mirror of
https://github.com/c64scene-ar/llvm-6502.git
synced 2024-08-21 18:29:46 +00:00
Summary: Set operation action for SINT_TO_FP and UINT_TO_FP nodes with v4i32, v8i8, v8i16 inputs to allow promotion of v4f16 results. Add tests for sitofp and uitofp for vec4, vec8, vec16, and i8, i16, i32, and i64 vectors. Only missing tests are for v16i8 and v16i16 as the shift operations are too complicated to write a proper check sequence. The conversions from v4i64 to v4f16 do not depend on this patch - v4i64 is split and the conversion gets handled while lowering v2i64. I am adding a test here for completeness. Reviewers: aemerson, rengolin, ab, jmolloy, srhines Subscribers: rengolin, aemerson, llvm-commits Differential Revision: http://reviews.llvm.org/D9166 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@235609 91177308-0d34-0410-b5e6-96231b3b80d8
222 lines
5.5 KiB
LLVM
222 lines
5.5 KiB
LLVM
; RUN: llc < %s -asm-verbose=false -mtriple=aarch64-none-eabi | FileCheck %s
|
|
|
|
define <4 x half> @add_h(<4 x half> %a, <4 x half> %b) {
|
|
entry:
|
|
; CHECK-LABEL: add_h:
|
|
; CHECK-DAG: fcvtl [[OP1:v[0-9]+\.4s]], v0.4h
|
|
; CHECK-DAG: fcvtl [[OP2:v[0-9]+\.4s]], v1.4h
|
|
; CHECK: fadd [[RES:v[0-9]+.4s]], [[OP1]], [[OP2]]
|
|
; CHECK: fcvtn v0.4h, [[RES]]
|
|
%0 = fadd <4 x half> %a, %b
|
|
ret <4 x half> %0
|
|
}
|
|
|
|
|
|
define <4 x half> @build_h4(<4 x half> %a) {
|
|
entry:
|
|
; CHECK-LABEL: build_h4:
|
|
; CHECK: movz [[GPR:w[0-9]+]], #0x3ccd
|
|
; CHECK: dup v0.4h, [[GPR]]
|
|
ret <4 x half> <half 0xH3CCD, half 0xH3CCD, half 0xH3CCD, half 0xH3CCD>
|
|
}
|
|
|
|
|
|
define <4 x half> @sub_h(<4 x half> %a, <4 x half> %b) {
|
|
entry:
|
|
; CHECK-LABEL: sub_h:
|
|
; CHECK-DAG: fcvtl [[OP1:v[0-9]+\.4s]], v0.4h
|
|
; CHECK-DAG: fcvtl [[OP2:v[0-9]+\.4s]], v1.4h
|
|
; CHECK: fsub [[RES:v[0-9]+.4s]], [[OP1]], [[OP2]]
|
|
; CHECK: fcvtn v0.4h, [[RES]]
|
|
%0 = fsub <4 x half> %a, %b
|
|
ret <4 x half> %0
|
|
}
|
|
|
|
|
|
define <4 x half> @mul_h(<4 x half> %a, <4 x half> %b) {
|
|
entry:
|
|
; CHECK-LABEL: mul_h:
|
|
; CHECK-DAG: fcvtl [[OP1:v[0-9]+\.4s]], v0.4h
|
|
; CHECK-DAG: fcvtl [[OP2:v[0-9]+\.4s]], v1.4h
|
|
; CHECK: fmul [[RES:v[0-9]+.4s]], [[OP1]], [[OP2]]
|
|
; CHECK: fcvtn v0.4h, [[RES]]
|
|
%0 = fmul <4 x half> %a, %b
|
|
ret <4 x half> %0
|
|
}
|
|
|
|
|
|
define <4 x half> @div_h(<4 x half> %a, <4 x half> %b) {
|
|
entry:
|
|
; CHECK-LABEL: div_h:
|
|
; CHECK-DAG: fcvtl [[OP1:v[0-9]+\.4s]], v0.4h
|
|
; CHECK-DAG: fcvtl [[OP2:v[0-9]+\.4s]], v1.4h
|
|
; CHECK: fdiv [[RES:v[0-9]+.4s]], [[OP1]], [[OP2]]
|
|
; CHECK: fcvtn v0.4h, [[RES]]
|
|
%0 = fdiv <4 x half> %a, %b
|
|
ret <4 x half> %0
|
|
}
|
|
|
|
|
|
define <4 x half> @load_h(<4 x half>* %a) {
|
|
entry:
|
|
; CHECK-LABEL: load_h:
|
|
; CHECK: ldr d0, [x0]
|
|
%0 = load <4 x half>, <4 x half>* %a, align 4
|
|
ret <4 x half> %0
|
|
}
|
|
|
|
|
|
define void @store_h(<4 x half>* %a, <4 x half> %b) {
|
|
entry:
|
|
; CHECK-LABEL: store_h:
|
|
; CHECK: str d0, [x0]
|
|
store <4 x half> %b, <4 x half>* %a, align 4
|
|
ret void
|
|
}
|
|
|
|
define <4 x half> @s_to_h(<4 x float> %a) {
|
|
; CHECK-LABEL: s_to_h:
|
|
; CHECK: fcvtn v0.4h, v0.4s
|
|
%1 = fptrunc <4 x float> %a to <4 x half>
|
|
ret <4 x half> %1
|
|
}
|
|
|
|
define <4 x half> @d_to_h(<4 x double> %a) {
|
|
; CHECK-LABEL: d_to_h:
|
|
; CHECK-DAG: fcvt
|
|
; CHECK-DAG: fcvt
|
|
; CHECK-DAG: fcvt
|
|
; CHECK-DAG: fcvt
|
|
; CHECK-DAG: ins
|
|
; CHECK-DAG: ins
|
|
; CHECK-DAG: ins
|
|
; CHECK-DAG: ins
|
|
%1 = fptrunc <4 x double> %a to <4 x half>
|
|
ret <4 x half> %1
|
|
}
|
|
|
|
define <4 x float> @h_to_s(<4 x half> %a) {
|
|
; CHECK-LABEL: h_to_s:
|
|
; CHECK: fcvtl v0.4s, v0.4h
|
|
%1 = fpext <4 x half> %a to <4 x float>
|
|
ret <4 x float> %1
|
|
}
|
|
|
|
define <4 x double> @h_to_d(<4 x half> %a) {
|
|
; CHECK-LABEL: h_to_d:
|
|
; CHECK-DAG: fcvt
|
|
; CHECK-DAG: fcvt
|
|
; CHECK-DAG: fcvt
|
|
; CHECK-DAG: fcvt
|
|
; CHECK-DAG: ins
|
|
; CHECK-DAG: ins
|
|
; CHECK-DAG: ins
|
|
; CHECK-DAG: ins
|
|
%1 = fpext <4 x half> %a to <4 x double>
|
|
ret <4 x double> %1
|
|
}
|
|
|
|
define <4 x half> @bitcast_i_to_h(float, <4 x i16> %a) {
|
|
; CHECK-LABEL: bitcast_i_to_h:
|
|
; CHECK: mov v0.16b, v1.16b
|
|
%2 = bitcast <4 x i16> %a to <4 x half>
|
|
ret <4 x half> %2
|
|
}
|
|
|
|
define <4 x i16> @bitcast_h_to_i(float, <4 x half> %a) {
|
|
; CHECK-LABEL: bitcast_h_to_i:
|
|
; CHECK: mov v0.16b, v1.16b
|
|
%2 = bitcast <4 x half> %a to <4 x i16>
|
|
ret <4 x i16> %2
|
|
}
|
|
|
|
|
|
define <4 x half> @sitofp_i8(<4 x i8> %a) #0 {
|
|
; CHECK-LABEL: sitofp_i8:
|
|
; CHECK-NEXT: shl [[OP1:v[0-9]+\.4h]], v0.4h, #8
|
|
; CHECK-NEXT: sshr [[OP2:v[0-9]+\.4h]], [[OP1]], #8
|
|
; CHECK-NEXT: sshll [[OP3:v[0-9]+\.4s]], [[OP2]], #0
|
|
; CHECK-NEXT: scvtf [[OP4:v[0-9]+\.4s]], [[OP3]]
|
|
; CHECK-NEXT: fcvtn v0.4h, [[OP4]]
|
|
; CHECK-NEXT: ret
|
|
%1 = sitofp <4 x i8> %a to <4 x half>
|
|
ret <4 x half> %1
|
|
}
|
|
|
|
|
|
define <4 x half> @sitofp_i16(<4 x i16> %a) #0 {
|
|
; CHECK-LABEL: sitofp_i16:
|
|
; CHECK-NEXT: sshll [[OP1:v[0-9]+\.4s]], v0.4h, #0
|
|
; CHECK-NEXT: scvtf [[OP2:v[0-9]+\.4s]], [[OP1]]
|
|
; CHECK-NEXT: fcvtn v0.4h, [[OP2]]
|
|
; CHECK-NEXT: ret
|
|
%1 = sitofp <4 x i16> %a to <4 x half>
|
|
ret <4 x half> %1
|
|
}
|
|
|
|
|
|
define <4 x half> @sitofp_i32(<4 x i32> %a) #0 {
|
|
; CHECK-LABEL: sitofp_i32:
|
|
; CHECK-NEXT: scvtf [[OP1:v[0-9]+\.4s]], v0.4s
|
|
; CHECK-NEXT: fcvtn v0.4h, [[OP1]]
|
|
%1 = sitofp <4 x i32> %a to <4 x half>
|
|
ret <4 x half> %1
|
|
}
|
|
|
|
|
|
define <4 x half> @sitofp_i64(<4 x i64> %a) #0 {
|
|
; CHECK-LABEL: sitofp_i64:
|
|
; CHECK-DAG: scvtf [[OP1:v[0-9]+\.2d]], v0.2d
|
|
; CHECK-DAG: scvtf [[OP2:v[0-9]+\.2d]], v1.2d
|
|
; CHECK-DAG: fcvtn [[OP3:v[0-9]+]].2s, [[OP1]]
|
|
; CHECK-NEXT: fcvtn2 [[OP3]].4s, [[OP2]]
|
|
; CHECK-NEXT: fcvtn v0.4h, [[OP3]].4s
|
|
%1 = sitofp <4 x i64> %a to <4 x half>
|
|
ret <4 x half> %1
|
|
}
|
|
|
|
define <4 x half> @uitofp_i8(<4 x i8> %a) #0 {
|
|
; CHECK-LABEL: uitofp_i8:
|
|
; CHECK-NEXT: bic v0.4h, #0xff, lsl #8
|
|
; CHECK-NEXT: ushll [[OP1:v[0-9]+\.4s]], v0.4h, #0
|
|
; CHECK-NEXT: ucvtf [[OP2:v[0-9]+\.4s]], [[OP1]]
|
|
; CHECK-NEXT: fcvtn v0.4h, [[OP2]]
|
|
; CHECK-NEXT: ret
|
|
%1 = uitofp <4 x i8> %a to <4 x half>
|
|
ret <4 x half> %1
|
|
}
|
|
|
|
|
|
define <4 x half> @uitofp_i16(<4 x i16> %a) #0 {
|
|
; CHECK-LABEL: uitofp_i16:
|
|
; CHECK-NEXT: ushll [[OP1:v[0-9]+\.4s]], v0.4h, #0
|
|
; CHECK-NEXT: ucvtf [[OP2:v[0-9]+\.4s]], [[OP1]]
|
|
; CHECK-NEXT: fcvtn v0.4h, [[OP2]]
|
|
; CHECK-NEXT: ret
|
|
%1 = uitofp <4 x i16> %a to <4 x half>
|
|
ret <4 x half> %1
|
|
}
|
|
|
|
|
|
define <4 x half> @uitofp_i32(<4 x i32> %a) #0 {
|
|
; CHECK-LABEL: uitofp_i32:
|
|
; CHECK-NEXT: ucvtf [[OP1:v[0-9]+\.4s]], v0.4s
|
|
; CHECK-NEXT: fcvtn v0.4h, [[OP1]]
|
|
%1 = uitofp <4 x i32> %a to <4 x half>
|
|
ret <4 x half> %1
|
|
}
|
|
|
|
|
|
define <4 x half> @uitofp_i64(<4 x i64> %a) #0 {
|
|
; CHECK-LABEL: uitofp_i64:
|
|
; CHECK-DAG: ucvtf [[OP1:v[0-9]+\.2d]], v0.2d
|
|
; CHECK-DAG: ucvtf [[OP2:v[0-9]+\.2d]], v1.2d
|
|
; CHECK-DAG: fcvtn [[OP3:v[0-9]+]].2s, [[OP1]]
|
|
; CHECK-NEXT: fcvtn2 [[OP3]].4s, [[OP2]]
|
|
; CHECK-NEXT: fcvtn v0.4h, [[OP3]].4s
|
|
%1 = uitofp <4 x i64> %a to <4 x half>
|
|
ret <4 x half> %1
|
|
}
|
|
|
|
attributes #0 = { nounwind }
|