7 Commits

Author SHA1 Message Date
Alexey Bataev
6b85fb1ef8
[SLP]Consider (f)sub, being operand of llvm.(f)abs/icmp eq/ne 0, commutative.
If (f)sub is only operand of llvm.(f)abs or icmp eq/ne 0 (int only), we can consider it as commutative operation, just need to drop wrapping flags for ineteger
operation.

https://alive2.llvm.org/ce/z/GxvxjB for correctness of abs with dropped
flags.

Reviewers: RKSimon

Reviewed By: RKSimon

Pull Request: https://github.com/llvm/llvm-project/pull/86196
2024-04-11 14:56:39 -04:00
Alexey Bataev
f7cc224044 Revert "[libclc] Refactor build system to allow in-tree builds (#87622)"
This reverts commit 9029e6ebdfd98a58b5c5646fd534c6c849148cda, which was
committed by mistake with the wrong message and fails  https://lab.llvm.org/buildbot/#/builders/221/builds/21958.
2024-04-11 11:08:47 -07:00
Fraser Cormack
9029e6ebdf [libclc] Refactor build system to allow in-tree builds (#87622)
The previous build system was adding custom "OpenCL" and "LLVM IR"
languages in CMake to build the builtin libraries. This was making it
harder to build in-tree because the tool binaries needed to be present
at configure time.

This commit refactors the build system to use custom commands to build
the bytecode files one by one, and link them all together into the final
bytecode library. It also enables in-tree builds by aliasing the
clang/llvm-link/etc. tool targets to internal targets, which are
imported from the LLVM installation directory when building out of tree.

Diffing (with llvm-diff) all of the final bytecode libraries in an
out-of-tree configuration against those built using the current tip
system shows no changes. Note that there are textual changes to metadata
IDs which confuse regular diff, and that llvm-diff 14 and below may show
false-positives.

This commit also removes a file listed in one of the SOURCEs which
didn't exist and which was preventing the use of
ENABLE_RUNTIME_SUBNORMAL when configuring CMake.
2024-04-11 10:56:45 -07:00
Alexey Bataev
4a1c53f9fa [SLP]Improve minbitwidth analysis for abs/smin/smax/umin/umax intrinsics.
https://alive2.llvm.org/ce/z/ivPZ26 for the abs transformations.

Reviewers: RKSimon

Reviewed By: RKSimon

Pull Request: https://github.com/llvm/llvm-project/pull/86135
2024-04-08 08:32:35 -07:00
Martin Storsjö
bd9486b4ec Revert "[SLP]Improve minbitwidth analysis for abs/smin/smax/umin/umax intrinsics."
This reverts commit 66b528078e4852412769375e35d2a672bf36a0ec.

This commit caused miscompilations, breaking tests in the libyuv
testsuite - see
https://github.com/llvm/llvm-project/pull/86135#issuecomment-2041049709
for more details.
2024-04-06 23:53:26 +03:00
Alexey Bataev
66b528078e
[SLP]Improve minbitwidth analysis for abs/smin/smax/umin/umax intrinsics.
https://alive2.llvm.org/ce/z/ivPZ26 for the abs transformations.

Reviewers: RKSimon

Reviewed By: RKSimon

Pull Request: https://github.com/llvm/llvm-project/pull/86135
2024-04-05 14:29:26 -04:00
Alexey Bataev
aa4cbaba1d [SLP][NFC]Add a test with @llvm.abs nodes, which can be analyzed for
better bitwidth.
2024-03-21 06:16:18 -07:00