Changes in X86InterleavedAccess.cpp

9d09d204 - Reapply "[X86] Limit X86InterleavedAccessGroup to handle the same type case only"

Wed, 19 May 2021 10:01:11 +0000

Reapply "[X86] Limit X86InterleavedAccessGroup to handle the same type case only"The current implementation assumes the destination type of shuffle is the same as the decomposed ones. Add the check to avoid crush when the condition is not satisfied.This fixes PR37616.Reviewed By: RKSimonDifferential Revision: https://reviews.llvm.org/D102751 List of files: /llvm-project-15.0.7/llvm/lib/Target/X86/X86InterleavedAccess.cpp

66513e2f - Revert "[X86] Limit X86InterleavedAccessGroup to handle the same type case only"

Wed, 19 May 2021 12:34:47 +0000

Revert "[X86] Limit X86InterleavedAccessGroup to handle the same type case only"This reverts commit ca23a38e373142a18ab56700ba4f3b947bfe9db0.Revert due to EXPENSIVE_CHECKS fail. List of files: /llvm-project-15.0.7/llvm/lib/Target/X86/X86InterleavedAccess.cpp

ca23a38e - [X86] Limit X86InterleavedAccessGroup to handle the same type case only

Wed, 19 May 2021 10:01:11 +0000

[X86] Limit X86InterleavedAccessGroup to handle the same type case onlyThe current implementation assumes the destination type of shuffle is the same as the decomposed ones. Add the check to avoid crush when the condition is not satisfied.This fixes PR37616.Reviewed By: RKSimonDifferential Revision: https://reviews.llvm.org/D102751 List of files: /llvm-project-15.0.7/llvm/lib/Target/X86/X86InterleavedAccess.cpp

49c2d703 - [X86] Make deinterleave8bitStride3 use unary CreateShuffleVector

Sun, 03 Jan 2021 16:09:48 +0000

[X86] Make deinterleave8bitStride3 use unary CreateShuffleVectorThis patch makes X86InterleavedAccessGroup::deinterleave8bitStride3 use the unary CreateShuffleVector.This is a continuation of D93923. There were a few missing replacements.IIUC, this patch does not cause change in the generated programs' semantics because thefunction inserts shufflevectors that only choose elements from the first vector.Reviewed By: nikicDifferential Revision: https://reviews.llvm.org/D93993 List of files: /llvm-project-15.0.7/llvm/lib/Target/X86/X86InterleavedAccess.cpp

9b296102 - Use unary CreateShuffleVector if possible

Tue, 29 Dec 2020 22:28:17 +0000

Use unary CreateShuffleVector if possibleAs mentioned in D93793, there are quite a few places where unary `IRBuilder::CreateShuffleVector(X, Mask)` can be usedinstead of `IRBuilder::CreateShuffleVector(X, Undef, Mask)`.Let's update them.Actually, it would have been more natural if the patches were made in this order:(1) let them use unary CreateShuffleVector first(2) update IRBuilder::CreateShuffleVector to use poison as a placeholder value (D93793)The order is swapped, but in terms of correctness it is still fine.Reviewed By: spatelDifferential Revision: https://reviews.llvm.org/D93923 List of files: /llvm-project-15.0.7/llvm/lib/Target/X86/X86InterleavedAccess.cpp

c5ba0d33 - [SVE] Make ElementCount and TypeSize use a new PolySize class

Wed, 23 Sep 2020 13:05:15 +0000

[SVE] Make ElementCount and TypeSize use a new PolySize classI have introduced a new template PolySize class, where the templateparameter determines the type of quantity, i.e. for an elementcount this is just an unsigned value. The ElementCount class isnow just a simple derivation of PolySize, whereas TypeSizeis more complicated because it still needs to contain the uint64_tcast operator, since there are still many places in the code thatrely upon this implicit cast. As such the class also still needssome of it's own operators.I've tried to minimise the amount of code in the base PolySizeclass, which led to a couple of changes:1. In some places we were relying on '==' operator comparisonsbetween ElementCounts and the scalar value 1. I didn't put thisoperator in the new PolySize class, and thought it was actuallyclearer to use the isScalar() function instead.2. I removed the isByteSized function and replaced it with callsto isKnownMultipleOf(8).I've also renamed NextPowerOf2 to be coefficientNextPowerOf2 sothat it's more consistent with coefficientDivideBy.Differential Revision: https://reviews.llvm.org/D88409 List of files: /llvm-project-15.0.7/llvm/lib/Target/X86/X86InterleavedAccess.cpp

0da1e7eb - [SVE] Remove calls to VectorType::getNumElements from X86

Mon, 29 Jun 2020 17:30:43 +0000

[SVE] Remove calls to VectorType::getNumElements from X86Reviewers: efriedma, RKSimon, craig.topper, fpetrogalli, c-rhodesReviewed By: RKSimonSubscribers: tschuett, hiraditya, rkruppe, psnobl, llvm-commitsTags: #llvmDifferential Revision: https://reviews.llvm.org/D82508 List of files: /llvm-project-15.0.7/llvm/lib/Target/X86/X86InterleavedAccess.cpp

5a99ec10 - [SVE] Eliminate calls to default-false VectorType::get() from X86

Fri, 29 May 2020 22:52:33 +0000

[SVE] Eliminate calls to default-false VectorType::get() from X86Reviewers: efriedma, sdesmalen, c-rhodes, craig.topperReviewed By: craig.topperSubscribers: tschuett, hiraditya, rkruppe, psnobl, llvm-commitsTags: #llvmDifferential Revision: https://reviews.llvm.org/D80331 List of files: /llvm-project-15.0.7/llvm/lib/Target/X86/X86InterleavedAccess.cpp

5b84ee4f - [Alignment] Fix misaligned interleaved loads

Wed, 20 May 2020 07:37:15 +0000

[Alignment] Fix misaligned interleaved loadsSummary: Tentatively fixing https://bugs.llvm.org/show_bug.cgi?id=45957Reviewers: craig.topper, nlopesSubscribers: hiraditya, llvm-commits, RKSimon, jdoerfert, efriedmaTags: #llvmDifferential Revision: https://reviews.llvm.org/D80276 List of files: /llvm-project-15.0.7/llvm/lib/Target/X86/X86InterleavedAccess.cpp

dd24fb38 - Clean up usages of asserting vector getters in Type

Fri, 17 Apr 2020 17:39:33 +0000

Clean up usages of asserting vector getters in TypeSummary:Remove usages of asserting vector getters in Type in preparation for theVectorType refactor. The existence of these functions complicates therefactor while adding little value.Reviewers: craig.topper, sdesmalen, efriedma, RKSimonReviewed By: efriedmaSubscribers: hiraditya, llvm-commitsTags: #llvmDifferential Revision: https://reviews.llvm.org/D77264 List of files: /llvm-project-15.0.7/llvm/lib/Target/X86/X86InterleavedAccess.cpp

166467e8 - [VectorUtils] Create shufflevector masks as int vectors instead of Constants

Fri, 17 Apr 2020 13:28:00 +0000

[VectorUtils] Create shufflevector masks as int vectors instead of ConstantsNo functionality change intended. List of files: /llvm-project-15.0.7/llvm/lib/Target/X86/X86InterleavedAccess.cpp

316b49d3 - Pass shufflevector indices as int instead of unsigned.

Wed, 15 Apr 2020 13:52:26 +0000

Pass shufflevector indices as int instead of unsigned.No functionality change intended. List of files: /llvm-project-15.0.7/llvm/lib/Target/X86/X86InterleavedAccess.cpp

1318ddbc - [VectorUtils] rename scaleShuffleMask to narrowShuffleMaskElts; NFC

Sat, 11 Apr 2020 14:05:49 +0000

[VectorUtils] rename scaleShuffleMask to narrowShuffleMaskElts; NFCAs proposed in D77881, we'll have the related widening operation,so this name becomes too vague.While here, change the function signature to take an 'int' ratherthan 'size_t' for the scaling factor, add an assert for overflow of32-bits, and improve the documentation comments. List of files: /llvm-project-15.0.7/llvm/lib/Target/X86/X86InterleavedAccess.cpp

f92563f9 - [VectorUtils][X86] De-templatize scaleShuffleMask and 2 X86 shuffle mask helpers and move their implementation to cpp files

Wed, 01 Apr 2020 06:15:10 +0000

[VectorUtils][X86] De-templatize scaleShuffleMask and 2 X86 shuffle mask helpers and move their implementation to cpp filesSummary: These were templated due to SelectionDAG using int masks for shuffles and IR using unsigned masks for shuffles. But now that D72467 has landed we have an int mask version of IRBuilder::CreateShuffleVector. So just use int instead of a templateReviewers: spatel, efriedma, RKSimonReviewed By: efriedmaSubscribers: hiraditya, llvm-commitsDifferential Revision: https://reviews.llvm.org/D77183 List of files: /llvm-project-15.0.7/llvm/lib/Target/X86/X86InterleavedAccess.cpp

7c362b25 - [IRBuilder] Fix unnecessary IRBuilder copies; NFC

Sun, 16 Feb 2020 16:57:18 +0000

[IRBuilder] Fix unnecessary IRBuilder copies; NFCFix a few cases where an IRBuilder is passed to a helper functionby value, while a by reference pass was intended. List of files: /llvm-project-15.0.7/llvm/lib/Target/X86/X86InterleavedAccess.cpp

59f95222 - [Alignment][NFC] Use Align with CreateAlignedStore

Thu, 23 Jan 2020 15:18:34 +0000

[Alignment][NFC] Use Align with CreateAlignedStoreSummary:This is patch is part of a series to introduce an Alignment type.See this thread for context: http://lists.llvm.org/pipermail/llvm-dev/2019-July/133851.htmlSee this patch for the introduction of the type: https://reviews.llvm.org/D64790Reviewers: courbet, bolluSubscribers: arsenm, jvesely, nhaehnle, hiraditya, kerbowa, cfe-commits, llvm-commitsTags: #clang, #llvmDifferential Revision: https://reviews.llvm.org/D73274 List of files: /llvm-project-15.0.7/llvm/lib/Target/X86/X86InterleavedAccess.cpp

279fa8e0 - [Alignement][NFC] Deprecate untyped CreateAlignedLoad

Thu, 23 Jan 2020 10:33:12 +0000

[Alignement][NFC] Deprecate untyped CreateAlignedLoadSummary:This is patch is part of a series to introduce an Alignment type.See this thread for context: http://lists.llvm.org/pipermail/llvm-dev/2019-July/133851.htmlSee this patch for the introduction of the type: https://reviews.llvm.org/D64790Reviewers: courbetSubscribers: arsenm, jvesely, nhaehnle, hiraditya, kerbowa, llvm-commitsTags: #llvmDifferential Revision: https://reviews.llvm.org/D73260 List of files: /llvm-project-15.0.7/llvm/lib/Target/X86/X86InterleavedAccess.cpp

9f4a4d39 - fix typo: "\t" => " "

Sun, 31 Mar 2019 14:49:00 +0000

fix typo: "\t" => " "Reviewers: llvm.org, JimReviewed By: JimSubscribers: arsenm, jvesely, nhaehnle, rupprecht, llvm-commitsTags: #llvmDifferential Revision: https://reviews.llvm.org/D59983llvm-svn: 357365 List of files: /llvm-project-15.0.7/llvm/lib/Target/X86/X86InterleavedAccess.cpp

7716075a - [opaque pointer types] Pass value type to GetElementPtr creation.

Fri, 01 Feb 2019 20:44:47 +0000

[opaque pointer types] Pass value type to GetElementPtr creation.This cleans up all GetElementPtr creation in LLVM to explicitly pass avalue type rather than deriving it from the pointer's element-type.Differential Revision: https://reviews.llvm.org/D57173llvm-svn: 352913 List of files: /llvm-project-15.0.7/llvm/lib/Target/X86/X86InterleavedAccess.cpp

14359ef1 - [opaque pointer types] Pass value type to LoadInst creation.

Fri, 01 Feb 2019 20:44:24 +0000

[opaque pointer types] Pass value type to LoadInst creation.This cleans up all LoadInst creation in LLVM to explicitly pass thevalue type rather than deriving it from the pointer's element-type.Differential Revision: https://reviews.llvm.org/D57172llvm-svn: 352911 List of files: /llvm-project-15.0.7/llvm/lib/Target/X86/X86InterleavedAccess.cpp