diff options
author | Tim Northover <tnorthover@apple.com> | 2013-08-01 09:20:35 +0000 |
---|---|---|
committer | Tim Northover <tnorthover@apple.com> | 2013-08-01 09:20:35 +0000 |
commit | 87773c318fcee853fb34a80a10c4347d523bdafb (patch) | |
tree | 6c8b6620d46529f553a508e9190a264534e0a0dd /test/CodeGen/AArch64/neon-frsqrt-frecp.ll | |
parent | 691aa094dafe54151b6f70168f066bd87c161e8d (diff) | |
download | llvm-87773c318fcee853fb34a80a10c4347d523bdafb.tar.gz llvm-87773c318fcee853fb34a80a10c4347d523bdafb.tar.bz2 llvm-87773c318fcee853fb34a80a10c4347d523bdafb.tar.xz |
AArch64: add initial NEON support
Patch by Ana Pazos.
- Completed implementation of instruction formats:
AdvSIMD three same
AdvSIMD modified immediate
AdvSIMD scalar pairwise
- Completed implementation of instruction classes
(some of the instructions in these classes
belong to yet unfinished instruction formats):
Vector Arithmetic
Vector Immediate
Vector Pairwise Arithmetic
- Initial implementation of instruction formats:
AdvSIMD scalar two-reg misc
AdvSIMD scalar three same
- Intial implementation of instruction class:
Scalar Arithmetic
- Initial clang changes to support arm v8 intrinsics.
Note: no clang changes for scalar intrinsics function name mangling yet.
- Comprehensive test cases for added instructions
To verify auto codegen, encoding, decoding, diagnosis, intrinsics.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@187567 91177308-0d34-0410-b5e6-96231b3b80d8
Diffstat (limited to 'test/CodeGen/AArch64/neon-frsqrt-frecp.ll')
-rw-r--r-- | test/CodeGen/AArch64/neon-frsqrt-frecp.ll | 54 |
1 files changed, 54 insertions, 0 deletions
diff --git a/test/CodeGen/AArch64/neon-frsqrt-frecp.ll b/test/CodeGen/AArch64/neon-frsqrt-frecp.ll new file mode 100644 index 0000000000..46fe25d74d --- /dev/null +++ b/test/CodeGen/AArch64/neon-frsqrt-frecp.ll @@ -0,0 +1,54 @@ +; RUN: llc < %s -verify-machineinstrs -mtriple=aarch64-none-linux-gnu -mattr=+neon | FileCheck %s + +; Set of tests for when the intrinsic is used. + +declare <2 x float> @llvm.arm.neon.vrsqrts.v2f32(<2 x float>, <2 x float>) +declare <4 x float> @llvm.arm.neon.vrsqrts.v4f32(<4 x float>, <4 x float>) +declare <2 x double> @llvm.arm.neon.vrsqrts.v2f64(<2 x double>, <2 x double>) + +define <2 x float> @frsqrts_from_intr_v2f32(<2 x float> %lhs, <2 x float> %rhs) { +; Using registers other than v0, v1 are possible, but would be odd. +; CHECK: frsqrts v0.2s, v0.2s, v1.2s + %val = call <2 x float> @llvm.arm.neon.vrsqrts.v2f32(<2 x float> %lhs, <2 x float> %rhs) + ret <2 x float> %val +} + +define <4 x float> @frsqrts_from_intr_v4f32(<4 x float> %lhs, <4 x float> %rhs) { +; Using registers other than v0, v1 are possible, but would be odd. +; CHECK: frsqrts v0.4s, v0.4s, v1.4s + %val = call <4 x float> @llvm.arm.neon.vrsqrts.v4f32(<4 x float> %lhs, <4 x float> %rhs) + ret <4 x float> %val +} + +define <2 x double> @frsqrts_from_intr_v2f64(<2 x double> %lhs, <2 x double> %rhs) { +; Using registers other than v0, v1 are possible, but would be odd. +; CHECK: frsqrts v0.2d, v0.2d, v1.2d + %val = call <2 x double> @llvm.arm.neon.vrsqrts.v2f64(<2 x double> %lhs, <2 x double> %rhs) + ret <2 x double> %val +} + +declare <2 x float> @llvm.arm.neon.vrecps.v2f32(<2 x float>, <2 x float>) +declare <4 x float> @llvm.arm.neon.vrecps.v4f32(<4 x float>, <4 x float>) +declare <2 x double> @llvm.arm.neon.vrecps.v2f64(<2 x double>, <2 x double>) + +define <2 x float> @frecps_from_intr_v2f32(<2 x float> %lhs, <2 x float> %rhs) { +; Using registers other than v0, v1 are possible, but would be odd. +; CHECK: frecps v0.2s, v0.2s, v1.2s + %val = call <2 x float> @llvm.arm.neon.vrecps.v2f32(<2 x float> %lhs, <2 x float> %rhs) + ret <2 x float> %val +} + +define <4 x float> @frecps_from_intr_v4f32(<4 x float> %lhs, <4 x float> %rhs) { +; Using registers other than v0, v1 are possible, but would be odd. +; CHECK: frecps v0.4s, v0.4s, v1.4s + %val = call <4 x float> @llvm.arm.neon.vrecps.v4f32(<4 x float> %lhs, <4 x float> %rhs) + ret <4 x float> %val +} + +define <2 x double> @frecps_from_intr_v2f64(<2 x double> %lhs, <2 x double> %rhs) { +; Using registers other than v0, v1 are possible, but would be odd. +; CHECK: frecps v0.2d, v0.2d, v1.2d + %val = call <2 x double> @llvm.arm.neon.vrecps.v2f64(<2 x double> %lhs, <2 x double> %rhs) + ret <2 x double> %val +} + |