mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-10-31 16:02:52 +01:00
dbac87d1fc
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. llvm-svn: 187567
79 lines
3.7 KiB
ArmAsm
79 lines
3.7 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 Absolute Difference and Accumulate (Signed, Unsigned)
|
|
//----------------------------------------------------------------------
|
|
uaba v0.8b, v1.8b, v2.8b
|
|
uaba v0.16b, v1.16b, v2.16b
|
|
uaba v0.4h, v1.4h, v2.4h
|
|
uaba v0.8h, v1.8h, v2.8h
|
|
uaba v0.2s, v1.2s, v2.2s
|
|
uaba v0.4s, v1.4s, v2.4s
|
|
|
|
// CHECK: uaba v0.8b, v1.8b, v2.8b // encoding: [0x20,0x7c,0x22,0x2e]
|
|
// CHECK: uaba v0.16b, v1.16b, v2.16b // encoding: [0x20,0x7c,0x22,0x6e]
|
|
// CHECK: uaba v0.4h, v1.4h, v2.4h // encoding: [0x20,0x7c,0x62,0x2e]
|
|
// CHECK: uaba v0.8h, v1.8h, v2.8h // encoding: [0x20,0x7c,0x62,0x6e]
|
|
// CHECK: uaba v0.2s, v1.2s, v2.2s // encoding: [0x20,0x7c,0xa2,0x2e]
|
|
// CHECK: uaba v0.4s, v1.4s, v2.4s // encoding: [0x20,0x7c,0xa2,0x6e]
|
|
|
|
|
|
saba v0.8b, v1.8b, v2.8b
|
|
saba v0.16b, v1.16b, v2.16b
|
|
saba v0.4h, v1.4h, v2.4h
|
|
saba v0.8h, v1.8h, v2.8h
|
|
saba v0.2s, v1.2s, v2.2s
|
|
saba v0.4s, v1.4s, v2.4s
|
|
|
|
// CHECK: saba v0.8b, v1.8b, v2.8b // encoding: [0x20,0x7c,0x22,0x0e]
|
|
// CHECK: saba v0.16b, v1.16b, v2.16b // encoding: [0x20,0x7c,0x22,0x4e]
|
|
// CHECK: saba v0.4h, v1.4h, v2.4h // encoding: [0x20,0x7c,0x62,0x0e]
|
|
// CHECK: saba v0.8h, v1.8h, v2.8h // encoding: [0x20,0x7c,0x62,0x4e]
|
|
// CHECK: saba v0.2s, v1.2s, v2.2s // encoding: [0x20,0x7c,0xa2,0x0e]
|
|
// CHECK: saba v0.4s, v1.4s, v2.4s // encoding: [0x20,0x7c,0xa2,0x4e]
|
|
|
|
//----------------------------------------------------------------------
|
|
// Vector Absolute Difference (Signed, Unsigned)
|
|
//----------------------------------------------------------------------
|
|
uabd v0.8b, v1.8b, v2.8b
|
|
uabd v0.16b, v1.16b, v2.16b
|
|
uabd v0.4h, v1.4h, v2.4h
|
|
uabd v0.8h, v1.8h, v2.8h
|
|
uabd v0.2s, v1.2s, v2.2s
|
|
uabd v0.4s, v1.4s, v2.4s
|
|
|
|
// CHECK: uabd v0.8b, v1.8b, v2.8b // encoding: [0x20,0x74,0x22,0x2e]
|
|
// CHECK: uabd v0.16b, v1.16b, v2.16b // encoding: [0x20,0x74,0x22,0x6e]
|
|
// CHECK: uabd v0.4h, v1.4h, v2.4h // encoding: [0x20,0x74,0x62,0x2e]
|
|
// CHECK: uabd v0.8h, v1.8h, v2.8h // encoding: [0x20,0x74,0x62,0x6e]
|
|
// CHECK: uabd v0.2s, v1.2s, v2.2s // encoding: [0x20,0x74,0xa2,0x2e]
|
|
// CHECK: uabd v0.4s, v1.4s, v2.4s // encoding: [0x20,0x74,0xa2,0x6e]
|
|
|
|
sabd v0.8b, v1.8b, v2.8b
|
|
sabd v0.16b, v1.16b, v2.16b
|
|
sabd v0.4h, v1.4h, v2.4h
|
|
sabd v0.8h, v1.8h, v2.8h
|
|
sabd v0.2s, v1.2s, v2.2s
|
|
sabd v0.4s, v1.4s, v2.4s
|
|
|
|
// CHECK: sabd v0.8b, v1.8b, v2.8b // encoding: [0x20,0x74,0x22,0x0e]
|
|
// CHECK: sabd v0.16b, v1.16b, v2.16b // encoding: [0x20,0x74,0x22,0x4e]
|
|
// CHECK: sabd v0.4h, v1.4h, v2.4h // encoding: [0x20,0x74,0x62,0x0e]
|
|
// CHECK: sabd v0.8h, v1.8h, v2.8h // encoding: [0x20,0x74,0x62,0x4e]
|
|
// CHECK: sabd v0.2s, v1.2s, v2.2s // encoding: [0x20,0x74,0xa2,0x0e]
|
|
// CHECK: sabd v0.4s, v1.4s, v2.4s // encoding: [0x20,0x74,0xa2,0x4e]
|
|
|
|
//----------------------------------------------------------------------
|
|
// Vector Absolute Difference (Floating Point)
|
|
//----------------------------------------------------------------------
|
|
fabd v0.2s, v1.2s, v2.2s
|
|
fabd v31.4s, v15.4s, v16.4s
|
|
fabd v7.2d, v8.2d, v25.2d
|
|
|
|
// CHECK: fabd v0.2s, v1.2s, v2.2s // encoding: [0x20,0xd4,0xa2,0x2e]
|
|
// CHECK: fabd v31.4s, v15.4s, v16.4s // encoding: [0xff,0xd5,0xb0,0x6e]
|
|
// CHECK: fabd v7.2d, v8.2d, v25.2d // encoding: [0x07,0xd5,0xf9,0x6e]
|
|
|