Restructure patterns to use a multiclass
  • Remove duplicated patterns
  • Add missing patterns for more f16 types
Feb 15 2021, 7:39 AM · Restricted Project

Feb 12 2021

bsmith requested review of D96599: [AArch64] Allow folding FMUL/FADD into FMA for FP16 types.
Feb 12 2021, 5:51 AM · Restricted Project
bsmith updated the diff for D96424: [AArch64][SVE] Add patterns to generate FMLA/FMLS/FNMLA/FNMLS/FMAD.

Updated patch:

  • Add some more FNMLA patterns that are required when things get combined slightly differently
  • Only return true in generateFMAsInMachineCombiner when the types in question are scalable, and not always when SVE is present.
  • Move generateFMAsInMachineCombiner into AArch64TargetLowering
  • Replace use of -fp-contract=fast in tests with fast flag on instructions
  • Pull out FP16 combining into a separate patch
Feb 12 2021, 5:44 AM · Restricted Project

Feb 11 2021

bsmith added a comment to D96424: [AArch64][SVE] Add patterns to generate FMLA/FMLS/FNMLA/FNMLS/FMAD.

Hello. Would the machine combiner method not be better, if it can reduce critical path length?

Feb 11 2021, 7:25 AM · Restricted Project

Feb 10 2021

bsmith requested review of D96424: [AArch64][SVE] Add patterns to generate FMLA/FMLS/FNMLA/FNMLS/FMAD.
Feb 10 2021, 8:07 AM · Restricted Project

Jan 29 2021

bsmith added reviewers for D95677: [AArch64][SVE] Add unpredicated ld1/st1 patterns for reg+reg addressing modes: paulwalker-arm, joechrisellis, peterwaller-arm.
Jan 29 2021, 5:38 AM · Restricted Project
bsmith requested review of D95677: [AArch64][SVE] Add unpredicated ld1/st1 patterns for reg+reg addressing modes.
Jan 29 2021, 5:36 AM · Restricted Project

Jan 28 2021

bsmith closed D94401: [AArch64][SVE] Allow accesses to SVE stack objects to use frame pointer.

Committed as rG42635856ed3c. (Forgot to add differential tag in commit message).

Jan 28 2021, 4:43 AM · Restricted Project
bsmith committed rG42635856ed3c: [AArch64][SVE] Allow accesses to SVE stack objects to use frame pointer (authored by bsmith).
[AArch64][SVE] Allow accesses to SVE stack objects to use frame pointer
Jan 28 2021, 4:41 AM

Jan 26 2021

bsmith added inline comments to D94401: [AArch64][SVE] Allow accesses to SVE stack objects to use frame pointer.
Jan 26 2021, 6:53 AM · Restricted Project

Jan 21 2021

bsmith updated the diff for D94401: [AArch64][SVE] Allow accesses to SVE stack objects to use frame pointer.
  • Remove duplicate tests
  • Some code cleanups as per review comments
Jan 21 2021, 4:03 AM · Restricted Project

Jan 19 2021

bsmith added inline comments to D94401: [AArch64][SVE] Allow accesses to SVE stack objects to use frame pointer.
Jan 19 2021, 9:06 AM · Restricted Project
bsmith resigned from D94230: [AArch64][SVE] Coalesce ptrue instrinsic calls where possible.

Resigning as reviewer to remove needs changes flag

Jan 19 2021, 6:10 AM · Restricted Project

Jan 11 2021

bsmith requested review of D94401: [AArch64][SVE] Allow accesses to SVE stack objects to use frame pointer.
Jan 11 2021, 4:29 AM · Restricted Project

Jan 7 2021

bsmith added a comment to D94230: [AArch64][SVE] Coalesce ptrue instrinsic calls where possible.

After much discussion I'm actually incorrect in this assertion, as I mistakenly thought that the ptrue's were ending up being passed straight into the load rather than through the existing svbool convertions. That said this case with (%4, %5 and %7 made not redundant) does now produce worse codegen with this pass:

Jan 7 2021, 7:01 AM · Restricted Project
bsmith requested changes to D94230: [AArch64][SVE] Coalesce ptrue instrinsic calls where possible.

I'm not sure this patch is correct as it's not taking into account how the predicates are used, for example in following case your patch replaces the ptrue_b32() predicate of the %5 8 x i16 load with a ptrue_b16(), which changes the behaviour.

Jan 7 2021, 5:22 AM · Restricted Project

Jan 5 2021

bsmith committed rGc73ae747cb0c: [AArch64][SVE] Add optimization to remove redundant ptest instructions (authored by bsmith).
[AArch64][SVE] Add optimization to remove redundant ptest instructions
Jan 5 2021, 7:30 AM
bsmith closed D93292: [AArch64][SVE] Add optimization to remove redundant ptest instructions.
Jan 5 2021, 7:30 AM · Restricted Project

Jan 4 2021

bsmith updated the diff for D93292: [AArch64][SVE] Add optimization to remove redundant ptest instructions.
  • Add comments to brk testcase to note what is being tested
Jan 4 2021, 5:13 AM · Restricted Project

Dec 18 2020

bsmith updated the diff for D93292: [AArch64][SVE] Add optimization to remove redundant ptest instructions.
  • Add missing tests for BRK* cases
  • Improve misleading comment
Dec 18 2020, 3:31 AM · Restricted Project

Dec 16 2020

bsmith committed rGe0b9c5df260c: [CostModel] Add costs for llvm.experimental.vector.{extract,insert} intrinsics (authored by bsmith).
[CostModel] Add costs for llvm.experimental.vector.{extract,insert} intrinsics
Dec 16 2020, 5:40 AM
bsmith closed D93043: [CostModel] Add costs for llvm.experimental.vector.{extract,insert} intrinsics.
Dec 16 2020, 5:40 AM · Restricted Project