T0b1/wasmtime
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Implement VhighBits & Vselect for interpreter

Browse Source 
Implemented the following Opcodes for the Cranelift interpreter:
- `VhighBits` to reduce a vector to a scalar integer formed by
concatenating the MSB of each lane.
- `Vselect` to select lanes from two vectors controlled by a boolean
vector.

Copyright (c) 2021, Arm Limited
This commit is contained in:
dheaton-arm
2021-09-14 12:28:21 +01:00
parent 9323762d71
commit 224a4b4094
3 changed files with 111 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

53
cranelift/filetests/filetests/runtests/simd-vhighbits.clif Normal file
View File

@@ -0,0 +1,53 @@
test interpret
test run
target aarch64
set enable_simd
target x86_64
function %vhighbits_i8x16(i8x16) -> i16 {
block0(v0: i8x16):
v1 = vhigh_bits.i16 v0
return v1
}
; run: %vhighbits_i8x16([0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]) == 0
; run: %vhighbits_i8x16([0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0]) == 0
; run: %vhighbits_i8x16([1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7]) == 0
; run: %vhighbits_i8x16([128 128 128 128 128 128 128 128 128 128 128 128 128 128 128 128]) == -1
; run: %vhighbits_i8x16([128 128 128 128 128 128 128 128 128 128 128 128 128 128 128 8]) == 32767
function %vhighbits_i16x8(i16x8) -> i8 {
block0(v0: i16x8):
v1 = vhigh_bits.i8 v0
return v1
}
; run: %vhighbits_i16x8([0 0 0 0 0 0 0 0]) == 0
; run: %vhighbits_i16x8([0 0 0 0 0 0 0 1]) == 0
; run: %vhighbits_i16x8([1 2 3 4 5 6 7 8]) == 0
; run: %vhighbits_i16x8([128 128 128 128 128 128 128 128]) == 0
; run: %vhighbits_i16x8([32768 32768 32768 32768 32768 32768 32768 0]) == 127
function %vhighbits_i32x4(i32x4) -> i8 {
block0(v0: i32x4):
v1 = vhigh_bits.i8 v0
return v1
}
; run: %vhighbits_i32x4([0 0 0 0]) == 0
; run: %vhighbits_i32x4([0 0 0 1]) == 0
; run: %vhighbits_i32x4([1 2 3 4]) == 0
; run: %vhighbits_i32x4([128 128 128 128]) == 0
; run: %vhighbits_i32x4([2147483648 2147483648 2147483648 2147483648]) == 15
; run: %vhighbits_i32x4([2147483648 0 2147483648 0]) == 5
function %vhighbits_i64x2(i64x2) -> i8 {
block0(v0: i64x2):
v1 = vhigh_bits.i8 v0
return v1
}
; run: %vhighbits_i64x2([0 0]) == 0
; run: %vhighbits_i64x2([0 1]) == 0
; run: %vhighbits_i64x2([1 2]) == 0
; run: %vhighbits_i64x2([128 128]) == 0
; run: %vhighbits_i64x2([18446744073709551615 18446744073709551615]) == 3
; run: %vhighbits_i64x2([18446744073709551615 0]) == 1