Skip to content

[PowerPC] Change default for auto gen stxvp for cpu=future #142826

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Jun 9, 2025
Merged

[PowerPC] Change default for auto gen stxvp for cpu=future #142826

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
5356180
[PowerPC] For cpu=future allow auto generation of stxvp instructions …
lei137 Jun 4, 2025
d2dfec6
address review comment
lei137 Jun 4, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 2 additions & 1 deletion llvm/lib/Target/PowerPC/PPC.td
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -482,7 +482,8 @@ def ProcessorFeatures {
// Future
// For future CPU we assume that all of the existing features from Power11
// still exist with the exception of those we know are Power11 specific.
list<SubtargetFeature> FutureAdditionalFeatures = [FeatureISAFuture];
list<SubtargetFeature> FutureAdditionalFeatures = [DirectivePwrFuture,
FeatureISAFuture];
list<SubtargetFeature> FutureSpecificFeatures = [];
list<SubtargetFeature> FutureInheritableFeatures =
!listconcat(P11InheritableFeatures, FutureAdditionalFeatures);
Expand Down
13 changes: 9 additions & 4 deletions llvm/lib/Target/PowerPC/PPCISelLowering.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1476,7 +1476,8 @@ PPCTargetLowering::PPCTargetLowering(const PPCTargetMachine &TM,

setMinFunctionAlignment(Align(4));

switch (Subtarget.getCPUDirective()) {
auto CPUDirective = Subtarget.getCPUDirective();
switch (CPUDirective) {
default: break;
case PPC::DIR_970:
case PPC::DIR_A2:
Expand Down Expand Up @@ -1508,15 +1509,14 @@ PPCTargetLowering::PPCTargetLowering(const PPCTargetMachine &TM,

// The Freescale cores do better with aggressive inlining of memcpy and
// friends. GCC uses same threshold of 128 bytes (= 32 word stores).
if (Subtarget.getCPUDirective() == PPC::DIR_E500mc ||
Subtarget.getCPUDirective() == PPC::DIR_E5500) {
if (CPUDirective == PPC::DIR_E500mc || CPUDirective == PPC::DIR_E5500) {
MaxStoresPerMemset = 32;
MaxStoresPerMemsetOptSize = 16;
MaxStoresPerMemcpy = 32;
MaxStoresPerMemcpyOptSize = 8;
MaxStoresPerMemmove = 32;
MaxStoresPerMemmoveOptSize = 8;
} else if (Subtarget.getCPUDirective() == PPC::DIR_A2) {
} else if (CPUDirective == PPC::DIR_A2) {
// The A2 also benefits from (very) aggressive inlining of memcpy and
// friends. The overhead of a the function call, even when warm, can be
// over one hundred cycles.
Expand All @@ -1529,6 +1529,11 @@ PPCTargetLowering::PPCTargetLowering(const PPCTargetMachine &TM,
MaxLoadsPerMemcmpOptSize = 4;
}

// Enable generation of STXVP instructions by default for mcpu=future.
if (CPUDirective == PPC::DIR_PWR_FUTURE &&
DisableAutoPairedVecSt.getNumOccurrences() == 0)
DisableAutoPairedVecSt = false;

IsStrictFPEnabled = true;

// Let the subtarget (CPU) decide if a predictable select is more expensive
Expand Down
6 changes: 3 additions & 3 deletions llvm/test/CodeGen/PowerPC/dmr-spill.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,12 +1,12 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu \
; RUN: -disable-auto-paired-vec-st=false -ppc-asm-full-reg-names \
; RUN: -ppc-asm-full-reg-names \
; RUN: -ppc-vsr-nums-as-vr -mcpu=future < %s | FileCheck %s
; RUN: llc -verify-machineinstrs -mtriple=powerpc64-unknown-aix \
; RUN: -disable-auto-paired-vec-st=false -ppc-asm-full-reg-names \
; RUN: -ppc-asm-full-reg-names \
; RUN: -ppc-vsr-nums-as-vr -mcpu=future < %s | FileCheck %s --check-prefix=AIX
; RUN: llc -verify-machineinstrs -mtriple=powerpc-unknown-aix \
; RUN: -disable-auto-paired-vec-st=false -ppc-asm-full-reg-names \
; RUN: -ppc-asm-full-reg-names \
; RUN: -ppc-vsr-nums-as-vr -mcpu=future < %s | FileCheck %s --check-prefix=AIX32

declare <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2pp(<1024 x i1>, <256 x i1>, <16 x i8>)
Expand Down
2 changes: 0 additions & 2 deletions llvm/test/CodeGen/PowerPC/mmaplus-acc-spill.ll
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,11 +1,9 @@
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; This test is a copy of mma-acc-spill.ll except that it uses mcpu=future.
; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu \
; RUN: -disable-auto-paired-vec-st=false \
; RUN: -mcpu=future -ppc-asm-full-reg-names \
; RUN: -ppc-vsr-nums-as-vr < %s | FileCheck %s
; RUN: llc -verify-machineinstrs -mtriple=powerpc64-unknown-linux-gnu \
; RUN: -disable-auto-paired-vec-st=false \
; RUN: -mcpu=future -ppc-asm-full-reg-names \
; RUN: -ppc-vsr-nums-as-vr < %s | FileCheck %s --check-prefix=CHECK-BE

Expand Down
Loading