Page MenuHomePhabricator
Differential D109253 Diff 370649 llvm/test/CodeGen/AArch64/fptoui-sat-vector.ll

Changeset View

Standalone View

View Options

llvm/test/CodeGen/AArch64/fptoui-sat-vector.ll

Show First 20 Lines • Show All 263 Lines • ▼ Show 20 Lines
; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload ; CHECK-NEXT: ldr q0, [sp] // 16-byte Folded Reload
; CHECK-NEXT: cmp w19, #0 ; CHECK-NEXT: cmp w19, #0
; CHECK-NEXT: csel w19, wzr, w0, lt ; CHECK-NEXT: csel w19, wzr, w0, lt
; CHECK-NEXT: bl __gttf2 ; CHECK-NEXT: bl __gttf2
; CHECK-NEXT: cmp w0, #0 ; CHECK-NEXT: cmp w0, #0
; CHECK-NEXT: csinv w8, w19, wzr, le ; CHECK-NEXT: csinv w8, w19, wzr, le
; CHECK-NEXT: ldp x30, x19, [sp, #16] // 16-byte Folded Reload ; CHECK-NEXT: ldp x30, x19, [sp, #16] // 16-byte Folded Reload
; CHECK-NEXT: fmov s0, w8 ; CHECK-NEXT: fmov s0, w8
; CHECK-NEXT: .cfi_adjust_cfa_offset 16
; CHECK-NEXT: add sp, sp, #32 ; CHECK-NEXT: add sp, sp, #32
; CHECK-NEXT: ret ; CHECK-NEXT: ret
%x = call <1 x i32> @llvm.fptoui.sat.v1f128.v1i32(<1 x fp128> %f) %x = call <1 x i32> @llvm.fptoui.sat.v1f128.v1i32(<1 x fp128> %f)
ret <1 x i32> %x ret <1 x i32> %x
} }
define <2 x i32> @test_unsigned_v2f128_v2i32(<2 x fp128> %f) { define <2 x i32> @test_unsigned_v2f128_v2i32(<2 x fp128> %f) {
; CHECK-LABEL: test_unsigned_v2f128_v2i32: ; CHECK-LABEL: test_unsigned_v2f128_v2i32:
Show All 37 Lines
; CHECK-NEXT: bl __gttf2 ; CHECK-NEXT: bl __gttf2
; CHECK-NEXT: cmp w0, #0 ; CHECK-NEXT: cmp w0, #0
; CHECK-NEXT: csinv w8, w19, wzr, le ; CHECK-NEXT: csinv w8, w19, wzr, le
; CHECK-NEXT: fmov s0, w8 ; CHECK-NEXT: fmov s0, w8
; CHECK-NEXT: mov v0.s[1], w20 ; CHECK-NEXT: mov v0.s[1], w20
; CHECK-NEXT: ldp x20, x19, [sp, #80] // 16-byte Folded Reload ; CHECK-NEXT: ldp x20, x19, [sp, #80] // 16-byte Folded Reload
; CHECK-NEXT: ldr x30, [sp, #64] // 8-byte Folded Reload ; CHECK-NEXT: ldr x30, [sp, #64] // 8-byte Folded Reload
; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0 ; CHECK-NEXT: // kill: def $d0 killed $d0 killed $q0
; CHECK-NEXT: .cfi_adjust_cfa_offset 64
; CHECK-NEXT: .cfi_adjust_cfa_offset 64
; CHECK-NEXT: add sp, sp, #96 ; CHECK-NEXT: add sp, sp, #96
; CHECK-NEXT: ret ; CHECK-NEXT: ret
%x = call <2 x i32> @llvm.fptoui.sat.v2f128.v2i32(<2 x fp128> %f) %x = call <2 x i32> @llvm.fptoui.sat.v2f128.v2i32(<2 x fp128> %f)
ret <2 x i32> %x ret <2 x i32> %x
} }
define <3 x i32> @test_unsigned_v3f128_v3i32(<3 x fp128> %f) { define <3 x i32> @test_unsigned_v3f128_v3i32(<3 x fp128> %f) {
; CHECK-LABEL: test_unsigned_v3f128_v3i32: ; CHECK-LABEL: test_unsigned_v3f128_v3i32:
▲ Show 20 Lines • Show All 52 Lines • ▼ Show 20 Lines
; CHECK-NEXT: csel w19, wzr, w0, lt ; CHECK-NEXT: csel w19, wzr, w0, lt
; CHECK-NEXT: bl __gttf2 ; CHECK-NEXT: bl __gttf2
; CHECK-NEXT: cmp w0, #0 ; CHECK-NEXT: cmp w0, #0
; CHECK-NEXT: ldr q0, [sp, #48] // 16-byte Folded Reload ; CHECK-NEXT: ldr q0, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: csinv w8, w19, wzr, le ; CHECK-NEXT: csinv w8, w19, wzr, le
; CHECK-NEXT: ldp x20, x19, [sp, #96] // 16-byte Folded Reload ; CHECK-NEXT: ldp x20, x19, [sp, #96] // 16-byte Folded Reload
; CHECK-NEXT: ldr x30, [sp, #80] // 8-byte Folded Reload ; CHECK-NEXT: ldr x30, [sp, #80] // 8-byte Folded Reload
; CHECK-NEXT: mov v0.s[2], w8 ; CHECK-NEXT: mov v0.s[2], w8
; CHECK-NEXT: .cfi_adjust_cfa_offset 80
; CHECK-NEXT: .cfi_adjust_cfa_offset 80
; CHECK-NEXT: add sp, sp, #112 ; CHECK-NEXT: add sp, sp, #112
; CHECK-NEXT: ret ; CHECK-NEXT: ret
%x = call <3 x i32> @llvm.fptoui.sat.v3f128.v3i32(<3 x fp128> %f) %x = call <3 x i32> @llvm.fptoui.sat.v3f128.v3i32(<3 x fp128> %f)
ret <3 x i32> %x ret <3 x i32> %x
} }
define <4 x i32> @test_unsigned_v4f128_v4i32(<4 x fp128> %f) { define <4 x i32> @test_unsigned_v4f128_v4i32(<4 x fp128> %f) {
; CHECK-LABEL: test_unsigned_v4f128_v4i32: ; CHECK-LABEL: test_unsigned_v4f128_v4i32:
▲ Show 20 Lines • Show All 68 Lines • ▼ Show 20 Lines
; CHECK-NEXT: csel w19, wzr, w0, lt ; CHECK-NEXT: csel w19, wzr, w0, lt
; CHECK-NEXT: bl __gttf2 ; CHECK-NEXT: bl __gttf2
; CHECK-NEXT: cmp w0, #0 ; CHECK-NEXT: cmp w0, #0
; CHECK-NEXT: ldr q0, [sp, #16] // 16-byte Folded Reload ; CHECK-NEXT: ldr q0, [sp, #16] // 16-byte Folded Reload
; CHECK-NEXT: csinv w8, w19, wzr, le ; CHECK-NEXT: csinv w8, w19, wzr, le
; CHECK-NEXT: ldp x20, x19, [sp, #112] // 16-byte Folded Reload ; CHECK-NEXT: ldp x20, x19, [sp, #112] // 16-byte Folded Reload
; CHECK-NEXT: ldr x30, [sp, #96] // 8-byte Folded Reload ; CHECK-NEXT: ldr x30, [sp, #96] // 8-byte Folded Reload
; CHECK-NEXT: mov v0.s[3], w8 ; CHECK-NEXT: mov v0.s[3], w8
; CHECK-NEXT: .cfi_adjust_cfa_offset 96
; CHECK-NEXT: .cfi_adjust_cfa_offset 96
; CHECK-NEXT: add sp, sp, #128 ; CHECK-NEXT: add sp, sp, #128
; CHECK-NEXT: ret ; CHECK-NEXT: ret
%x = call <4 x i32> @llvm.fptoui.sat.v4f128.v4i32(<4 x fp128> %f) %x = call <4 x i32> @llvm.fptoui.sat.v4f128.v4i32(<4 x fp128> %f)
ret <4 x i32> %x ret <4 x i32> %x
} }
; ;
; FP16 to unsigned 32-bit -- Vector size variation ; FP16 to unsigned 32-bit -- Vector size variation
▲ Show 20 Lines • Show All 461 Lines • ▼ Show 20 Lines
; CHECK-NEXT: fcmp s0, s9 ; CHECK-NEXT: fcmp s0, s9
; CHECK-NEXT: csinv x8, x8, xzr, le ; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: csel x1, x21, x9, gt ; CHECK-NEXT: csel x1, x21, x9, gt
; CHECK-NEXT: ldp x30, x21, [sp, #32] // 16-byte Folded Reload ; CHECK-NEXT: ldp x30, x21, [sp, #32] // 16-byte Folded Reload
; CHECK-NEXT: ldp d9, d8, [sp, #16] // 16-byte Folded Reload ; CHECK-NEXT: ldp d9, d8, [sp, #16] // 16-byte Folded Reload
; CHECK-NEXT: fmov d0, x8 ; CHECK-NEXT: fmov d0, x8
; CHECK-NEXT: mov v0.d[1], x1 ; CHECK-NEXT: mov v0.d[1], x1
; CHECK-NEXT: fmov x0, d0 ; CHECK-NEXT: fmov x0, d0
; CHECK-NEXT: .cfi_adjust_cfa_offset 16
; CHECK-NEXT: .cfi_adjust_cfa_offset 16
; CHECK-NEXT: .cfi_adjust_cfa_offset 16
; CHECK-NEXT: add sp, sp, #64 ; CHECK-NEXT: add sp, sp, #64
; CHECK-NEXT: ret ; CHECK-NEXT: ret
%x = call <2 x i100> @llvm.fptoui.sat.v2f32.v2i100(<2 x float> %f) %x = call <2 x i100> @llvm.fptoui.sat.v2f32.v2i100(<2 x float> %f)
ret <2 x i100> %x ret <2 x i100> %x
} }
define <2 x i128> @test_unsigned_v2f32_v2i128(<2 x float> %f) { define <2 x i128> @test_unsigned_v2f32_v2i128(<2 x float> %f) {
; CHECK-LABEL: test_unsigned_v2f32_v2i128: ; CHECK-LABEL: test_unsigned_v2f32_v2i128:
Show All 34 Lines
; CHECK-NEXT: fcmp s0, s9 ; CHECK-NEXT: fcmp s0, s9
; CHECK-NEXT: csinv x8, x8, xzr, le ; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: ldr x30, [sp, #32] // 8-byte Folded Reload ; CHECK-NEXT: ldr x30, [sp, #32] // 8-byte Folded Reload
; CHECK-NEXT: ldp d9, d8, [sp, #16] // 16-byte Folded Reload ; CHECK-NEXT: ldp d9, d8, [sp, #16] // 16-byte Folded Reload
; CHECK-NEXT: csinv x1, x9, xzr, le ; CHECK-NEXT: csinv x1, x9, xzr, le
; CHECK-NEXT: fmov d0, x8 ; CHECK-NEXT: fmov d0, x8
; CHECK-NEXT: mov v0.d[1], x1 ; CHECK-NEXT: mov v0.d[1], x1
; CHECK-NEXT: fmov x0, d0 ; CHECK-NEXT: fmov x0, d0
; CHECK-NEXT: .cfi_adjust_cfa_offset 16
; CHECK-NEXT: .cfi_adjust_cfa_offset 16
; CHECK-NEXT: .cfi_adjust_cfa_offset 16
; CHECK-NEXT: add sp, sp, #64 ; CHECK-NEXT: add sp, sp, #64
; CHECK-NEXT: ret ; CHECK-NEXT: ret
%x = call <2 x i128> @llvm.fptoui.sat.v2f32.v2i128(<2 x float> %f) %x = call <2 x i128> @llvm.fptoui.sat.v2f32.v2i128(<2 x float> %f)
ret <2 x i128> %x ret <2 x i128> %x
} }
; ;
; 2-Vector double to unsigned integer -- result size variation ; 2-Vector double to unsigned integer -- result size variation
▲ Show 20 Lines • Show All 201 Lines • ▼ Show 20 Lines
; CHECK-NEXT: fcmp d0, d9 ; CHECK-NEXT: fcmp d0, d9
; CHECK-NEXT: csinv x8, x8, xzr, le ; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: csel x1, x21, x9, gt ; CHECK-NEXT: csel x1, x21, x9, gt
; CHECK-NEXT: ldp x30, x21, [sp, #32] // 16-byte Folded Reload ; CHECK-NEXT: ldp x30, x21, [sp, #32] // 16-byte Folded Reload
; CHECK-NEXT: ldp d9, d8, [sp, #16] // 16-byte Folded Reload ; CHECK-NEXT: ldp d9, d8, [sp, #16] // 16-byte Folded Reload
; CHECK-NEXT: fmov d0, x8 ; CHECK-NEXT: fmov d0, x8
; CHECK-NEXT: mov v0.d[1], x1 ; CHECK-NEXT: mov v0.d[1], x1
; CHECK-NEXT: fmov x0, d0 ; CHECK-NEXT: fmov x0, d0
; CHECK-NEXT: .cfi_adjust_cfa_offset 16
; CHECK-NEXT: .cfi_adjust_cfa_offset 16
; CHECK-NEXT: .cfi_adjust_cfa_offset 16
; CHECK-NEXT: add sp, sp, #64 ; CHECK-NEXT: add sp, sp, #64
; CHECK-NEXT: ret ; CHECK-NEXT: ret
%x = call <2 x i100> @llvm.fptoui.sat.v2f64.v2i100(<2 x double> %f) %x = call <2 x i100> @llvm.fptoui.sat.v2f64.v2i100(<2 x double> %f)
ret <2 x i100> %x ret <2 x i100> %x
} }
define <2 x i128> @test_unsigned_v2f64_v2i128(<2 x double> %f) { define <2 x i128> @test_unsigned_v2f64_v2i128(<2 x double> %f) {
; CHECK-LABEL: test_unsigned_v2f64_v2i128: ; CHECK-LABEL: test_unsigned_v2f64_v2i128:
Show All 33 Lines
; CHECK-NEXT: fcmp d0, d9 ; CHECK-NEXT: fcmp d0, d9
; CHECK-NEXT: csinv x8, x8, xzr, le ; CHECK-NEXT: csinv x8, x8, xzr, le
; CHECK-NEXT: ldr x30, [sp, #32] // 8-byte Folded Reload ; CHECK-NEXT: ldr x30, [sp, #32] // 8-byte Folded Reload
; CHECK-NEXT: ldp d9, d8, [sp, #16] // 16-byte Folded Reload ; CHECK-NEXT: ldp d9, d8, [sp, #16] // 16-byte Folded Reload
; CHECK-NEXT: csinv x1, x9, xzr, le ; CHECK-NEXT: csinv x1, x9, xzr, le
; CHECK-NEXT: fmov d0, x8 ; CHECK-NEXT: fmov d0, x8
; CHECK-NEXT: mov v0.d[1], x1 ; CHECK-NEXT: mov v0.d[1], x1
; CHECK-NEXT: fmov x0, d0 ; CHECK-NEXT: fmov x0, d0
; CHECK-NEXT: .cfi_adjust_cfa_offset 16
; CHECK-NEXT: .cfi_adjust_cfa_offset 16
; CHECK-NEXT: .cfi_adjust_cfa_offset 16
; CHECK-NEXT: add sp, sp, #64 ; CHECK-NEXT: add sp, sp, #64
; CHECK-NEXT: ret ; CHECK-NEXT: ret
%x = call <2 x i128> @llvm.fptoui.sat.v2f64.v2i128(<2 x double> %f) %x = call <2 x i128> @llvm.fptoui.sat.v2f64.v2i128(<2 x double> %f)
ret <2 x i128> %x ret <2 x i128> %x
} }
; ;
; 4-Vector half to unsigned integer -- result size variation ; 4-Vector half to unsigned integer -- result size variation
▲ Show 20 Lines • Show All 373 Lines • ▼ Show 20 Lines
; CHECK-NEXT: ldp x20, x19, [sp, #80] // 16-byte Folded Reload ; CHECK-NEXT: ldp x20, x19, [sp, #80] // 16-byte Folded Reload
; CHECK-NEXT: ldp x22, x21, [sp, #64] // 16-byte Folded Reload ; CHECK-NEXT: ldp x22, x21, [sp, #64] // 16-byte Folded Reload
; CHECK-NEXT: ldp x24, x23, [sp, #48] // 16-byte Folded Reload ; CHECK-NEXT: ldp x24, x23, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: ldp x30, x25, [sp, #32] // 16-byte Folded Reload ; CHECK-NEXT: ldp x30, x25, [sp, #32] // 16-byte Folded Reload
; CHECK-NEXT: ldp d9, d8, [sp, #16] // 16-byte Folded Reload ; CHECK-NEXT: ldp d9, d8, [sp, #16] // 16-byte Folded Reload
; CHECK-NEXT: fmov d0, x8 ; CHECK-NEXT: fmov d0, x8
; CHECK-NEXT: mov v0.d[1], x1 ; CHECK-NEXT: mov v0.d[1], x1
; CHECK-NEXT: fmov x0, d0 ; CHECK-NEXT: fmov x0, d0
; CHECK-NEXT: .cfi_adjust_cfa_offset 16
; CHECK-NEXT: .cfi_adjust_cfa_offset 16
; CHECK-NEXT: .cfi_adjust_cfa_offset 16
; CHECK-NEXT: .cfi_adjust_cfa_offset 16
; CHECK-NEXT: .cfi_adjust_cfa_offset 16
; CHECK-NEXT: add sp, sp, #96 ; CHECK-NEXT: add sp, sp, #96
; CHECK-NEXT: ret ; CHECK-NEXT: ret
%x = call <4 x i100> @llvm.fptoui.sat.v4f16.v4i100(<4 x half> %f) %x = call <4 x i100> @llvm.fptoui.sat.v4f16.v4i100(<4 x half> %f)
ret <4 x i100> %x ret <4 x i100> %x
} }
define <4 x i128> @test_unsigned_v4f16_v4i128(<4 x half> %f) { define <4 x i128> @test_unsigned_v4f16_v4i128(<4 x half> %f) {
; CHECK-LABEL: test_unsigned_v4f16_v4i128: ; CHECK-LABEL: test_unsigned_v4f16_v4i128:
▲ Show 20 Lines • Show All 69 Lines • ▼ Show 20 Lines
; CHECK-NEXT: ldp x22, x21, [sp, #64] // 16-byte Folded Reload ; CHECK-NEXT: ldp x22, x21, [sp, #64] // 16-byte Folded Reload
; CHECK-NEXT: ldp x24, x23, [sp, #48] // 16-byte Folded Reload ; CHECK-NEXT: ldp x24, x23, [sp, #48] // 16-byte Folded Reload
; CHECK-NEXT: ldr x30, [sp, #32] // 8-byte Folded Reload ; CHECK-NEXT: ldr x30, [sp, #32] // 8-byte Folded Reload
; CHECK-NEXT: ldp d9, d8, [sp, #16] // 16-byte Folded Reload ; CHECK-NEXT: ldp d9, d8, [sp, #16] // 16-byte Folded Reload
; CHECK-NEXT: csinv x1, x9, xzr, le ; CHECK-NEXT: csinv x1, x9, xzr, le
; CHECK-NEXT: fmov d0, x8 ; CHECK-NEXT: fmov d0, x8
; CHECK-NEXT: mov v0.d[1], x1 ; CHECK-NEXT: mov v0.d[1], x1
; CHECK-NEXT: fmov x0, d0 ; CHECK-NEXT: fmov x0, d0
; CHECK-NEXT: .cfi_adjust_cfa_offset 16
; CHECK-NEXT: .cfi_adjust_cfa_offset 16
; CHECK-NEXT: .cfi_adjust_cfa_offset 16
; CHECK-NEXT: .cfi_adjust_cfa_offset 16
; CHECK-NEXT: .cfi_adjust_cfa_offset 16
; CHECK-NEXT: add sp, sp, #96 ; CHECK-NEXT: add sp, sp, #96
; CHECK-NEXT: ret ; CHECK-NEXT: ret
%x = call <4 x i128> @llvm.fptoui.sat.v4f16.v4i128(<4 x half> %f) %x = call <4 x i128> @llvm.fptoui.sat.v4f16.v4i128(<4 x half> %f)
ret <4 x i128> %x ret <4 x i128> %x
} }