Skip to content

[TLI] Add SLEEFGNUABI mappings for fmod/fmodf fixed-width. #75803

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

Jump to bottom
Merged
merged 1 commit into from
Dec 20, 2023
Merged

[TLI] Add SLEEFGNUABI mappings for fmod/fmodf fixed-width. #75803

merged 1 commit into from
Dec 20, 2023
+51 −142

Conversation

paschalis-mpeis
Copy link
Member

Cleanup test sleef-calls-aarch64.ll:

  • make the util update script's regex more clear
  • eliminate scalar epilogues in tests

@llvmbot llvmbot added llvm:analysis Includes value tracking, cost tables and constant folding llvm:transforms labels Dec 18, 2023
@llvmbot
Copy link
Member

llvmbot commented Dec 18, 2023 •
edited
Loading

@llvm/pr-subscribers-llvm-analysis

@llvm/pr-subscribers-llvm-transforms

Author: Paschalis Mpeis (paschalis-mpeis)

Changes

Cleanup test sleef-calls-aarch64.ll:

  • make the util update script's regex more clear
  • eliminate scalar epilogues in tests

Patch is 46.92 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/75803.diff

2 Files Affected:

  • (modified) llvm/include/llvm/Analysis/VecFuncs.def (+2)
  • (modified) llvm/test/Transforms/LoopVectorize/AArch64/sleef-calls-aarch64.ll (+49-142)
diff --git a/llvm/include/llvm/Analysis/VecFuncs.def b/llvm/include/llvm/Analysis/VecFuncs.def
index c628e72b24d12b..4bffcdee6f9caa 100644
--- a/llvm/include/llvm/Analysis/VecFuncs.def
+++ b/llvm/include/llvm/Analysis/VecFuncs.def
@@ -493,6 +493,7 @@ TLI_DEFINE_VECFUNC( "llvm.exp2.f64", "_ZGVnN2v_exp2", FIXED(2), "_ZGV_LLVM_N2v")
 
 TLI_DEFINE_VECFUNC( "exp10", "_ZGVnN2v_exp10", FIXED(2), "_ZGV_LLVM_N2v")
 TLI_DEFINE_VECFUNC( "llvm.exp10.f64", "_ZGVnN2v_exp10", FIXED(2), "_ZGV_LLVM_N2v")
+TLI_DEFINE_VECFUNC("fmod", "_ZGVnN2vv_fmod", FIXED(2), "_ZGV_LLVM_N2vv")
 
 TLI_DEFINE_VECFUNC( "lgamma", "_ZGVnN2v_lgamma", FIXED(2), "_ZGV_LLVM_N2v")
 
@@ -546,6 +547,7 @@ TLI_DEFINE_VECFUNC( "llvm.exp2.f32", "_ZGVnN4v_exp2f", FIXED(4), "_ZGV_LLVM_N4v"
 
 TLI_DEFINE_VECFUNC( "exp10f", "_ZGVnN4v_exp10f", FIXED(4), "_ZGV_LLVM_N4v")
 TLI_DEFINE_VECFUNC( "llvm.exp10.f32", "_ZGVnN4v_exp10f", FIXED(4), "_ZGV_LLVM_N4v")
+TLI_DEFINE_VECFUNC("fmodf", "_ZGVnN4vv_fmodf", FIXED(4), "_ZGV_LLVM_N4vv")
 
 TLI_DEFINE_VECFUNC( "lgammaf", "_ZGVnN4v_lgammaf", FIXED(4), "_ZGV_LLVM_N4v")
 
diff --git a/llvm/test/Transforms/LoopVectorize/AArch64/sleef-calls-aarch64.ll b/llvm/test/Transforms/LoopVectorize/AArch64/sleef-calls-aarch64.ll
index d7dc122edaf7e2..bd39dcb3371a92 100644
--- a/llvm/test/Transforms/LoopVectorize/AArch64/sleef-calls-aarch64.ll
+++ b/llvm/test/Transforms/LoopVectorize/AArch64/sleef-calls-aarch64.ll
@@ -1,8 +1,7 @@
-; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --filter "(_)|(cos|exp[^e]|fmod|gamma|log|pow|sin|sqrt|tan)|(ret)" --version 2
-; RUN: opt -vector-library=sleefgnuabi -passes=inject-tli-mappings,loop-vectorize -force-vector-interleave=1 -S < %s | FileCheck %s --check-prefix=NEON
-; RUN: opt -mattr=+sve -vector-library=sleefgnuabi -passes=inject-tli-mappings,loop-vectorize -force-vector-interleave=1 -S < %s | FileCheck %s --check-prefix=SVE
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --filter "(_)|(cos|expf?\(|exp2|exp10|fmod|gamma|log|pow|sin|sqrt|tan)|(ret)" --version 2
+; RUN: opt -vector-library=sleefgnuabi -passes=inject-tli-mappings,loop-vectorize,simplifycfg -prefer-predicate-over-epilogue=predicate-dont-vectorize -force-vector-interleave=1 -S < %s | FileCheck %s --check-prefix=NEON
+; RUN: opt -mattr=+sve -vector-library=sleefgnuabi -passes=inject-tli-mappings,loop-vectorize,simplifycfg -prefer-predicate-over-epilogue=predicate-dont-vectorize -force-vector-interleave=1 -S < %s | FileCheck %s --check-prefix=SVE
 
-target datalayout = "e-m:e-i8:8:32-i16:16:32-i64:64-i128:128-n32:64-S128"
 target triple = "aarch64-unknown-linux-gnu"
 
 declare double @acos(double)
@@ -12,13 +11,11 @@ define void @acos_f64(double* nocapture %varray) {
 ; NEON-LABEL: define void @acos_f64
 ; NEON-SAME: (ptr nocapture [[VARRAY:%.*]]) {
 ; NEON:    [[TMP2:%.*]] = call <2 x double> @_ZGVnN2v_acos(<2 x double> [[TMP1:%.*]])
-; NEON:    [[CALL:%.*]] = tail call double @acos(double [[CONV:%.*]]) #[[ATTR0:[0-9]+]]
 ; NEON:    ret void
 ;
 ; SVE-LABEL: define void @acos_f64
 ; SVE-SAME: (ptr nocapture [[VARRAY:%.*]]) #[[ATTR0:[0-9]+]] {
-; SVE:    [[TMP12:%.*]] = call <vscale x 2 x double> @_ZGVsMxv_acos(<vscale x 2 x double> [[TMP11:%.*]], <vscale x 2 x i1> shufflevector (<vscale x 2 x i1> insertelement (<vscale x 2 x i1> poison, i1 true, i64 0), <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer))
-; SVE:    [[CALL:%.*]] = tail call double @acos(double [[CONV:%.*]]) #[[ATTR2:[0-9]+]]
+; SVE:    [[TMP20:%.*]] = call <vscale x 2 x double> @_ZGVsMxv_acos(<vscale x 2 x double> [[TMP19:%.*]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK:%.*]])
 ; SVE:    ret void
 ;
   entry:
@@ -43,13 +40,11 @@ define void @acos_f32(float* nocapture %varray) {
 ; NEON-LABEL: define void @acos_f32
 ; NEON-SAME: (ptr nocapture [[VARRAY:%.*]]) {
 ; NEON:    [[TMP2:%.*]] = call <4 x float> @_ZGVnN4v_acosf(<4 x float> [[TMP1:%.*]])
-; NEON:    [[CALL:%.*]] = tail call float @acosf(float [[CONV:%.*]]) #[[ATTR1:[0-9]+]]
 ; NEON:    ret void
 ;
 ; SVE-LABEL: define void @acos_f32
 ; SVE-SAME: (ptr nocapture [[VARRAY:%.*]]) #[[ATTR0]] {
-; SVE:    [[TMP12:%.*]] = call <vscale x 4 x float> @_ZGVsMxv_acosf(<vscale x 4 x float> [[TMP11:%.*]], <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
-; SVE:    [[CALL:%.*]] = tail call float @acosf(float [[CONV:%.*]]) #[[ATTR3:[0-9]+]]
+; SVE:    [[TMP20:%.*]] = call <vscale x 4 x float> @_ZGVsMxv_acosf(<vscale x 4 x float> [[TMP19:%.*]], <vscale x 4 x i1> [[ACTIVE_LANE_MASK:%.*]])
 ; SVE:    ret void
 ;
   entry:
@@ -77,13 +72,11 @@ define void @asin_f64(double* nocapture %varray) {
 ; NEON-LABEL: define void @asin_f64
 ; NEON-SAME: (ptr nocapture [[VARRAY:%.*]]) {
 ; NEON:    [[TMP2:%.*]] = call <2 x double> @_ZGVnN2v_asin(<2 x double> [[TMP1:%.*]])
-; NEON:    [[CALL:%.*]] = tail call double @asin(double [[CONV:%.*]]) #[[ATTR2:[0-9]+]]
 ; NEON:    ret void
 ;
 ; SVE-LABEL: define void @asin_f64
 ; SVE-SAME: (ptr nocapture [[VARRAY:%.*]]) #[[ATTR0]] {
-; SVE:    [[TMP12:%.*]] = call <vscale x 2 x double> @_ZGVsMxv_asin(<vscale x 2 x double> [[TMP11:%.*]], <vscale x 2 x i1> shufflevector (<vscale x 2 x i1> insertelement (<vscale x 2 x i1> poison, i1 true, i64 0), <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer))
-; SVE:    [[CALL:%.*]] = tail call double @asin(double [[CONV:%.*]]) #[[ATTR4:[0-9]+]]
+; SVE:    [[TMP20:%.*]] = call <vscale x 2 x double> @_ZGVsMxv_asin(<vscale x 2 x double> [[TMP19:%.*]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK:%.*]])
 ; SVE:    ret void
 ;
   entry:
@@ -108,13 +101,11 @@ define void @asin_f32(float* nocapture %varray) {
 ; NEON-LABEL: define void @asin_f32
 ; NEON-SAME: (ptr nocapture [[VARRAY:%.*]]) {
 ; NEON:    [[TMP2:%.*]] = call <4 x float> @_ZGVnN4v_asinf(<4 x float> [[TMP1:%.*]])
-; NEON:    [[CALL:%.*]] = tail call float @asinf(float [[CONV:%.*]]) #[[ATTR3:[0-9]+]]
 ; NEON:    ret void
 ;
 ; SVE-LABEL: define void @asin_f32
 ; SVE-SAME: (ptr nocapture [[VARRAY:%.*]]) #[[ATTR0]] {
-; SVE:    [[TMP12:%.*]] = call <vscale x 4 x float> @_ZGVsMxv_asinf(<vscale x 4 x float> [[TMP11:%.*]], <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
-; SVE:    [[CALL:%.*]] = tail call float @asinf(float [[CONV:%.*]]) #[[ATTR5:[0-9]+]]
+; SVE:    [[TMP20:%.*]] = call <vscale x 4 x float> @_ZGVsMxv_asinf(<vscale x 4 x float> [[TMP19:%.*]], <vscale x 4 x i1> [[ACTIVE_LANE_MASK:%.*]])
 ; SVE:    ret void
 ;
   entry:
@@ -142,13 +133,11 @@ define void @atan_f64(double* nocapture %varray) {
 ; NEON-LABEL: define void @atan_f64
 ; NEON-SAME: (ptr nocapture [[VARRAY:%.*]]) {
 ; NEON:    [[TMP2:%.*]] = call <2 x double> @_ZGVnN2v_atan(<2 x double> [[TMP1:%.*]])
-; NEON:    [[CALL:%.*]] = tail call double @atan(double [[CONV:%.*]]) #[[ATTR4:[0-9]+]]
 ; NEON:    ret void
 ;
 ; SVE-LABEL: define void @atan_f64
 ; SVE-SAME: (ptr nocapture [[VARRAY:%.*]]) #[[ATTR0]] {
-; SVE:    [[TMP12:%.*]] = call <vscale x 2 x double> @_ZGVsMxv_atan(<vscale x 2 x double> [[TMP11:%.*]], <vscale x 2 x i1> shufflevector (<vscale x 2 x i1> insertelement (<vscale x 2 x i1> poison, i1 true, i64 0), <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer))
-; SVE:    [[CALL:%.*]] = tail call double @atan(double [[CONV:%.*]]) #[[ATTR6:[0-9]+]]
+; SVE:    [[TMP20:%.*]] = call <vscale x 2 x double> @_ZGVsMxv_atan(<vscale x 2 x double> [[TMP19:%.*]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK:%.*]])
 ; SVE:    ret void
 ;
   entry:
@@ -173,13 +162,11 @@ define void @atan_f32(float* nocapture %varray) {
 ; NEON-LABEL: define void @atan_f32
 ; NEON-SAME: (ptr nocapture [[VARRAY:%.*]]) {
 ; NEON:    [[TMP2:%.*]] = call <4 x float> @_ZGVnN4v_atanf(<4 x float> [[TMP1:%.*]])
-; NEON:    [[CALL:%.*]] = tail call float @atanf(float [[CONV:%.*]]) #[[ATTR5:[0-9]+]]
 ; NEON:    ret void
 ;
 ; SVE-LABEL: define void @atan_f32
 ; SVE-SAME: (ptr nocapture [[VARRAY:%.*]]) #[[ATTR0]] {
-; SVE:    [[TMP12:%.*]] = call <vscale x 4 x float> @_ZGVsMxv_atanf(<vscale x 4 x float> [[TMP11:%.*]], <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
-; SVE:    [[CALL:%.*]] = tail call float @atanf(float [[CONV:%.*]]) #[[ATTR7:[0-9]+]]
+; SVE:    [[TMP20:%.*]] = call <vscale x 4 x float> @_ZGVsMxv_atanf(<vscale x 4 x float> [[TMP19:%.*]], <vscale x 4 x i1> [[ACTIVE_LANE_MASK:%.*]])
 ; SVE:    ret void
 ;
   entry:
@@ -207,13 +194,11 @@ define void @atan2_f64(double* nocapture %varray) {
 ; NEON-LABEL: define void @atan2_f64
 ; NEON-SAME: (ptr nocapture [[VARRAY:%.*]]) {
 ; NEON:    [[TMP2:%.*]] = call <2 x double> @_ZGVnN2vv_atan2(<2 x double> [[TMP1:%.*]], <2 x double> [[TMP1]])
-; NEON:    [[CALL:%.*]] = tail call double @atan2(double [[CONV:%.*]], double [[CONV]]) #[[ATTR6:[0-9]+]]
 ; NEON:    ret void
 ;
 ; SVE-LABEL: define void @atan2_f64
 ; SVE-SAME: (ptr nocapture [[VARRAY:%.*]]) #[[ATTR0]] {
-; SVE:    [[TMP12:%.*]] = call <vscale x 2 x double> @_ZGVsMxvv_atan2(<vscale x 2 x double> [[TMP11:%.*]], <vscale x 2 x double> [[TMP11]], <vscale x 2 x i1> shufflevector (<vscale x 2 x i1> insertelement (<vscale x 2 x i1> poison, i1 true, i64 0), <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer))
-; SVE:    [[CALL:%.*]] = tail call double @atan2(double [[CONV:%.*]], double [[CONV]]) #[[ATTR8:[0-9]+]]
+; SVE:    [[TMP20:%.*]] = call <vscale x 2 x double> @_ZGVsMxvv_atan2(<vscale x 2 x double> [[TMP19:%.*]], <vscale x 2 x double> [[TMP19]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK:%.*]])
 ; SVE:    ret void
 ;
   entry:
@@ -238,13 +223,11 @@ define void @atan2_f32(float* nocapture %varray) {
 ; NEON-LABEL: define void @atan2_f32
 ; NEON-SAME: (ptr nocapture [[VARRAY:%.*]]) {
 ; NEON:    [[TMP2:%.*]] = call <4 x float> @_ZGVnN4vv_atan2f(<4 x float> [[TMP1:%.*]], <4 x float> [[TMP1]])
-; NEON:    [[CALL:%.*]] = tail call float @atan2f(float [[CONV:%.*]], float [[CONV]]) #[[ATTR7:[0-9]+]]
 ; NEON:    ret void
 ;
 ; SVE-LABEL: define void @atan2_f32
 ; SVE-SAME: (ptr nocapture [[VARRAY:%.*]]) #[[ATTR0]] {
-; SVE:    [[TMP12:%.*]] = call <vscale x 4 x float> @_ZGVsMxvv_atan2f(<vscale x 4 x float> [[TMP11:%.*]], <vscale x 4 x float> [[TMP11]], <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
-; SVE:    [[CALL:%.*]] = tail call float @atan2f(float [[CONV:%.*]], float [[CONV]]) #[[ATTR9:[0-9]+]]
+; SVE:    [[TMP20:%.*]] = call <vscale x 4 x float> @_ZGVsMxvv_atan2f(<vscale x 4 x float> [[TMP19:%.*]], <vscale x 4 x float> [[TMP19]], <vscale x 4 x i1> [[ACTIVE_LANE_MASK:%.*]])
 ; SVE:    ret void
 ;
   entry:
@@ -272,13 +255,11 @@ define void @atanh_f64(double* nocapture %varray) {
 ; NEON-LABEL: define void @atanh_f64
 ; NEON-SAME: (ptr nocapture [[VARRAY:%.*]]) {
 ; NEON:    [[TMP2:%.*]] = call <2 x double> @_ZGVnN2v_atanh(<2 x double> [[TMP1:%.*]])
-; NEON:    [[CALL:%.*]] = tail call double @atanh(double [[CONV:%.*]]) #[[ATTR8:[0-9]+]]
 ; NEON:    ret void
 ;
 ; SVE-LABEL: define void @atanh_f64
 ; SVE-SAME: (ptr nocapture [[VARRAY:%.*]]) #[[ATTR0]] {
-; SVE:    [[TMP12:%.*]] = call <vscale x 2 x double> @_ZGVsMxv_atanh(<vscale x 2 x double> [[TMP11:%.*]], <vscale x 2 x i1> shufflevector (<vscale x 2 x i1> insertelement (<vscale x 2 x i1> poison, i1 true, i64 0), <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer))
-; SVE:    [[CALL:%.*]] = tail call double @atanh(double [[CONV:%.*]]) #[[ATTR10:[0-9]+]]
+; SVE:    [[TMP20:%.*]] = call <vscale x 2 x double> @_ZGVsMxv_atanh(<vscale x 2 x double> [[TMP19:%.*]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK:%.*]])
 ; SVE:    ret void
 ;
   entry:
@@ -303,13 +284,11 @@ define void @atanh_f32(float* nocapture %varray) {
 ; NEON-LABEL: define void @atanh_f32
 ; NEON-SAME: (ptr nocapture [[VARRAY:%.*]]) {
 ; NEON:    [[TMP2:%.*]] = call <4 x float> @_ZGVnN4v_atanhf(<4 x float> [[TMP1:%.*]])
-; NEON:    [[CALL:%.*]] = tail call float @atanhf(float [[CONV:%.*]]) #[[ATTR9:[0-9]+]]
 ; NEON:    ret void
 ;
 ; SVE-LABEL: define void @atanh_f32
 ; SVE-SAME: (ptr nocapture [[VARRAY:%.*]]) #[[ATTR0]] {
-; SVE:    [[TMP12:%.*]] = call <vscale x 4 x float> @_ZGVsMxv_atanhf(<vscale x 4 x float> [[TMP11:%.*]], <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
-; SVE:    [[CALL:%.*]] = tail call float @atanhf(float [[CONV:%.*]]) #[[ATTR11:[0-9]+]]
+; SVE:    [[TMP20:%.*]] = call <vscale x 4 x float> @_ZGVsMxv_atanhf(<vscale x 4 x float> [[TMP19:%.*]], <vscale x 4 x i1> [[ACTIVE_LANE_MASK:%.*]])
 ; SVE:    ret void
 ;
   entry:
@@ -337,13 +316,11 @@ define void @cos_f64(double* nocapture %varray) {
 ; NEON-LABEL: define void @cos_f64
 ; NEON-SAME: (ptr nocapture [[VARRAY:%.*]]) {
 ; NEON:    [[TMP2:%.*]] = call <2 x double> @_ZGVnN2v_cos(<2 x double> [[TMP1:%.*]])
-; NEON:    [[CALL:%.*]] = tail call double @cos(double [[CONV:%.*]]) #[[ATTR10:[0-9]+]]
 ; NEON:    ret void
 ;
 ; SVE-LABEL: define void @cos_f64
 ; SVE-SAME: (ptr nocapture [[VARRAY:%.*]]) #[[ATTR0]] {
-; SVE:    [[TMP12:%.*]] = call <vscale x 2 x double> @_ZGVsMxv_cos(<vscale x 2 x double> [[TMP11:%.*]], <vscale x 2 x i1> shufflevector (<vscale x 2 x i1> insertelement (<vscale x 2 x i1> poison, i1 true, i64 0), <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer))
-; SVE:    [[CALL:%.*]] = tail call double @cos(double [[CONV:%.*]]) #[[ATTR12:[0-9]+]]
+; SVE:    [[TMP20:%.*]] = call <vscale x 2 x double> @_ZGVsMxv_cos(<vscale x 2 x double> [[TMP19:%.*]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK:%.*]])
 ; SVE:    ret void
 ;
   entry:
@@ -368,13 +345,11 @@ define void @cos_f32(float* nocapture %varray) {
 ; NEON-LABEL: define void @cos_f32
 ; NEON-SAME: (ptr nocapture [[VARRAY:%.*]]) {
 ; NEON:    [[TMP2:%.*]] = call <4 x float> @_ZGVnN4v_cosf(<4 x float> [[TMP1:%.*]])
-; NEON:    [[CALL:%.*]] = tail call float @cosf(float [[CONV:%.*]]) #[[ATTR11:[0-9]+]]
 ; NEON:    ret void
 ;
 ; SVE-LABEL: define void @cos_f32
 ; SVE-SAME: (ptr nocapture [[VARRAY:%.*]]) #[[ATTR0]] {
-; SVE:    [[TMP12:%.*]] = call <vscale x 4 x float> @_ZGVsMxv_cosf(<vscale x 4 x float> [[TMP11:%.*]], <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
-; SVE:    [[CALL:%.*]] = tail call float @cosf(float [[CONV:%.*]]) #[[ATTR13:[0-9]+]]
+; SVE:    [[TMP20:%.*]] = call <vscale x 4 x float> @_ZGVsMxv_cosf(<vscale x 4 x float> [[TMP19:%.*]], <vscale x 4 x i1> [[ACTIVE_LANE_MASK:%.*]])
 ; SVE:    ret void
 ;
   entry:
@@ -402,13 +377,11 @@ define void @cosh_f64(double* nocapture %varray) {
 ; NEON-LABEL: define void @cosh_f64
 ; NEON-SAME: (ptr nocapture [[VARRAY:%.*]]) {
 ; NEON:    [[TMP2:%.*]] = call <2 x double> @_ZGVnN2v_cosh(<2 x double> [[TMP1:%.*]])
-; NEON:    [[CALL:%.*]] = tail call double @cosh(double [[CONV:%.*]]) #[[ATTR12:[0-9]+]]
 ; NEON:    ret void
 ;
 ; SVE-LABEL: define void @cosh_f64
 ; SVE-SAME: (ptr nocapture [[VARRAY:%.*]]) #[[ATTR0]] {
-; SVE:    [[TMP12:%.*]] = call <vscale x 2 x double> @_ZGVsMxv_cosh(<vscale x 2 x double> [[TMP11:%.*]], <vscale x 2 x i1> shufflevector (<vscale x 2 x i1> insertelement (<vscale x 2 x i1> poison, i1 true, i64 0), <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer))
-; SVE:    [[CALL:%.*]] = tail call double @cosh(double [[CONV:%.*]]) #[[ATTR14:[0-9]+]]
+; SVE:    [[TMP20:%.*]] = call <vscale x 2 x double> @_ZGVsMxv_cosh(<vscale x 2 x double> [[TMP19:%.*]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK:%.*]])
 ; SVE:    ret void
 ;
   entry:
@@ -433,13 +406,11 @@ define void @cosh_f32(float* nocapture %varray) {
 ; NEON-LABEL: define void @cosh_f32
 ; NEON-SAME: (ptr nocapture [[VARRAY:%.*]]) {
 ; NEON:    [[TMP2:%.*]] = call <4 x float> @_ZGVnN4v_coshf(<4 x float> [[TMP1:%.*]])
-; NEON:    [[CALL:%.*]] = tail call float @coshf(float [[CONV:%.*]]) #[[ATTR13:[0-9]+]]
 ; NEON:    ret void
 ;
 ; SVE-LABEL: define void @cosh_f32
 ; SVE-SAME: (ptr nocapture [[VARRAY:%.*]]) #[[ATTR0]] {
-; SVE:    [[TMP12:%.*]] = call <vscale x 4 x float> @_ZGVsMxv_coshf(<vscale x 4 x float> [[TMP11:%.*]], <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
-; SVE:    [[CALL:%.*]] = tail call float @coshf(float [[CONV:%.*]]) #[[ATTR15:[0-9]+]]
+; SVE:    [[TMP20:%.*]] = call <vscale x 4 x float> @_ZGVsMxv_coshf(<vscale x 4 x float> [[TMP19:%.*]], <vscale x 4 x i1> [[ACTIVE_LANE_MASK:%.*]])
 ; SVE:    ret void
 ;
   entry:
@@ -467,13 +438,11 @@ define void @exp_f64(double* nocapture %varray) {
 ; NEON-LABEL: define void @exp_f64
 ; NEON-SAME: (ptr nocapture [[VARRAY:%.*]]) {
 ; NEON:    [[TMP2:%.*]] = call <2 x double> @_ZGVnN2v_exp(<2 x double> [[TMP1:%.*]])
-; NEON:    [[CALL:%.*]] = tail call double @exp(double [[CONV:%.*]]) #[[ATTR14:[0-9]+]]
 ; NEON:    ret void
 ;
 ; SVE-LABEL: define void @exp_f64
 ; SVE-SAME: (ptr nocapture [[VARRAY:%.*]]) #[[ATTR0]] {
-; SVE:    [[TMP12:%.*]] = call <vscale x 2 x double> @_ZGVsMxv_exp(<vscale x 2 x double> [[TMP11:%.*]], <vscale x 2 x i1> shufflevector (<vscale x 2 x i1> insertelement (<vscale x 2 x i1> poison, i1 true, i64 0), <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer))
-; SVE:    [[CALL:%.*]] = tail call double @exp(double [[CONV:%.*]]) #[[ATTR16:[0-9]+]]
+; SVE:    [[TMP20:%.*]] = call <vscale x 2 x double> @_ZGVsMxv_exp(<vscale x 2 x double> [[TMP19:%.*]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK:%.*]])
 ; SVE:    ret void
 ;
   entry:
@@ -498,13 +467,11 @@ define void @exp_f32(float* nocapture %varray) {
 ; NEON-LABEL: define void @exp_f32
 ; NEON-SAME: (ptr nocapture [[VARRAY:%.*]]) {
 ; NEON:    [[TMP2:%.*]] = call <4 x float> @_ZGVnN4v_expf(<4 x float> [[TMP1:%.*]])
-; NEON:    [[CALL:%.*]] = tail call float @expf(float [[CONV:%.*]]) #[[ATTR15:[0-9]+]]
 ; NEON:    ret void
 ;
 ; SVE-LABEL: define void @exp_f32
 ; SVE-SAME: (ptr nocapture [[VARRAY:%.*]]) #[[ATTR0]] {
-; SVE:    [[TMP12:%.*]] = call <vscale x 4 x float> @_ZGVsMxv_expf(<vscale x 4 x float> [[TMP11:%.*]], <vscale x 4 x i1> shufflevector (<vscale x 4 x i1> insertelement (<vscale x 4 x i1> poison, i1 true, i64 0), <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer))
-; SVE:    [[CALL:%.*]] = tail call float @expf(float [[CONV:%.*]]) #[[ATTR17:[0-9]+]]
+; SVE:    [[TMP20:%.*]] = call <vscale x 4 x float> @_ZGVsMxv_expf(<vscale x 4 x float> [[TMP19:%.*]], <vscale x 4 x i1> [[ACTIVE_LANE_MASK:%.*]])
 ; SVE:    ret void
 ;
   entry:
@@ -532,13 +499,11 @@ define void @exp2_f64(double* nocapture %varray) {
 ; NEON-LABEL: define void @exp2_f64
 ; NEON-SAME: (ptr nocapture [[VARRAY:%.*]]) {
 ; NEON:    [[TMP2:%.*]] = call <2 x double> @_ZGVnN2v_exp2(<2 x double> [[TMP1:%.*]])
-; NEON:    [[CALL:%.*]] = tail call double @exp2(double [[CONV:%.*]]) #[[ATTR16:[0-9]+]]
 ; NEON:    ret void
 ;
 ; SVE-LABEL: define void @exp2_f64
 ; SVE-SAME: (ptr nocapture [[VARRAY:%.*]]) #[[ATTR0]] {
-; SVE:    [[TMP12:%.*]] = call <vscale x 2 x double> @_ZGVsMxv_exp2(<vscale x 2 x double> [[TMP11:%.*]], <vscale x 2 x i1> shufflevector (<vscale x 2 x i1> insertelement (<vscale x 2 x i1> poison, i1 true, i64 0), <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer))
-; SVE:    [[CALL:%.*]] = tail call double @exp2(double [[CONV:%.*]]) #[[ATTR18:[0-9]+]]
+; SVE:    [[TMP20:%.*]] = call <vscale x 2 x double> @_ZGVsMxv_exp2(<vscale x 2 x double> [[TMP19:%.*]], <vscale x 2 x i1> [[ACTIVE_LANE_MASK:%.*]])
 ; SVE:    ret void
 ;
   entry:
@@ -563,13 +528,11 @@ define void @exp2_f32(float* nocapture %varray) {
 ; NEON-LABEL: define void @exp2_f32
 ; NEON-SAME: (ptr nocapture [[VARRAY:%.*]]) {
 ; NEON:    [[TMP2:%.*]] = call <4 x float> @_ZGVnN4v_exp2f(<4 x float> [[TMP1:%.*]])
-; NEON:    [[CALL:%.*]] = tail call float @exp2f(float [[CONV:%.*]]) #[[ATTR17:[0-9]+]]
 ; NEON:    ret void
 ;
 ; SVE-LABEL: define void @exp2_f32
 ; SVE-SAME: (ptr nocapture [[VARRAY:%.*]]) #[[ATTR0]] {
-; SVE:    [[TMP12:%.*]] = call <vscale x 4 x float> @_ZGVsMxv_exp2f(<vscale x 4 x float> [[TMP11:%.*]], <vscale x 4 x i1> shufflevector (<vscale...
[truncated]

paulwalker-arm
paulwalker-arm approved these changes Dec 18, 2023
View reviewed changes
Copy link
Collaborator

@paulwalker-arm paulwalker-arm left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It's worth mentioning this is specific to SLEEFGNUABI.

@paschalis-mpeis
[TLI] Add SLEEFGNUABI mappings for fmod/fmodf fixed-width.
65d447f 
Cleanup test sleef-calls-aarch64.ll:
- make the util update script's regex more clear
- eliminate scalar epilogues in tests
@paschalis-mpeis paschalis-mpeis force-pushed the users/paschalis-mpeis/frem-tli-mappings-sleef branch from b4024d1 to 65d447f Compare December 19, 2023 12:38
@paschalis-mpeis paschalis-mpeis changed the title [TLI] Add SLEEF mappings for fmod/fmodf fixed-width. [TLI] Add SLEEFGNUABI mappings for fmod/fmodf fixed-width. Dec 19, 2023
@paschalis-mpeis paschalis-mpeis merged commit 2349731 into main Dec 20, 2023
@paschalis-mpeis paschalis-mpeis deleted the users/paschalis-mpeis/frem-tli-mappings-sleef branch December 20, 2023 09:08
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@paulwalker-arm paulwalker-arm paulwalker-arm approved these changes

@mgabka mgabka mgabka approved these changes

Assignees
No one assigned
Labels
llvm:analysis Includes value tracking, cost tables and constant folding llvm:transforms
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
4 participants
@paschalis-mpeis @llvmbot @paulwalker-arm @mgabka