RE: [PATCH] D84703: [clang codegen][AArch64] Use llvm.aarch64.neon.fcvtzs/u where it's necessary

[Eli Friedman via cfe-commits]

Committed dca23ed; should be fixed now.

-Eli

From: Azhar Mohammed 
Sent: Monday, August 3, 2020 11:01 AM
To: Eli Friedman ; Eli 
Friedman via Phabricator ; Eli Friedman 

Cc: sander.desma...@arm.com; sjoerd.mei...@arm.com; t.p.northo...@gmail.com; 
cameron.mcina...@nyu.edu; blitzrak...@gmail.com; ju...@samsung.com; Anna Welker 
via Phabricator via llvm-commits ; 
mlek...@skidmore.edu; kanh...@a-bix.com; Jonathan Coe via cfe-commits 
; shen...@google.com; kristof.be...@arm.com; 
daniel.k...@arm.com
Subject: [EXT] Re: [PATCH] D84703: [clang codegen][AArch64] Use 
llvm.aarch64.neon.fcvtzs/u where it's necessary

Hey Eli,

Looks like this is causing the test-suite build to fail. Can you please take a 
look.
Refer to 
http://green.lab.llvm.org/green/job/test-suite-verify-machineinstrs-aarch64-O3/8035/consoleFull.
FAILED: 
SingleSource/UnitTests/Vector/AArch64/CMakeFiles/aarch64_neon_intrinsics.dir/aarch64_neon_intrinsics.c.o
/Users/buildslave/jenkins/workspace/test-suite-verify-machineinstrs-aarch64-O3/test-suite-build/tools/timeit
 --summary 
SingleSource/UnitTests/Vector/AArch64/CMakeFiles/aarch64_neon_intrinsics.dir/aarch64_neon_intrinsics.c.o.time
 
/Users/buildslave/jenkins/workspace/test-suite-verify-machineinstrs-aarch64-O3/compiler/bin/clang
 -DNDEBUG -B 
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin
 -Wno-unused-command-line-argument -mllvm -verify-machineinstrs -O3 -arch arm64 
-isysroot 
/Applications/Xcode.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS13.5.sdk
 -w -Werror=date-time -std=c99 -MD -MT 
SingleSource/UnitTests/Vector/AArch64/CMakeFiles/aarch64_neon_intrinsics.dir/aarch64_neon_intrinsics.c.o
 -MF 
SingleSource/UnitTests/Vector/AArch64/CMakeFiles/aarch64_neon_intrinsics.dir/aarch64_neon_intrinsics.c.o.d
 -o 
SingleSource/UnitTests/Vector/AArch64/CMakeFiles/aarch64_neon_intrinsics.dir/aarch64_neon_intrinsics.c.o
 -c 
/Users/buildslave/jenkins/workspace/test-suite-verify-machineinstrs-aarch64-O3/test-suite/SingleSource/UnitTests/Vector/AArch64/aarch64_neon_intrinsics.c
fatal error: error in backend: Cannot select: intrinsic 
%llvm.aarch64.neon.fcvtzs

Thanks
Azhar


On Jul 30, 2020, at 3:42 PM, Eli Friedman via Phabricator via llvm-commits 
mailto:llvm-comm...@lists.llvm.org>> wrote:

This revision was landed with ongoing or failed builds.
This revision was automatically updated to reflect the committed changes.
Closed by commit rG8dfb5d767e70: [clang codegen][AArch64] Use 
llvm.aarch64.neon.fcvtzs/u where its necessary (authored by efriedma).

Changed prior to commit:
 https://reviews.llvm.org/D84703?vs=281052=282069#toc

Repository:
 rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
 https://reviews.llvm.org/D84703/new/

https://reviews.llvm.org/D84703

Files:
 clang/lib/CodeGen/CGBuiltin.cpp
 clang/test/CodeGen/aarch64-neon-fcvt-intrinsics.c
 clang/test/CodeGen/aarch64-neon-intrinsics.c
 clang/test/CodeGen/aarch64-neon-misc.c
 clang/test/CodeGen/aarch64-v8.2a-fp16-intrinsics.c
 clang/test/CodeGen/aarch64-v8.2a-neon-intrinsics.c

___
llvm-commits mailing list
llvm-comm...@lists.llvm.org<mailto:llvm-comm...@lists.llvm.org>
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Re: [PATCH] D84703: [clang codegen][AArch64] Use llvm.aarch64.neon.fcvtzs/u where it's necessary

[Azhar Mohammed via cfe-commits]

Hey Eli,

Looks like this is causing the test-suite build to fail. Can you please take a 
look.

Refer to 
http://green.lab.llvm.org/green/job/test-suite-verify-machineinstrs-aarch64-O3/8035/consoleFull
 
.

FAILED: 
SingleSource/UnitTests/Vector/AArch64/CMakeFiles/aarch64_neon_intrinsics.dir/aarch64_neon_intrinsics.c.o
 
/Users/buildslave/jenkins/workspace/test-suite-verify-machineinstrs-aarch64-O3/test-suite-build/tools/timeit
 --summary 
SingleSource/UnitTests/Vector/AArch64/CMakeFiles/aarch64_neon_intrinsics.dir/aarch64_neon_intrinsics.c.o.time
 
/Users/buildslave/jenkins/workspace/test-suite-verify-machineinstrs-aarch64-O3/compiler/bin/clang
 -DNDEBUG -B 
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin
 -Wno-unused-command-line-argument -mllvm -verify-machineinstrs -O3 -arch arm64 
-isysroot 
/Applications/Xcode.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS13.5.sdk
 -w -Werror=date-time -std=c99 -MD -MT 
SingleSource/UnitTests/Vector/AArch64/CMakeFiles/aarch64_neon_intrinsics.dir/aarch64_neon_intrinsics.c.o
 -MF 
SingleSource/UnitTests/Vector/AArch64/CMakeFiles/aarch64_neon_intrinsics.dir/aarch64_neon_intrinsics.c.o.d
 -o 
SingleSource/UnitTests/Vector/AArch64/CMakeFiles/aarch64_neon_intrinsics.dir/aarch64_neon_intrinsics.c.o
 -c 
/Users/buildslave/jenkins/workspace/test-suite-verify-machineinstrs-aarch64-O3/test-suite/SingleSource/UnitTests/Vector/AArch64/aarch64_neon_intrinsics.c
fatal error: error in backend: Cannot select: intrinsic 
%llvm.aarch64.neon.fcvtzs

Thanks
Azhar

> On Jul 30, 2020, at 3:42 PM, Eli Friedman via Phabricator via llvm-commits 
>  wrote:
> 
> This revision was landed with ongoing or failed builds.
> This revision was automatically updated to reflect the committed changes.
> Closed by commit rG8dfb5d767e70: [clang codegen][AArch64] Use 
> llvm.aarch64.neon.fcvtzs/u where its necessary (authored by efriedma).
> 
> Changed prior to commit:
>  https://reviews.llvm.org/D84703?vs=281052=282069#toc
> 
> Repository:
>  rG LLVM Github Monorepo
> 
> CHANGES SINCE LAST ACTION
>  https://reviews.llvm.org/D84703/new/
> 
> https://reviews.llvm.org/D84703
> 
> Files:
>  clang/lib/CodeGen/CGBuiltin.cpp
>  clang/test/CodeGen/aarch64-neon-fcvt-intrinsics.c
>  clang/test/CodeGen/aarch64-neon-intrinsics.c
>  clang/test/CodeGen/aarch64-neon-misc.c
>  clang/test/CodeGen/aarch64-v8.2a-fp16-intrinsics.c
>  clang/test/CodeGen/aarch64-v8.2a-neon-intrinsics.c
> 
> ___
> llvm-commits mailing list
> llvm-comm...@lists.llvm.org
> https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D84703: [clang codegen][AArch64] Use llvm.aarch64.neon.fcvtzs/u where it's necessary

[Eli Friedman via Phabricator via cfe-commits]

This revision was landed with ongoing or failed builds.
This revision was automatically updated to reflect the committed changes.
Closed by commit rG8dfb5d767e70: [clang codegen][AArch64] Use 
llvm.aarch64.neon.fcvtzs/u where its necessary (authored by efriedma).

Changed prior to commit:
  https://reviews.llvm.org/D84703?vs=281052=282069#toc

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D84703/new/

https://reviews.llvm.org/D84703

Files:
  clang/lib/CodeGen/CGBuiltin.cpp
  clang/test/CodeGen/aarch64-neon-fcvt-intrinsics.c
  clang/test/CodeGen/aarch64-neon-intrinsics.c
  clang/test/CodeGen/aarch64-neon-misc.c
  clang/test/CodeGen/aarch64-v8.2a-fp16-intrinsics.c
  clang/test/CodeGen/aarch64-v8.2a-neon-intrinsics.c

Index: clang/test/CodeGen/aarch64-v8.2a-neon-intrinsics.c
===
--- clang/test/CodeGen/aarch64-v8.2a-neon-intrinsics.c
+++ clang/test/CodeGen/aarch64-v8.2a-neon-intrinsics.c
@@ -130,28 +130,28 @@
 }
 
 // CHECK-LABEL: test_vcvt_s16_f16
-// CHECK:  [[VCVT:%.*]] = fptosi <4 x half> %a to <4 x i16>
+// CHECK:  [[VCVT:%.*]] = call <4 x i16> @llvm.aarch64.neon.fcvtzs.v4i16.v4f16(<4 x half> %a)
 // CHECK:  ret <4 x i16> [[VCVT]]
 int16x4_t test_vcvt_s16_f16 (float16x4_t a) {
   return vcvt_s16_f16(a);
 }
 
 // CHECK-LABEL: test_vcvtq_s16_f16
-// CHECK:  [[VCVT:%.*]] = fptosi <8 x half> %a to <8 x i16>
+// CHECK:  [[VCVT:%.*]] = call <8 x i16> @llvm.aarch64.neon.fcvtzs.v8i16.v8f16(<8 x half> %a)
 // CHECK:  ret <8 x i16> [[VCVT]]
 int16x8_t test_vcvtq_s16_f16 (float16x8_t a) {
   return vcvtq_s16_f16(a);
 }
 
 // CHECK-LABEL: test_vcvt_u16_f16
-// CHECK:  [[VCVT:%.*]] = fptoui <4 x half> %a to <4 x i16>
+// CHECK:  [[VCVT:%.*]] = call <4 x i16> @llvm.aarch64.neon.fcvtzu.v4i16.v4f16(<4 x half> %a)
 // CHECK:  ret <4 x i16> [[VCVT]]
 uint16x4_t test_vcvt_u16_f16 (float16x4_t a) {
   return vcvt_u16_f16(a);
 }
 
 // CHECK-LABEL: test_vcvtq_u16_f16
-// CHECK:  [[VCVT:%.*]] = fptoui <8 x half> %a to <8 x i16>
+// CHECK:  [[VCVT:%.*]] = call <8 x i16> @llvm.aarch64.neon.fcvtzu.v8i16.v8f16(<8 x half> %a)
 // CHECK:  ret <8 x i16> [[VCVT]]
 uint16x8_t test_vcvtq_u16_f16 (float16x8_t a) {
   return vcvtq_u16_f16(a);
Index: clang/test/CodeGen/aarch64-v8.2a-fp16-intrinsics.c
===
--- clang/test/CodeGen/aarch64-v8.2a-fp16-intrinsics.c
+++ clang/test/CodeGen/aarch64-v8.2a-fp16-intrinsics.c
@@ -97,42 +97,44 @@
 }
 
 // CHECK-LABEL: test_vcvth_s16_f16
-// CHECK:  [[VCVT:%.*]] = fptosi half %a to i16
-// CHECK:  ret i16 [[VCVT]]
+// CHECK:  [[VCVT:%.*]] = call i32 @llvm.aarch64.neon.fcvtzs.i32.f16(half %a)
+// CHECK:  [[TRUNC:%.*]] = trunc i32 [[VCVT]] to i16
+// CHECK:  ret i16 [[TRUNC]]
 int16_t test_vcvth_s16_f16 (float16_t a) {
   return vcvth_s16_f16(a);
 }
 
 // CHECK-LABEL: test_vcvth_s32_f16
-// CHECK:  [[VCVT:%.*]] = fptosi half %a to i32
+// CHECK:  [[VCVT:%.*]] = call i32 @llvm.aarch64.neon.fcvtzs.i32.f16(half %a)
 // CHECK:  ret i32 [[VCVT]]
 int32_t test_vcvth_s32_f16 (float16_t a) {
   return vcvth_s32_f16(a);
 }
 
 // CHECK-LABEL: test_vcvth_s64_f16
-// CHECK:  [[VCVT:%.*]] = fptosi half %a to i64
+// CHECK:  [[VCVT:%.*]] = call i64 @llvm.aarch64.neon.fcvtzs.i64.f16(half %a)
 // CHECK:  ret i64 [[VCVT]]
 int64_t test_vcvth_s64_f16 (float16_t a) {
   return vcvth_s64_f16(a);
 }
 
 // CHECK-LABEL: test_vcvth_u16_f16
-// CHECK:  [[VCVT:%.*]] = fptoui half %a to i16
-// CHECK:  ret i16 [[VCVT]]
+// CHECK:  [[VCVT:%.*]] = call i32 @llvm.aarch64.neon.fcvtzu.i32.f16(half %a)
+// CHECK:  [[TRUNC:%.*]] = trunc i32 [[VCVT]] to i16
+// CHECK:  ret i16 [[TRUNC]]
 uint16_t test_vcvth_u16_f16 (float16_t a) {
   return vcvth_u16_f16(a);
 }
 
 // CHECK-LABEL: test_vcvth_u32_f16
-// CHECK:  [[VCVT:%.*]] = fptoui half %a to i32
+// CHECK:  [[VCVT:%.*]] = call i32 @llvm.aarch64.neon.fcvtzu.i32.f16(half %a)
 // CHECK:  ret i32 [[VCVT]]
 uint32_t test_vcvth_u32_f16 (float16_t a) {
   return vcvth_u32_f16(a);
 }
 
 // CHECK-LABEL: test_vcvth_u64_f16
-// CHECK:  [[VCVT:%.*]] = fptoui half %a to i64
+// CHECK:  [[VCVT:%.*]] = call i64 @llvm.aarch64.neon.fcvtzu.i64.f16(half %a)
 // CHECK:  ret i64 [[VCVT]]
 uint64_t test_vcvth_u64_f16 (float16_t a) {
   return vcvth_u64_f16(a);
Index: clang/test/CodeGen/aarch64-neon-misc.c
===
--- clang/test/CodeGen/aarch64-neon-misc.c
+++ clang/test/CodeGen/aarch64-neon-misc.c
@@ -2311,7 +2311,7 @@
 
 // CHECK-LABEL: @test_vcvt_s32_f32(
 // CHECK:   [[TMP0:%.*]] = bitcast <2 x float> %a to <8 x i8>
-// CHECK:   [[TMP1:%.*]] = fptosi <2 x float> %a to <2 x i32>
+// CHECK:   [[TMP1:%.*]] = call <2 x i32> @llvm.aarch64.neon.fcvtzs.v2i32.v2f32(<2 x float> %a)
 // CHECK:   ret <2 x i32> [[TMP1]]
 int32x2_t test_vcvt_s32_f32(float32x2_t a) {
   return vcvt_s32_f32(a);
@@ -2319,7 +2319,7 @@
 
 // CHECK-LABEL: @test_vcvtq_s32_f32(
 // CHECK:   [[TMP0:%.*]] = bitcast <4 

[PATCH] D84703: [clang codegen][AArch64] Use llvm.aarch64.neon.fcvtzs/u where it's necessary

[Sjoerd Meijer via Phabricator via cfe-commits]

SjoerdMeijer accepted this revision.
SjoerdMeijer added a comment.
This revision is now accepted and ready to land.

Cheers, LGTM


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D84703/new/

https://reviews.llvm.org/D84703

___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D84703: [clang codegen][AArch64] Use llvm.aarch64.neon.fcvtzs/u where it's necessary

[Eli Friedman via Phabricator via cfe-commits]

efriedma created this revision.
efriedma added reviewers: sdesmalen, SjoerdMeijer, t.p.northover, 
cameron.mcinally.
Herald added subscribers: cfe-commits, danielkiss, kristof.beyls.
Herald added a project: clang.

fptosi/fptoui have similar, but not identical, semantics.  In particular, the 
behavior on overflow is different.

Fixes https://bugs.llvm.org/show_bug.cgi?id=46844 for 64-bit.  (This is missing 
the corresponding change for 32-bit; the 32-bit equivalent to 
llvm.aarch64.neon.fcvtzs isn't implemented, as far as I can tell.)


Repository:
  rG LLVM Github Monorepo

https://reviews.llvm.org/D84703

Files:
  clang/lib/CodeGen/CGBuiltin.cpp
  clang/test/CodeGen/aarch64-neon-fcvt-intrinsics.c
  clang/test/CodeGen/aarch64-neon-intrinsics.c
  clang/test/CodeGen/aarch64-neon-misc.c
  clang/test/CodeGen/aarch64-v8.2a-fp16-intrinsics.c
  clang/test/CodeGen/aarch64-v8.2a-neon-intrinsics.c

Index: clang/test/CodeGen/aarch64-v8.2a-neon-intrinsics.c
===
--- clang/test/CodeGen/aarch64-v8.2a-neon-intrinsics.c
+++ clang/test/CodeGen/aarch64-v8.2a-neon-intrinsics.c
@@ -130,28 +130,28 @@
 }
 
 // CHECK-LABEL: test_vcvt_s16_f16
-// CHECK:  [[VCVT:%.*]] = fptosi <4 x half> %a to <4 x i16>
+// CHECK:  [[VCVT:%.*]] = call <4 x i16> @llvm.aarch64.neon.fcvtzs.v4i16.v4f16(<4 x half> %a)
 // CHECK:  ret <4 x i16> [[VCVT]]
 int16x4_t test_vcvt_s16_f16 (float16x4_t a) {
   return vcvt_s16_f16(a);
 }
 
 // CHECK-LABEL: test_vcvtq_s16_f16
-// CHECK:  [[VCVT:%.*]] = fptosi <8 x half> %a to <8 x i16>
+// CHECK:  [[VCVT:%.*]] = call <8 x i16> @llvm.aarch64.neon.fcvtzs.v8i16.v8f16(<8 x half> %a)
 // CHECK:  ret <8 x i16> [[VCVT]]
 int16x8_t test_vcvtq_s16_f16 (float16x8_t a) {
   return vcvtq_s16_f16(a);
 }
 
 // CHECK-LABEL: test_vcvt_u16_f16
-// CHECK:  [[VCVT:%.*]] = fptoui <4 x half> %a to <4 x i16>
+// CHECK:  [[VCVT:%.*]] = call <4 x i16> @llvm.aarch64.neon.fcvtzu.v4i16.v4f16(<4 x half> %a)
 // CHECK:  ret <4 x i16> [[VCVT]]
 uint16x4_t test_vcvt_u16_f16 (float16x4_t a) {
   return vcvt_u16_f16(a);
 }
 
 // CHECK-LABEL: test_vcvtq_u16_f16
-// CHECK:  [[VCVT:%.*]] = fptoui <8 x half> %a to <8 x i16>
+// CHECK:  [[VCVT:%.*]] = call <8 x i16> @llvm.aarch64.neon.fcvtzu.v8i16.v8f16(<8 x half> %a)
 // CHECK:  ret <8 x i16> [[VCVT]]
 uint16x8_t test_vcvtq_u16_f16 (float16x8_t a) {
   return vcvtq_u16_f16(a);
Index: clang/test/CodeGen/aarch64-v8.2a-fp16-intrinsics.c
===
--- clang/test/CodeGen/aarch64-v8.2a-fp16-intrinsics.c
+++ clang/test/CodeGen/aarch64-v8.2a-fp16-intrinsics.c
@@ -97,42 +97,44 @@
 }
 
 // CHECK-LABEL: test_vcvth_s16_f16
-// CHECK:  [[VCVT:%.*]] = fptosi half %a to i16
-// CHECK:  ret i16 [[VCVT]]
+// CHECK:  [[VCVT:%.*]] = call i32 @llvm.aarch64.neon.fcvtzs.i32.f16(half %a)
+// CHECK:  [[TRUNC:%.*]] = trunc i32 [[VCVT]] to i16
+// CHECK:  ret i16 [[TRUNC]]
 int16_t test_vcvth_s16_f16 (float16_t a) {
   return vcvth_s16_f16(a);
 }
 
 // CHECK-LABEL: test_vcvth_s32_f16
-// CHECK:  [[VCVT:%.*]] = fptosi half %a to i32
+// CHECK:  [[VCVT:%.*]] = call i32 @llvm.aarch64.neon.fcvtzs.i32.f16(half %a)
 // CHECK:  ret i32 [[VCVT]]
 int32_t test_vcvth_s32_f16 (float16_t a) {
   return vcvth_s32_f16(a);
 }
 
 // CHECK-LABEL: test_vcvth_s64_f16
-// CHECK:  [[VCVT:%.*]] = fptosi half %a to i64
+// CHECK:  [[VCVT:%.*]] = call i64 @llvm.aarch64.neon.fcvtzs.i64.f16(half %a)
 // CHECK:  ret i64 [[VCVT]]
 int64_t test_vcvth_s64_f16 (float16_t a) {
   return vcvth_s64_f16(a);
 }
 
 // CHECK-LABEL: test_vcvth_u16_f16
-// CHECK:  [[VCVT:%.*]] = fptoui half %a to i16
-// CHECK:  ret i16 [[VCVT]]
+// CHECK:  [[VCVT:%.*]] = call i32 @llvm.aarch64.neon.fcvtzu.i32.f16(half %a)
+// CHECK:  [[TRUNC:%.*]] = trunc i32 [[VCVT]] to i16
+// CHECK:  ret i16 [[TRUNC]]
 uint16_t test_vcvth_u16_f16 (float16_t a) {
   return vcvth_u16_f16(a);
 }
 
 // CHECK-LABEL: test_vcvth_u32_f16
-// CHECK:  [[VCVT:%.*]] = fptoui half %a to i32
+// CHECK:  [[VCVT:%.*]] = call i32 @llvm.aarch64.neon.fcvtzu.i32.f16(half %a)
 // CHECK:  ret i32 [[VCVT]]
 uint32_t test_vcvth_u32_f16 (float16_t a) {
   return vcvth_u32_f16(a);
 }
 
 // CHECK-LABEL: test_vcvth_u64_f16
-// CHECK:  [[VCVT:%.*]] = fptoui half %a to i64
+// CHECK:  [[VCVT:%.*]] = call i64 @llvm.aarch64.neon.fcvtzu.i64.f16(half %a)
 // CHECK:  ret i64 [[VCVT]]
 uint64_t test_vcvth_u64_f16 (float16_t a) {
   return vcvth_u64_f16(a);
Index: clang/test/CodeGen/aarch64-neon-misc.c
===
--- clang/test/CodeGen/aarch64-neon-misc.c
+++ clang/test/CodeGen/aarch64-neon-misc.c
@@ -2311,7 +2311,7 @@
 
 // CHECK-LABEL: @test_vcvt_s32_f32(
 // CHECK:   [[TMP0:%.*]] = bitcast <2 x float> %a to <8 x i8>
-// CHECK:   [[TMP1:%.*]] = fptosi <2 x float> %a to <2 x i32>
+// CHECK:   [[TMP1:%.*]] = call <2 x i32> @llvm.aarch64.neon.fcvtzs.v2i32.v2f32(<2 x float> %a)
 // CHECK:   ret <2 x i32> [[TMP1]]
 int32x2_t test_vcvt_s32_f32(float32x2_t