llvm-6502/test/MC/AArch64/neon-max-min-pairwise.s
Tim Northover 87773c318f 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
2013-08-01 09:20:35 +00:00

111 lines
5.3 KiB
ArmAsm

// RUN: llvm-mc -triple aarch64-none-linux-gnu -mattr=+neon -show-encoding < %s | FileCheck %s
// Check that the assembler can handle the documented syntax for AArch64
//----------------------------------------------------------------------
// Vector Maximum Pairwise (Signed and Unsigned Integer)
//----------------------------------------------------------------------
smaxp v0.8b, v1.8b, v2.8b
smaxp v0.16b, v1.16b, v2.16b
smaxp v0.4h, v1.4h, v2.4h
smaxp v0.8h, v1.8h, v2.8h
smaxp v0.2s, v1.2s, v2.2s
smaxp v0.4s, v1.4s, v2.4s
// CHECK: smaxp v0.8b, v1.8b, v2.8b // encoding: [0x20,0xa4,0x22,0x0e]
// CHECK: smaxp v0.16b, v1.16b, v2.16b // encoding: [0x20,0xa4,0x22,0x4e]
// CHECK: smaxp v0.4h, v1.4h, v2.4h // encoding: [0x20,0xa4,0x62,0x0e]
// CHECK: smaxp v0.8h, v1.8h, v2.8h // encoding: [0x20,0xa4,0x62,0x4e]
// CHECK: smaxp v0.2s, v1.2s, v2.2s // encoding: [0x20,0xa4,0xa2,0x0e]
// CHECK: smaxp v0.4s, v1.4s, v2.4s // encoding: [0x20,0xa4,0xa2,0x4e]
umaxp v0.8b, v1.8b, v2.8b
umaxp v0.16b, v1.16b, v2.16b
umaxp v0.4h, v1.4h, v2.4h
umaxp v0.8h, v1.8h, v2.8h
umaxp v0.2s, v1.2s, v2.2s
umaxp v0.4s, v1.4s, v2.4s
// CHECK: umaxp v0.8b, v1.8b, v2.8b // encoding: [0x20,0xa4,0x22,0x2e]
// CHECK: umaxp v0.16b, v1.16b, v2.16b // encoding: [0x20,0xa4,0x22,0x6e]
// CHECK: umaxp v0.4h, v1.4h, v2.4h // encoding: [0x20,0xa4,0x62,0x2e]
// CHECK: umaxp v0.8h, v1.8h, v2.8h // encoding: [0x20,0xa4,0x62,0x6e]
// CHECK: umaxp v0.2s, v1.2s, v2.2s // encoding: [0x20,0xa4,0xa2,0x2e]
// CHECK: umaxp v0.4s, v1.4s, v2.4s // encoding: [0x20,0xa4,0xa2,0x6e]
//----------------------------------------------------------------------
// Vector Minimum Pairwise (Signed and Unsigned Integer)
//----------------------------------------------------------------------
sminp v0.8b, v1.8b, v2.8b
sminp v0.16b, v1.16b, v2.16b
sminp v0.4h, v1.4h, v2.4h
sminp v0.8h, v1.8h, v2.8h
sminp v0.2s, v1.2s, v2.2s
sminp v0.4s, v1.4s, v2.4s
// CHECK: sminp v0.8b, v1.8b, v2.8b // encoding: [0x20,0xac,0x22,0x0e]
// CHECK: sminp v0.16b, v1.16b, v2.16b // encoding: [0x20,0xac,0x22,0x4e]
// CHECK: sminp v0.4h, v1.4h, v2.4h // encoding: [0x20,0xac,0x62,0x0e]
// CHECK: sminp v0.8h, v1.8h, v2.8h // encoding: [0x20,0xac,0x62,0x4e]
// CHECK: sminp v0.2s, v1.2s, v2.2s // encoding: [0x20,0xac,0xa2,0x0e]
// CHECK: sminp v0.4s, v1.4s, v2.4s // encoding: [0x20,0xac,0xa2,0x4e]
uminp v0.8b, v1.8b, v2.8b
uminp v0.16b, v1.16b, v2.16b
uminp v0.4h, v1.4h, v2.4h
uminp v0.8h, v1.8h, v2.8h
uminp v0.2s, v1.2s, v2.2s
uminp v0.4s, v1.4s, v2.4s
// CHECK: uminp v0.8b, v1.8b, v2.8b // encoding: [0x20,0xac,0x22,0x2e]
// CHECK: uminp v0.16b, v1.16b, v2.16b // encoding: [0x20,0xac,0x22,0x6e]
// CHECK: uminp v0.4h, v1.4h, v2.4h // encoding: [0x20,0xac,0x62,0x2e]
// CHECK: uminp v0.8h, v1.8h, v2.8h // encoding: [0x20,0xac,0x62,0x6e]
// CHECK: uminp v0.2s, v1.2s, v2.2s // encoding: [0x20,0xac,0xa2,0x2e]
// CHECK: uminp v0.4s, v1.4s, v2.4s // encoding: [0x20,0xac,0xa2,0x6e]
//----------------------------------------------------------------------
// Vector Maximum Pairwise (Floating Point)
//----------------------------------------------------------------------
fmaxp v0.2s, v1.2s, v2.2s
fmaxp v31.4s, v15.4s, v16.4s
fmaxp v7.2d, v8.2d, v25.2d
// CHECK: fmaxp v0.2s, v1.2s, v2.2s // encoding: [0x20,0xf4,0x22,0x2e]
// CHECK: fmaxp v31.4s, v15.4s, v16.4s // encoding: [0xff,0xf5,0x30,0x6e]
// CHECK: fmaxp v7.2d, v8.2d, v25.2d // encoding: [0x07,0xf5,0x79,0x6e]
//----------------------------------------------------------------------
// Vector Minimum Pairwise (Floating Point)
//----------------------------------------------------------------------
fminp v10.2s, v15.2s, v22.2s
fminp v3.4s, v5.4s, v6.4s
fminp v17.2d, v13.2d, v2.2d
// CHECK: fminp v10.2s, v15.2s, v22.2s // encoding: [0xea,0xf5,0xb6,0x2e]
// CHECK: fminp v3.4s, v5.4s, v6.4s // encoding: [0xa3,0xf4,0xa6,0x6e]
// CHECK: fminp v17.2d, v13.2d, v2.2d // encoding: [0xb1,0xf5,0xe2,0x6e]
//----------------------------------------------------------------------
// Vector maxNum Pairwise (Floating Point)
//----------------------------------------------------------------------
fmaxnmp v0.2s, v1.2s, v2.2s
fmaxnmp v31.4s, v15.4s, v16.4s
fmaxnmp v7.2d, v8.2d, v25.2d
// CHECK: fmaxnmp v0.2s, v1.2s, v2.2s // encoding: [0x20,0xc4,0x22,0x2e]
// CHECK: fmaxnmp v31.4s, v15.4s, v16.4s // encoding: [0xff,0xc5,0x30,0x6e]
// CHECK: fmaxnmp v7.2d, v8.2d, v25.2d // encoding: [0x07,0xc5,0x79,0x6e]
//----------------------------------------------------------------------
// Vector minNum Pairwise (Floating Point)
//----------------------------------------------------------------------
fminnmp v10.2s, v15.2s, v22.2s
fminnmp v3.4s, v5.4s, v6.4s
fminnmp v17.2d, v13.2d, v2.2d
// CHECK: fminnmp v10.2s, v15.2s, v22.2s // encoding: [0xea,0xc5,0xb6,0x2e]
// CHECK: fminnmp v3.4s, v5.4s, v6.4s // encoding: [0xa3,0xc4,0xa6,0x6e]
// CHECK: fminnmp v17.2d, v13.2d, v2.2d // encoding: [0xb1,0xc5,0xe2,0x6e]