mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-23 11:13:28 +01:00
a49fa69bbb
This adds a legalization rule for G_ZEXT, G_ANYEXT, and G_SEXT which allows extends whenever the types will fit in registers (or the source is an s1). Update tests. Add GISel checks throughout all of arm64-vabs.ll, where we now select a good portion of the code. Add GISel checks to arm64-subvector-extend.ll, which has a good number of vector extends in it. Differential Revision: https://reviews.llvm.org/D60889 llvm-svn: 359222
148 lines
3.7 KiB
LLVM
148 lines
3.7 KiB
LLVM
; RUN: llc < %s -mtriple=arm64-eabi -aarch64-neon-syntax=apple -asm-verbose=false | FileCheck %s
|
|
; RUN: llc < %s -mtriple=arm64-eabi -aarch64-neon-syntax=apple -asm-verbose=false -global-isel -global-isel-abort=2 -pass-remarks-missed=gisel* 2>&1 | FileCheck %s --check-prefixes=CHECK,FALLBACK
|
|
|
|
; Test efficient codegen of vector extends up from legal type to 128 bit
|
|
; and 256 bit vector types.
|
|
|
|
;-----
|
|
; Vectors of i16.
|
|
;-----
|
|
|
|
; FALLBACK-NOT: remark:{{.*}}(in function: func1)
|
|
define <8 x i16> @func1(<8 x i8> %v0) nounwind {
|
|
; CHECK-LABEL: func1:
|
|
; CHECK-NEXT: ushll.8h v0, v0, #0
|
|
; CHECK-NEXT: ret
|
|
%r = zext <8 x i8> %v0 to <8 x i16>
|
|
ret <8 x i16> %r
|
|
}
|
|
|
|
; FALLBACK-NOT: remark:{{.*}}(in function: func2)
|
|
define <8 x i16> @func2(<8 x i8> %v0) nounwind {
|
|
; CHECK-LABEL: func2:
|
|
; CHECK-NEXT: sshll.8h v0, v0, #0
|
|
; CHECK-NEXT: ret
|
|
%r = sext <8 x i8> %v0 to <8 x i16>
|
|
ret <8 x i16> %r
|
|
}
|
|
|
|
define <16 x i16> @func3(<16 x i8> %v0) nounwind {
|
|
; CHECK-LABEL: func3:
|
|
; CHECK-NEXT: ushll2.8h v1, v0, #0
|
|
; CHECK-NEXT: ushll.8h v0, v0, #0
|
|
; CHECK-NEXT: ret
|
|
%r = zext <16 x i8> %v0 to <16 x i16>
|
|
ret <16 x i16> %r
|
|
}
|
|
|
|
define <16 x i16> @func4(<16 x i8> %v0) nounwind {
|
|
; CHECK-LABEL: func4:
|
|
; CHECK-NEXT: sshll2.8h v1, v0, #0
|
|
; CHECK-NEXT: sshll.8h v0, v0, #0
|
|
; CHECK-NEXT: ret
|
|
%r = sext <16 x i8> %v0 to <16 x i16>
|
|
ret <16 x i16> %r
|
|
}
|
|
|
|
;-----
|
|
; Vectors of i32.
|
|
;-----
|
|
|
|
; FALLBACK-NOT: remark:{{.*}}(in function: afunc1)
|
|
define <4 x i32> @afunc1(<4 x i16> %v0) nounwind {
|
|
; CHECK-LABEL: afunc1:
|
|
; CHECK-NEXT: ushll.4s v0, v0, #0
|
|
; CHECK-NEXT: ret
|
|
%r = zext <4 x i16> %v0 to <4 x i32>
|
|
ret <4 x i32> %r
|
|
}
|
|
|
|
; FALLBACK-NOT: remark:{{.*}}(in function: afunc2)
|
|
define <4 x i32> @afunc2(<4 x i16> %v0) nounwind {
|
|
; CHECK-LABEL: afunc2:
|
|
; CHECK-NEXT: sshll.4s v0, v0, #0
|
|
; CHECK-NEXT: ret
|
|
%r = sext <4 x i16> %v0 to <4 x i32>
|
|
ret <4 x i32> %r
|
|
}
|
|
|
|
define <8 x i32> @afunc3(<8 x i16> %v0) nounwind {
|
|
; CHECK-LABEL: afunc3:
|
|
; CHECK-NEXT: ushll2.4s v1, v0, #0
|
|
; CHECK-NEXT: ushll.4s v0, v0, #0
|
|
; CHECK-NEXT: ret
|
|
%r = zext <8 x i16> %v0 to <8 x i32>
|
|
ret <8 x i32> %r
|
|
}
|
|
|
|
define <8 x i32> @afunc4(<8 x i16> %v0) nounwind {
|
|
; CHECK-LABEL: afunc4:
|
|
; CHECK-NEXT: sshll2.4s v1, v0, #0
|
|
; CHECK-NEXT: sshll.4s v0, v0, #0
|
|
; CHECK-NEXT: ret
|
|
%r = sext <8 x i16> %v0 to <8 x i32>
|
|
ret <8 x i32> %r
|
|
}
|
|
|
|
define <8 x i32> @bfunc1(<8 x i8> %v0) nounwind {
|
|
; CHECK-LABEL: bfunc1:
|
|
; CHECK-NEXT: ushll.8h v0, v0, #0
|
|
; CHECK-NEXT: ushll2.4s v1, v0, #0
|
|
; CHECK-NEXT: ushll.4s v0, v0, #0
|
|
; CHECK-NEXT: ret
|
|
%r = zext <8 x i8> %v0 to <8 x i32>
|
|
ret <8 x i32> %r
|
|
}
|
|
|
|
define <8 x i32> @bfunc2(<8 x i8> %v0) nounwind {
|
|
; CHECK-LABEL: bfunc2:
|
|
; CHECK-NEXT: sshll.8h v0, v0, #0
|
|
; CHECK-NEXT: sshll2.4s v1, v0, #0
|
|
; CHECK-NEXT: sshll.4s v0, v0, #0
|
|
; CHECK-NEXT: ret
|
|
%r = sext <8 x i8> %v0 to <8 x i32>
|
|
ret <8 x i32> %r
|
|
}
|
|
|
|
;-----
|
|
; Vectors of i64.
|
|
;-----
|
|
|
|
define <4 x i64> @zfunc1(<4 x i32> %v0) nounwind {
|
|
; CHECK-LABEL: zfunc1:
|
|
; CHECK-NEXT: ushll2.2d v1, v0, #0
|
|
; CHECK-NEXT: ushll.2d v0, v0, #0
|
|
; CHECK-NEXT: ret
|
|
%r = zext <4 x i32> %v0 to <4 x i64>
|
|
ret <4 x i64> %r
|
|
}
|
|
|
|
define <4 x i64> @zfunc2(<4 x i32> %v0) nounwind {
|
|
; CHECK-LABEL: zfunc2:
|
|
; CHECK-NEXT: sshll2.2d v1, v0, #0
|
|
; CHECK-NEXT: sshll.2d v0, v0, #0
|
|
; CHECK-NEXT: ret
|
|
%r = sext <4 x i32> %v0 to <4 x i64>
|
|
ret <4 x i64> %r
|
|
}
|
|
|
|
define <4 x i64> @bfunc3(<4 x i16> %v0) nounwind {
|
|
; CHECK-LABEL: func3:
|
|
; CHECK-NEXT: ushll.4s v0, v0, #0
|
|
; CHECK-NEXT: ushll2.2d v1, v0, #0
|
|
; CHECK-NEXT: ushll.2d v0, v0, #0
|
|
; CHECK-NEXT: ret
|
|
%r = zext <4 x i16> %v0 to <4 x i64>
|
|
ret <4 x i64> %r
|
|
}
|
|
|
|
define <4 x i64> @cfunc4(<4 x i16> %v0) nounwind {
|
|
; CHECK-LABEL: func4:
|
|
; CHECK-NEXT: sshll.4s v0, v0, #0
|
|
; CHECK-NEXT: sshll2.2d v1, v0, #0
|
|
; CHECK-NEXT: sshll.2d v0, v0, #0
|
|
; CHECK-NEXT: ret
|
|
%r = sext <4 x i16> %v0 to <4 x i64>
|
|
ret <4 x i64> %r
|
|
}
|