blob: ad53487025699e41facd5dddccd4d4a15869ffe0 [file] [edit]
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
; RUN: llc -mtriple=amdgpu9.00-amd-amdhsa < %s | FileCheck -check-prefix=GFX9 %s
; RUN: llc -mtriple=amdgpu12.00-amd-amdhsa -mattr=+real-true16 < %s | FileCheck -check-prefixes=GFX12,GFX12-TRUE16 %s
; RUN: llc -mtriple=amdgpu12.00-amd-amdhsa -mattr=-real-true16 < %s | FileCheck -check-prefixes=GFX12,GFX12-FAKE16 %s
define <4 x i8> @test_bitextract_v4i8_const(<4 x b32> %src) {
; GFX9-LABEL: test_bitextract_v4i8_const:
; GFX9: ; %bb.0:
; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX9-NEXT: v_mov_b32_e32 v4, 8
; GFX9-NEXT: v_lshrrev_b32_sdwa v3, v4, v3 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
; GFX9-NEXT: v_lshrrev_b32_sdwa v1, v4, v1 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
; GFX9-NEXT: v_or_b32_sdwa v2, v2, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_1 src1_sel:DWORD
; GFX9-NEXT: v_or_b32_sdwa v0, v0, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_1 src1_sel:DWORD
; GFX9-NEXT: v_lshlrev_b32_e32 v3, 16, v2
; GFX9-NEXT: v_lshl_or_b32 v1, v2, 16, v1
; GFX9-NEXT: v_lshrrev_b32_e32 v1, 8, v1
; GFX9-NEXT: v_lshrrev_b32_e32 v3, 24, v3
; GFX9-NEXT: s_setpc_b64 s[30:31]
;
; GFX12-TRUE16-LABEL: test_bitextract_v4i8_const:
; GFX12-TRUE16: ; %bb.0:
; GFX12-TRUE16-NEXT: s_wait_loadcnt_dscnt 0x0
; GFX12-TRUE16-NEXT: s_wait_expcnt 0x0
; GFX12-TRUE16-NEXT: s_wait_samplecnt 0x0
; GFX12-TRUE16-NEXT: s_wait_bvhcnt 0x0
; GFX12-TRUE16-NEXT: s_wait_kmcnt 0x0
; GFX12-TRUE16-NEXT: v_lshrrev_b32_e32 v3, 8, v3
; GFX12-TRUE16-NEXT: v_lshrrev_b16 v0.h, 8, v2.l
; GFX12-TRUE16-NEXT: v_lshrrev_b32_e32 v1, 8, v1
; GFX12-TRUE16-NEXT: v_lshrrev_b16 v0.l, 8, v0.l
; GFX12-TRUE16-NEXT: s_delay_alu instid0(VALU_DEP_4) | instskip(NEXT) | instid1(VALU_DEP_3)
; GFX12-TRUE16-NEXT: v_lshlrev_b16 v2.l, 8, v3.l
; GFX12-TRUE16-NEXT: v_lshlrev_b16 v1.l, 8, v1.l
; GFX12-TRUE16-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2)
; GFX12-TRUE16-NEXT: v_or_b16 v2.l, v0.h, v2.l
; GFX12-TRUE16-NEXT: v_or_b16 v0.l, v0.l, v1.l
; GFX12-TRUE16-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_2)
; GFX12-TRUE16-NEXT: v_mov_b16_e32 v1.h, v2.l
; GFX12-TRUE16-NEXT: v_lshlrev_b32_e32 v3, 16, v2
; GFX12-TRUE16-NEXT: v_lshrrev_b32_e32 v1, 8, v1
; GFX12-TRUE16-NEXT: s_delay_alu instid0(VALU_DEP_2)
; GFX12-TRUE16-NEXT: v_lshrrev_b32_e32 v3, 24, v3
; GFX12-TRUE16-NEXT: s_setpc_b64 s[30:31]
;
; GFX12-FAKE16-LABEL: test_bitextract_v4i8_const:
; GFX12-FAKE16: ; %bb.0:
; GFX12-FAKE16-NEXT: s_wait_loadcnt_dscnt 0x0
; GFX12-FAKE16-NEXT: s_wait_expcnt 0x0
; GFX12-FAKE16-NEXT: s_wait_samplecnt 0x0
; GFX12-FAKE16-NEXT: s_wait_bvhcnt 0x0
; GFX12-FAKE16-NEXT: s_wait_kmcnt 0x0
; GFX12-FAKE16-NEXT: v_lshrrev_b32_e32 v3, 8, v3
; GFX12-FAKE16-NEXT: v_lshrrev_b32_e32 v1, 8, v1
; GFX12-FAKE16-NEXT: v_lshrrev_b16 v2, 8, v2
; GFX12-FAKE16-NEXT: v_lshrrev_b16 v0, 8, v0
; GFX12-FAKE16-NEXT: s_delay_alu instid0(VALU_DEP_4) | instskip(NEXT) | instid1(VALU_DEP_4)
; GFX12-FAKE16-NEXT: v_lshlrev_b16 v3, 8, v3
; GFX12-FAKE16-NEXT: v_lshlrev_b16 v1, 8, v1
; GFX12-FAKE16-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(NEXT) | instid1(VALU_DEP_2)
; GFX12-FAKE16-NEXT: v_or_b32_e32 v2, v2, v3
; GFX12-FAKE16-NEXT: v_and_b32_e32 v3, 0xffff, v1
; GFX12-FAKE16-NEXT: v_or_b32_e32 v0, v0, v1
; GFX12-FAKE16-NEXT: s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3)
; GFX12-FAKE16-NEXT: v_lshlrev_b32_e32 v4, 16, v2
; GFX12-FAKE16-NEXT: v_lshl_or_b32 v3, v2, 16, v3
; GFX12-FAKE16-NEXT: s_delay_alu instid0(VALU_DEP_1) | instskip(NEXT) | instid1(VALU_DEP_3)
; GFX12-FAKE16-NEXT: v_lshrrev_b32_e32 v1, 8, v3
; GFX12-FAKE16-NEXT: v_lshrrev_b32_e32 v3, 24, v4
; GFX12-FAKE16-NEXT: s_setpc_b64 s[30:31]
%result = bitextract <4 x i8>, <4 x b32> %src, <4 x i32> splat (i32 8)
ret <4 x i8> %result
}
define <4 x i8> @test_bitextract_v4i8_var(<4 x b32> %src, <4 x i32> %off) {
; GFX9-LABEL: test_bitextract_v4i8_var:
; GFX9: ; %bb.0:
; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX9-NEXT: v_lshrrev_b32_sdwa v3, v7, v3 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
; GFX9-NEXT: v_lshrrev_b32_e32 v2, v6, v2
; GFX9-NEXT: v_lshrrev_b32_sdwa v1, v5, v1 dst_sel:BYTE_1 dst_unused:UNUSED_PAD src0_sel:DWORD src1_sel:DWORD
; GFX9-NEXT: v_lshrrev_b32_e32 v0, v4, v0
; GFX9-NEXT: v_or_b32_sdwa v2, v2, v3 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
; GFX9-NEXT: v_or_b32_sdwa v0, v0, v1 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_0 src1_sel:DWORD
; GFX9-NEXT: v_lshlrev_b32_e32 v3, 16, v2
; GFX9-NEXT: v_lshl_or_b32 v1, v2, 16, v1
; GFX9-NEXT: v_lshrrev_b32_e32 v1, 8, v1
; GFX9-NEXT: v_lshrrev_b32_e32 v3, 24, v3
; GFX9-NEXT: s_setpc_b64 s[30:31]
;
; GFX12-TRUE16-LABEL: test_bitextract_v4i8_var:
; GFX12-TRUE16: ; %bb.0:
; GFX12-TRUE16-NEXT: s_wait_loadcnt_dscnt 0x0
; GFX12-TRUE16-NEXT: s_wait_expcnt 0x0
; GFX12-TRUE16-NEXT: s_wait_samplecnt 0x0
; GFX12-TRUE16-NEXT: s_wait_bvhcnt 0x0
; GFX12-TRUE16-NEXT: s_wait_kmcnt 0x0
; GFX12-TRUE16-NEXT: v_lshrrev_b32_e32 v3, v7, v3
; GFX12-TRUE16-NEXT: v_lshrrev_b32_e32 v6, v6, v2
; GFX12-TRUE16-NEXT: v_lshrrev_b32_e32 v1, v5, v1
; GFX12-TRUE16-NEXT: v_lshrrev_b32_e32 v0, v4, v0
; GFX12-TRUE16-NEXT: s_delay_alu instid0(VALU_DEP_4) | instskip(NEXT) | instid1(VALU_DEP_4)
; GFX12-TRUE16-NEXT: v_lshlrev_b16 v2.l, 8, v3.l
; GFX12-TRUE16-NEXT: v_and_b16 v2.h, v6.l, 0xff
; GFX12-TRUE16-NEXT: s_delay_alu instid0(VALU_DEP_4) | instskip(NEXT) | instid1(VALU_DEP_4)
; GFX12-TRUE16-NEXT: v_lshlrev_b16 v1.l, 8, v1.l
; GFX12-TRUE16-NEXT: v_and_b16 v0.l, v0.l, 0xff
; GFX12-TRUE16-NEXT: s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_2)
; GFX12-TRUE16-NEXT: v_or_b16 v2.l, v2.h, v2.l
; GFX12-TRUE16-NEXT: v_or_b16 v0.l, v0.l, v1.l
; GFX12-TRUE16-NEXT: s_delay_alu instid0(VALU_DEP_2) | instskip(SKIP_1) | instid1(VALU_DEP_2)
; GFX12-TRUE16-NEXT: v_mov_b16_e32 v1.h, v2.l
; GFX12-TRUE16-NEXT: v_lshlrev_b32_e32 v3, 16, v2
; GFX12-TRUE16-NEXT: v_lshrrev_b32_e32 v1, 8, v1
; GFX12-TRUE16-NEXT: s_delay_alu instid0(VALU_DEP_2)
; GFX12-TRUE16-NEXT: v_lshrrev_b32_e32 v3, 24, v3
; GFX12-TRUE16-NEXT: s_setpc_b64 s[30:31]
;
; GFX12-FAKE16-LABEL: test_bitextract_v4i8_var:
; GFX12-FAKE16: ; %bb.0:
; GFX12-FAKE16-NEXT: s_wait_loadcnt_dscnt 0x0
; GFX12-FAKE16-NEXT: s_wait_expcnt 0x0
; GFX12-FAKE16-NEXT: s_wait_samplecnt 0x0
; GFX12-FAKE16-NEXT: s_wait_bvhcnt 0x0
; GFX12-FAKE16-NEXT: s_wait_kmcnt 0x0
; GFX12-FAKE16-NEXT: v_lshrrev_b32_e32 v3, v7, v3
; GFX12-FAKE16-NEXT: v_lshrrev_b32_e32 v2, v6, v2
; GFX12-FAKE16-NEXT: v_lshrrev_b32_e32 v1, v5, v1
; GFX12-FAKE16-NEXT: v_lshrrev_b32_e32 v0, v4, v0
; GFX12-FAKE16-NEXT: s_delay_alu instid0(VALU_DEP_4) | instskip(NEXT) | instid1(VALU_DEP_4)
; GFX12-FAKE16-NEXT: v_lshlrev_b16 v3, 8, v3
; GFX12-FAKE16-NEXT: v_and_b32_e32 v2, 0xff, v2
; GFX12-FAKE16-NEXT: s_delay_alu instid0(VALU_DEP_4) | instskip(NEXT) | instid1(VALU_DEP_4)
; GFX12-FAKE16-NEXT: v_lshlrev_b16 v1, 8, v1
; GFX12-FAKE16-NEXT: v_and_b32_e32 v0, 0xff, v0
; GFX12-FAKE16-NEXT: s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3)
; GFX12-FAKE16-NEXT: v_or_b32_e32 v2, v2, v3
; GFX12-FAKE16-NEXT: v_and_b32_e32 v3, 0xffff, v1
; GFX12-FAKE16-NEXT: s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_3)
; GFX12-FAKE16-NEXT: v_or_b32_e32 v0, v0, v1
; GFX12-FAKE16-NEXT: v_lshlrev_b32_e32 v4, 16, v2
; GFX12-FAKE16-NEXT: s_delay_alu instid0(VALU_DEP_3) | instskip(NEXT) | instid1(VALU_DEP_1)
; GFX12-FAKE16-NEXT: v_lshl_or_b32 v3, v2, 16, v3
; GFX12-FAKE16-NEXT: v_lshrrev_b32_e32 v1, 8, v3
; GFX12-FAKE16-NEXT: s_delay_alu instid0(VALU_DEP_3)
; GFX12-FAKE16-NEXT: v_lshrrev_b32_e32 v3, 24, v4
; GFX12-FAKE16-NEXT: s_setpc_b64 s[30:31]
%result = bitextract <4 x i8>, <4 x b32> %src, <4 x i32> %off
ret <4 x i8> %result
}
define <2 x float> @test_bitextract_v2f32(<2 x b64> %src) {
; GFX9-LABEL: test_bitextract_v2f32:
; GFX9: ; %bb.0:
; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX9-NEXT: v_mov_b32_e32 v1, v3
; GFX9-NEXT: s_setpc_b64 s[30:31]
;
; GFX12-LABEL: test_bitextract_v2f32:
; GFX12: ; %bb.0:
; GFX12-NEXT: s_wait_loadcnt_dscnt 0x0
; GFX12-NEXT: s_wait_expcnt 0x0
; GFX12-NEXT: s_wait_samplecnt 0x0
; GFX12-NEXT: s_wait_bvhcnt 0x0
; GFX12-NEXT: s_wait_kmcnt 0x0
; GFX12-NEXT: v_mov_b32_e32 v1, v3
; GFX12-NEXT: s_setpc_b64 s[30:31]
%result = bitextract <2 x float>, <2 x b64> %src, <2 x i32> <i32 0, i32 32>
ret <2 x float> %result
}
define <2 x ptr> @test_bitextract_v2ptr(<2 x b64> %src) {
; GFX9-LABEL: test_bitextract_v2ptr:
; GFX9: ; %bb.0:
; GFX9-NEXT: s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; GFX9-NEXT: s_setpc_b64 s[30:31]
;
; GFX12-LABEL: test_bitextract_v2ptr:
; GFX12: ; %bb.0:
; GFX12-NEXT: s_wait_loadcnt_dscnt 0x0
; GFX12-NEXT: s_wait_expcnt 0x0
; GFX12-NEXT: s_wait_samplecnt 0x0
; GFX12-NEXT: s_wait_bvhcnt 0x0
; GFX12-NEXT: s_wait_kmcnt 0x0
; GFX12-NEXT: s_setpc_b64 s[30:31]
%result = bitextract <2 x ptr>, <2 x b64> %src, <2 x i32> zeroinitializer
ret <2 x ptr> %result
}