let PrintMethod = "printSendMsg";
}
-def isGCN : Predicate<"Subtarget.getGeneration() "
+def isGCN : Predicate<"Subtarget->getGeneration() "
">= AMDGPUSubtarget::SOUTHERN_ISLANDS">;
def isSICI : Predicate<
- "Subtarget.getGeneration() == AMDGPUSubtarget::SOUTHERN_ISLANDS ||"
- "Subtarget.getGeneration() == AMDGPUSubtarget::SEA_ISLANDS"
+ "Subtarget->getGeneration() == AMDGPUSubtarget::SOUTHERN_ISLANDS ||"
+ "Subtarget->getGeneration() == AMDGPUSubtarget::SEA_ISLANDS"
>;
-def isCI : Predicate<"Subtarget.getGeneration() "
+def isCI : Predicate<"Subtarget->getGeneration() "
">= AMDGPUSubtarget::SEA_ISLANDS">;
def isVI : Predicate <
- "Subtarget.getGeneration() >= AMDGPUSubtarget::VOLCANIC_ISLANDS"
+ "Subtarget->getGeneration() >= AMDGPUSubtarget::VOLCANIC_ISLANDS"
>;
def HasFlatAddressSpace : Predicate<"Subtarget.hasFlatAddressSpace()">;
defm S_BREV_B64 : SOP1_64 <sop1<0x0c, 0x09>, "s_brev_b64", []>;
let Defs = [SCC] in {
- //defm S_BCNT0_I32_B32 : SOP1_BCNT0 <sop1<0x0d, 0x0a>, "s_bcnt0_i32_b32", []>;
- //defm S_BCNT0_I32_B64 : SOP1_BCNT0 <sop1<0x0e, 0x0b>, "s_bcnt0_i32_b64", []>;
+ defm S_BCNT0_I32_B32 : SOP1_32 <sop1<0x0d, 0x0a>, "s_bcnt0_i32_b32", []>;
+ defm S_BCNT0_I32_B64 : SOP1_32_64 <sop1<0x0e, 0x0b>, "s_bcnt0_i32_b64", []>;
defm S_BCNT1_I32_B32 : SOP1_32 <sop1<0x0f, 0x0c>, "s_bcnt1_i32_b32",
[(set i32:$dst, (ctpop i32:$src0))]
>;
defm S_BCNT1_I32_B64 : SOP1_32_64 <sop1<0x10, 0x0d>, "s_bcnt1_i32_b64", []>;
} // End Defs = [SCC]
-//defm S_FF0_I32_B32 : SOP1_32 <sop1<0x11, 0x0e>, "s_ff0_i32_b32", []>;
-//defm S_FF0_I32_B64 : SOP1_FF0 <sop1<0x12, 0x0f>, "s_ff0_i32_b64", []>;
+defm S_FF0_I32_B32 : SOP1_32 <sop1<0x11, 0x0e>, "s_ff0_i32_b32", []>;
+defm S_FF0_I32_B64 : SOP1_32_64 <sop1<0x12, 0x0f>, "s_ff0_i32_b64", []>;
defm S_FF1_I32_B32 : SOP1_32 <sop1<0x13, 0x10>, "s_ff1_i32_b32",
[(set i32:$dst, (cttz_zero_undef i32:$src0))]
>;
-////defm S_FF1_I32_B64 : SOP1_FF1 <sop1<0x14, 0x11>, "s_ff1_i32_b64", []>;
+defm S_FF1_I32_B64 : SOP1_32_64 <sop1<0x14, 0x11>, "s_ff1_i32_b64", []>;
defm S_FLBIT_I32_B32 : SOP1_32 <sop1<0x15, 0x12>, "s_flbit_i32_b32",
[(set i32:$dst, (ctlz_zero_undef i32:$src0))]
>;
-//defm S_FLBIT_I32_B64 : SOP1_32 <sop1<0x16, 0x13>, "s_flbit_i32_b64", []>;
-defm S_FLBIT_I32 : SOP1_32 <sop1<0x17, 0x14>, "s_flbit_i32", []>;
-//defm S_FLBIT_I32_I64 : SOP1_32 <sop1<0x18, 0x15>, "s_flbit_i32_i64", []>;
+defm S_FLBIT_I32_B64 : SOP1_32_64 <sop1<0x16, 0x13>, "s_flbit_i32_b64", []>;
+defm S_FLBIT_I32 : SOP1_32 <sop1<0x17, 0x14>, "s_flbit_i32",
+ [(set i32:$dst, (int_AMDGPU_flbit_i32 i32:$src0))]
+>;
+defm S_FLBIT_I32_I64 : SOP1_32_64 <sop1<0x18, 0x15>, "s_flbit_i32_i64", []>;
defm S_SEXT_I32_I8 : SOP1_32 <sop1<0x19, 0x16>, "s_sext_i32_i8",
[(set i32:$dst, (sext_inreg i32:$src0, i8))]
>;
[(set i32:$dst, (sext_inreg i32:$src0, i16))]
>;
-////defm S_BITSET0_B32 : SOP1_BITSET0 <sop1<0x1b, 0x18>, "s_bitset0_b32", []>;
-////defm S_BITSET0_B64 : SOP1_BITSET0 <sop1<0x1c, 0x19>, "s_bitset0_b64", []>;
-////defm S_BITSET1_B32 : SOP1_BITSET1 <sop1<0x1d, 0x1a>, "s_bitset1_b32", []>;
-////defm S_BITSET1_B64 : SOP1_BITSET1 <sop1<0x1e, 0x1b>, "s_bitset1_b64", []>;
+defm S_BITSET0_B32 : SOP1_32 <sop1<0x1b, 0x18>, "s_bitset0_b32", []>;
+defm S_BITSET0_B64 : SOP1_64 <sop1<0x1c, 0x19>, "s_bitset0_b64", []>;
+defm S_BITSET1_B32 : SOP1_32 <sop1<0x1d, 0x1a>, "s_bitset1_b32", []>;
+defm S_BITSET1_B64 : SOP1_64 <sop1<0x1e, 0x1b>, "s_bitset1_b64", []>;
defm S_GETPC_B64 : SOP1_64_0 <sop1<0x1f, 0x1c>, "s_getpc_b64", []>;
defm S_SETPC_B64 : SOP1_64 <sop1<0x20, 0x1d>, "s_setpc_b64", []>;
defm S_SWAPPC_B64 : SOP1_64 <sop1<0x21, 0x1e>, "s_swappc_b64", []>;
defm S_MOVRELS_B64 : SOP1_64 <sop1<0x2f, 0x2b>, "s_movrels_b64", []>;
defm S_MOVRELD_B32 : SOP1_32 <sop1<0x30, 0x2c>, "s_movreld_b32", []>;
defm S_MOVRELD_B64 : SOP1_64 <sop1<0x31, 0x2d>, "s_movreld_b64", []>;
-//defm S_CBRANCH_JOIN : SOP1_ <sop1<0x32, 0x2e>, "s_cbranch_join", []>;
+defm S_CBRANCH_JOIN : SOP1_1 <sop1<0x32, 0x2e>, "s_cbranch_join", []>;
defm S_MOV_REGRD_B32 : SOP1_32 <sop1<0x33, 0x2f>, "s_mov_regrd_b32", []>;
let Defs = [SCC] in {
defm S_ABS_I32 : SOP1_32 <sop1<0x34, 0x30>, "s_abs_i32", []>;
defm S_BFE_I64 : SOP2_64_32 <sop2<0x2a, 0x28>, "s_bfe_i64", []>;
} // End Defs = [SCC]
-//defm S_CBRANCH_G_FORK : SOP2_ <sop2<0x2b, 0x29>, "s_cbranch_g_fork", []>;
+let sdst = 0 in {
+defm S_CBRANCH_G_FORK : SOP2_m <
+ sop2<0x2b, 0x29>, "s_cbranch_g_fork", (outs),
+ (ins SReg_64:$src0, SReg_64:$src1), "s_cbranch_g_fork $src0, $src1", []
+>;
+}
+
let Defs = [SCC] in {
defm S_ABSDIFF_I32 : SOP2_32 <sop2<0x2c, 0x2a>, "s_absdiff_i32", []>;
} // End Defs = [SCC]
// DS Instructions
//===----------------------------------------------------------------------===//
-
-def DS_ADD_U32 : DS_1A1D_NORET <0x0, "ds_add_u32", VGPR_32>;
-def DS_SUB_U32 : DS_1A1D_NORET <0x1, "ds_sub_u32", VGPR_32>;
-def DS_RSUB_U32 : DS_1A1D_NORET <0x2, "ds_rsub_u32", VGPR_32>;
-def DS_INC_U32 : DS_1A1D_NORET <0x3, "ds_inc_u32", VGPR_32>;
-def DS_DEC_U32 : DS_1A1D_NORET <0x4, "ds_dec_u32", VGPR_32>;
-def DS_MIN_I32 : DS_1A1D_NORET <0x5, "ds_min_i32", VGPR_32>;
-def DS_MAX_I32 : DS_1A1D_NORET <0x6, "ds_max_i32", VGPR_32>;
-def DS_MIN_U32 : DS_1A1D_NORET <0x7, "ds_min_u32", VGPR_32>;
-def DS_MAX_U32 : DS_1A1D_NORET <0x8, "ds_max_u32", VGPR_32>;
-def DS_AND_B32 : DS_1A1D_NORET <0x9, "ds_and_b32", VGPR_32>;
-def DS_OR_B32 : DS_1A1D_NORET <0xa, "ds_or_b32", VGPR_32>;
-def DS_XOR_B32 : DS_1A1D_NORET <0xb, "ds_xor_b32", VGPR_32>;
-def DS_MSKOR_B32 : DS_1A1D_NORET <0xc, "ds_mskor_b32", VGPR_32>;
-def DS_CMPST_B32 : DS_1A2D_NORET <0x10, "ds_cmpst_b32", VGPR_32>;
-def DS_CMPST_F32 : DS_1A2D_NORET <0x11, "ds_cmpst_f32", VGPR_32>;
-def DS_MIN_F32 : DS_1A1D_NORET <0x12, "ds_min_f32", VGPR_32>;
-def DS_MAX_F32 : DS_1A1D_NORET <0x13, "ds_max_f32", VGPR_32>;
-
-def DS_ADD_RTN_U32 : DS_1A1D_RET <0x20, "ds_add_rtn_u32", VGPR_32, "ds_add_u32">;
-def DS_SUB_RTN_U32 : DS_1A1D_RET <0x21, "ds_sub_rtn_u32", VGPR_32, "ds_sub_u32">;
-def DS_RSUB_RTN_U32 : DS_1A1D_RET <0x22, "ds_rsub_rtn_u32", VGPR_32, "ds_rsub_u32">;
-def DS_INC_RTN_U32 : DS_1A1D_RET <0x23, "ds_inc_rtn_u32", VGPR_32, "ds_inc_u32">;
-def DS_DEC_RTN_U32 : DS_1A1D_RET <0x24, "ds_dec_rtn_u32", VGPR_32, "ds_dec_u32">;
-def DS_MIN_RTN_I32 : DS_1A1D_RET <0x25, "ds_min_rtn_i32", VGPR_32, "ds_min_i32">;
-def DS_MAX_RTN_I32 : DS_1A1D_RET <0x26, "ds_max_rtn_i32", VGPR_32, "ds_max_i32">;
-def DS_MIN_RTN_U32 : DS_1A1D_RET <0x27, "ds_min_rtn_u32", VGPR_32, "ds_min_u32">;
-def DS_MAX_RTN_U32 : DS_1A1D_RET <0x28, "ds_max_rtn_u32", VGPR_32, "ds_max_u32">;
-def DS_AND_RTN_B32 : DS_1A1D_RET <0x29, "ds_and_rtn_b32", VGPR_32, "ds_and_b32">;
-def DS_OR_RTN_B32 : DS_1A1D_RET <0x2a, "ds_or_rtn_b32", VGPR_32, "ds_or_b32">;
-def DS_XOR_RTN_B32 : DS_1A1D_RET <0x2b, "ds_xor_rtn_b32", VGPR_32, "ds_xor_b32">;
-def DS_MSKOR_RTN_B32 : DS_1A1D_RET <0x2c, "ds_mskor_rtn_b32", VGPR_32, "ds_mskor_b32">;
-def DS_WRXCHG_RTN_B32 : DS_1A1D_RET <0x2d, "ds_wrxchg_rtn_b32", VGPR_32>;
-//def DS_WRXCHG2_RTN_B32 : DS_2A0D_RET <0x2e, "ds_wrxchg2_rtn_b32", VGPR_32, "ds_wrxchg2_b32">;
-//def DS_WRXCHG2ST64_RTN_B32 : DS_2A0D_RET <0x2f, "ds_wrxchg2_rtn_b32", VGPR_32, "ds_wrxchg2st64_b32">;
-def DS_CMPST_RTN_B32 : DS_1A2D_RET <0x30, "ds_cmpst_rtn_b32", VGPR_32, "ds_cmpst_b32">;
-def DS_CMPST_RTN_F32 : DS_1A2D_RET <0x31, "ds_cmpst_rtn_f32", VGPR_32, "ds_cmpst_f32">;
-def DS_MIN_RTN_F32 : DS_1A1D_RET <0x32, "ds_min_rtn_f32", VGPR_32, "ds_min_f32">;
-def DS_MAX_RTN_F32 : DS_1A1D_RET <0x33, "ds_max_rtn_f32", VGPR_32, "ds_max_f32">;
-
+defm DS_ADD_U32 : DS_1A1D_NORET <0x0, "ds_add_u32", VGPR_32>;
+defm DS_SUB_U32 : DS_1A1D_NORET <0x1, "ds_sub_u32", VGPR_32>;
+defm DS_RSUB_U32 : DS_1A1D_NORET <0x2, "ds_rsub_u32", VGPR_32>;
+defm DS_INC_U32 : DS_1A1D_NORET <0x3, "ds_inc_u32", VGPR_32>;
+defm DS_DEC_U32 : DS_1A1D_NORET <0x4, "ds_dec_u32", VGPR_32>;
+defm DS_MIN_I32 : DS_1A1D_NORET <0x5, "ds_min_i32", VGPR_32>;
+defm DS_MAX_I32 : DS_1A1D_NORET <0x6, "ds_max_i32", VGPR_32>;
+defm DS_MIN_U32 : DS_1A1D_NORET <0x7, "ds_min_u32", VGPR_32>;
+defm DS_MAX_U32 : DS_1A1D_NORET <0x8, "ds_max_u32", VGPR_32>;
+defm DS_AND_B32 : DS_1A1D_NORET <0x9, "ds_and_b32", VGPR_32>;
+defm DS_OR_B32 : DS_1A1D_NORET <0xa, "ds_or_b32", VGPR_32>;
+defm DS_XOR_B32 : DS_1A1D_NORET <0xb, "ds_xor_b32", VGPR_32>;
+defm DS_MSKOR_B32 : DS_1A2D_NORET <0xc, "ds_mskor_b32", VGPR_32>;
+let mayLoad = 0 in {
+defm DS_WRITE_B32 : DS_1A1D_NORET <0xd, "ds_write_b32", VGPR_32>;
+defm DS_WRITE2_B32 : DS_1A1D_Off8_NORET <0xe, "ds_write2_b32", VGPR_32>;
+defm DS_WRITE2ST64_B32 : DS_1A1D_Off8_NORET <0xf, "ds_write2st64_b32", VGPR_32>;
+}
+defm DS_CMPST_B32 : DS_1A2D_NORET <0x10, "ds_cmpst_b32", VGPR_32>;
+defm DS_CMPST_F32 : DS_1A2D_NORET <0x11, "ds_cmpst_f32", VGPR_32>;
+defm DS_MIN_F32 : DS_1A2D_NORET <0x12, "ds_min_f32", VGPR_32>;
+defm DS_MAX_F32 : DS_1A2D_NORET <0x13, "ds_max_f32", VGPR_32>;
+
+defm DS_GWS_INIT : DS_1A_GDS <0x19, "ds_gws_init">;
+defm DS_GWS_SEMA_V : DS_1A_GDS <0x1a, "ds_gws_sema_v">;
+defm DS_GWS_SEMA_BR : DS_1A_GDS <0x1b, "ds_gws_sema_br">;
+defm DS_GWS_SEMA_P : DS_1A_GDS <0x1c, "ds_gws_sema_p">;
+defm DS_GWS_BARRIER : DS_1A_GDS <0x1d, "ds_gws_barrier">;
+let mayLoad = 0 in {
+defm DS_WRITE_B8 : DS_1A1D_NORET <0x1e, "ds_write_b8", VGPR_32>;
+defm DS_WRITE_B16 : DS_1A1D_NORET <0x1f, "ds_write_b16", VGPR_32>;
+}
+defm DS_ADD_RTN_U32 : DS_1A1D_RET <0x20, "ds_add_rtn_u32", VGPR_32, "ds_add_u32">;
+defm DS_SUB_RTN_U32 : DS_1A1D_RET <0x21, "ds_sub_rtn_u32", VGPR_32, "ds_sub_u32">;
+defm DS_RSUB_RTN_U32 : DS_1A1D_RET <0x22, "ds_rsub_rtn_u32", VGPR_32, "ds_rsub_u32">;
+defm DS_INC_RTN_U32 : DS_1A1D_RET <0x23, "ds_inc_rtn_u32", VGPR_32, "ds_inc_u32">;
+defm DS_DEC_RTN_U32 : DS_1A1D_RET <0x24, "ds_dec_rtn_u32", VGPR_32, "ds_dec_u32">;
+defm DS_MIN_RTN_I32 : DS_1A1D_RET <0x25, "ds_min_rtn_i32", VGPR_32, "ds_min_i32">;
+defm DS_MAX_RTN_I32 : DS_1A1D_RET <0x26, "ds_max_rtn_i32", VGPR_32, "ds_max_i32">;
+defm DS_MIN_RTN_U32 : DS_1A1D_RET <0x27, "ds_min_rtn_u32", VGPR_32, "ds_min_u32">;
+defm DS_MAX_RTN_U32 : DS_1A1D_RET <0x28, "ds_max_rtn_u32", VGPR_32, "ds_max_u32">;
+defm DS_AND_RTN_B32 : DS_1A1D_RET <0x29, "ds_and_rtn_b32", VGPR_32, "ds_and_b32">;
+defm DS_OR_RTN_B32 : DS_1A1D_RET <0x2a, "ds_or_rtn_b32", VGPR_32, "ds_or_b32">;
+defm DS_XOR_RTN_B32 : DS_1A1D_RET <0x2b, "ds_xor_rtn_b32", VGPR_32, "ds_xor_b32">;
+defm DS_MSKOR_RTN_B32 : DS_1A2D_RET <0x2c, "ds_mskor_rtn_b32", VGPR_32, "ds_mskor_b32">;
+defm DS_WRXCHG_RTN_B32 : DS_1A1D_RET <0x2d, "ds_wrxchg_rtn_b32", VGPR_32>;
+defm DS_WRXCHG2_RTN_B32 : DS_1A2D_RET <
+ 0x2e, "ds_wrxchg2_rtn_b32", VReg_64, "", VGPR_32
+>;
+defm DS_WRXCHG2ST64_RTN_B32 : DS_1A2D_RET <
+ 0x2f, "ds_wrxchg2st64_rtn_b32", VReg_64, "", VGPR_32
+>;
+defm DS_CMPST_RTN_B32 : DS_1A2D_RET <0x30, "ds_cmpst_rtn_b32", VGPR_32, "ds_cmpst_b32">;
+defm DS_CMPST_RTN_F32 : DS_1A2D_RET <0x31, "ds_cmpst_rtn_f32", VGPR_32, "ds_cmpst_f32">;
+defm DS_MIN_RTN_F32 : DS_1A2D_RET <0x32, "ds_min_rtn_f32", VGPR_32, "ds_min_f32">;
+defm DS_MAX_RTN_F32 : DS_1A2D_RET <0x33, "ds_max_rtn_f32", VGPR_32, "ds_max_f32">;
let SubtargetPredicate = isCI in {
-def DS_WRAP_RTN_F32 : DS_1A1D_RET <0x34, "ds_wrap_rtn_f32", VGPR_32, "ds_wrap_f32">;
+defm DS_WRAP_RTN_F32 : DS_1A1D_RET <0x34, "ds_wrap_rtn_f32", VGPR_32, "ds_wrap_f32">;
} // End isCI
+defm DS_SWIZZLE_B32 : DS_1A_RET <0x35, "ds_swizzle_b32", VGPR_32>;
+let mayStore = 0 in {
+defm DS_READ_B32 : DS_1A_RET <0x36, "ds_read_b32", VGPR_32>;
+defm DS_READ2_B32 : DS_1A_Off8_RET <0x37, "ds_read2_b32", VReg_64>;
+defm DS_READ2ST64_B32 : DS_1A_Off8_RET <0x38, "ds_read2st64_b32", VReg_64>;
+defm DS_READ_I8 : DS_1A_RET <0x39, "ds_read_i8", VGPR_32>;
+defm DS_READ_U8 : DS_1A_RET <0x3a, "ds_read_u8", VGPR_32>;
+defm DS_READ_I16 : DS_1A_RET <0x3b, "ds_read_i16", VGPR_32>;
+defm DS_READ_U16 : DS_1A_RET <0x3c, "ds_read_u16", VGPR_32>;
+}
+defm DS_CONSUME : DS_0A_RET <0x3d, "ds_consume">;
+defm DS_APPEND : DS_0A_RET <0x3e, "ds_append">;
+defm DS_ORDERED_COUNT : DS_1A_RET_GDS <0x3f, "ds_ordered_count">;
+defm DS_ADD_U64 : DS_1A1D_NORET <0x40, "ds_add_u64", VReg_64>;
+defm DS_SUB_U64 : DS_1A1D_NORET <0x41, "ds_sub_u64", VReg_64>;
+defm DS_RSUB_U64 : DS_1A1D_NORET <0x42, "ds_rsub_u64", VReg_64>;
+defm DS_INC_U64 : DS_1A1D_NORET <0x43, "ds_inc_u64", VReg_64>;
+defm DS_DEC_U64 : DS_1A1D_NORET <0x44, "ds_dec_u64", VReg_64>;
+defm DS_MIN_I64 : DS_1A1D_NORET <0x45, "ds_min_i64", VReg_64>;
+defm DS_MAX_I64 : DS_1A1D_NORET <0x46, "ds_max_i64", VReg_64>;
+defm DS_MIN_U64 : DS_1A1D_NORET <0x47, "ds_min_u64", VReg_64>;
+defm DS_MAX_U64 : DS_1A1D_NORET <0x48, "ds_max_u64", VReg_64>;
+defm DS_AND_B64 : DS_1A1D_NORET <0x49, "ds_and_b64", VReg_64>;
+defm DS_OR_B64 : DS_1A1D_NORET <0x4a, "ds_or_b64", VReg_64>;
+defm DS_XOR_B64 : DS_1A1D_NORET <0x4b, "ds_xor_b64", VReg_64>;
+defm DS_MSKOR_B64 : DS_1A2D_NORET <0x4c, "ds_mskor_b64", VReg_64>;
+let mayLoad = 0 in {
+defm DS_WRITE_B64 : DS_1A1D_NORET <0x4d, "ds_write_b64", VReg_64>;
+defm DS_WRITE2_B64 : DS_1A1D_Off8_NORET <0x4E, "ds_write2_b64", VReg_64>;
+defm DS_WRITE2ST64_B64 : DS_1A1D_Off8_NORET <0x4f, "ds_write2st64_b64", VReg_64>;
+}
+defm DS_CMPST_B64 : DS_1A2D_NORET <0x50, "ds_cmpst_b64", VReg_64>;
+defm DS_CMPST_F64 : DS_1A2D_NORET <0x51, "ds_cmpst_f64", VReg_64>;
+defm DS_MIN_F64 : DS_1A1D_NORET <0x52, "ds_min_f64", VReg_64>;
+defm DS_MAX_F64 : DS_1A1D_NORET <0x53, "ds_max_f64", VReg_64>;
+
+defm DS_ADD_RTN_U64 : DS_1A1D_RET <0x60, "ds_add_rtn_u64", VReg_64, "ds_add_u64">;
+defm DS_SUB_RTN_U64 : DS_1A1D_RET <0x61, "ds_sub_rtn_u64", VReg_64, "ds_sub_u64">;
+defm DS_RSUB_RTN_U64 : DS_1A1D_RET <0x62, "ds_rsub_rtn_u64", VReg_64, "ds_rsub_u64">;
+defm DS_INC_RTN_U64 : DS_1A1D_RET <0x63, "ds_inc_rtn_u64", VReg_64, "ds_inc_u64">;
+defm DS_DEC_RTN_U64 : DS_1A1D_RET <0x64, "ds_dec_rtn_u64", VReg_64, "ds_dec_u64">;
+defm DS_MIN_RTN_I64 : DS_1A1D_RET <0x65, "ds_min_rtn_i64", VReg_64, "ds_min_i64">;
+defm DS_MAX_RTN_I64 : DS_1A1D_RET <0x66, "ds_max_rtn_i64", VReg_64, "ds_max_i64">;
+defm DS_MIN_RTN_U64 : DS_1A1D_RET <0x67, "ds_min_rtn_u64", VReg_64, "ds_min_u64">;
+defm DS_MAX_RTN_U64 : DS_1A1D_RET <0x68, "ds_max_rtn_u64", VReg_64, "ds_max_u64">;
+defm DS_AND_RTN_B64 : DS_1A1D_RET <0x69, "ds_and_rtn_b64", VReg_64, "ds_and_b64">;
+defm DS_OR_RTN_B64 : DS_1A1D_RET <0x6a, "ds_or_rtn_b64", VReg_64, "ds_or_b64">;
+defm DS_XOR_RTN_B64 : DS_1A1D_RET <0x6b, "ds_xor_rtn_b64", VReg_64, "ds_xor_b64">;
+defm DS_MSKOR_RTN_B64 : DS_1A2D_RET <0x6c, "ds_mskor_rtn_b64", VReg_64, "ds_mskor_b64">;
+defm DS_WRXCHG_RTN_B64 : DS_1A1D_RET <0x6d, "ds_wrxchg_rtn_b64", VReg_64, "ds_wrxchg_b64">;
+defm DS_WRXCHG2_RTN_B64 : DS_1A2D_RET <0x6e, "ds_wrxchg2_rtn_b64", VReg_128, "ds_wrxchg2_b64", VReg_64>;
+defm DS_WRXCHG2ST64_RTN_B64 : DS_1A2D_RET <0x6f, "ds_wrxchg2st64_rtn_b64", VReg_128, "ds_wrxchg2st64_b64", VReg_64>;
+defm DS_CMPST_RTN_B64 : DS_1A2D_RET <0x70, "ds_cmpst_rtn_b64", VReg_64, "ds_cmpst_b64">;
+defm DS_CMPST_RTN_F64 : DS_1A2D_RET <0x71, "ds_cmpst_rtn_f64", VReg_64, "ds_cmpst_f64">;
+defm DS_MIN_RTN_F64 : DS_1A1D_RET <0x72, "ds_min_rtn_f64", VReg_64, "ds_min_f64">;
+defm DS_MAX_RTN_F64 : DS_1A1D_RET <0x73, "ds_max_rtn_f64", VReg_64, "ds_max_f64">;
+
+let mayStore = 0 in {
+defm DS_READ_B64 : DS_1A_RET <0x76, "ds_read_b64", VReg_64>;
+defm DS_READ2_B64 : DS_1A_Off8_RET <0x77, "ds_read2_b64", VReg_128>;
+defm DS_READ2ST64_B64 : DS_1A_Off8_RET <0x78, "ds_read2st64_b64", VReg_128>;
+}
-
-def DS_ADD_U64 : DS_1A1D_NORET <0x40, "ds_add_u64", VReg_64>;
-def DS_SUB_U64 : DS_1A1D_NORET <0x41, "ds_sub_u64", VReg_64>;
-def DS_RSUB_U64 : DS_1A1D_NORET <0x42, "ds_rsub_u64", VReg_64>;
-def DS_INC_U64 : DS_1A1D_NORET <0x43, "ds_inc_u64", VReg_64>;
-def DS_DEC_U64 : DS_1A1D_NORET <0x44, "ds_dec_u64", VReg_64>;
-def DS_MIN_I64 : DS_1A1D_NORET <0x45, "ds_min_i64", VReg_64>;
-def DS_MAX_I64 : DS_1A1D_NORET <0x46, "ds_max_i64", VReg_64>;
-def DS_MIN_U64 : DS_1A1D_NORET <0x47, "ds_min_u64", VReg_64>;
-def DS_MAX_U64 : DS_1A1D_NORET <0x48, "ds_max_u64", VReg_64>;
-def DS_AND_B64 : DS_1A1D_NORET <0x49, "ds_and_b64", VReg_64>;
-def DS_OR_B64 : DS_1A1D_NORET <0x4a, "ds_or_b64", VReg_64>;
-def DS_XOR_B64 : DS_1A1D_NORET <0x4b, "ds_xor_b64", VReg_64>;
-def DS_MSKOR_B64 : DS_1A1D_NORET <0x4c, "ds_mskor_b64", VReg_64>;
-def DS_CMPST_B64 : DS_1A2D_NORET <0x50, "ds_cmpst_b64", VReg_64>;
-def DS_CMPST_F64 : DS_1A2D_NORET <0x51, "ds_cmpst_f64", VReg_64>;
-def DS_MIN_F64 : DS_1A1D_NORET <0x52, "ds_min_f64", VReg_64>;
-def DS_MAX_F64 : DS_1A1D_NORET <0x53, "ds_max_f64", VReg_64>;
-
-def DS_ADD_RTN_U64 : DS_1A1D_RET <0x60, "ds_add_rtn_u64", VReg_64, "ds_add_u64">;
-def DS_SUB_RTN_U64 : DS_1A1D_RET <0x61, "ds_sub_rtn_u64", VReg_64, "ds_sub_u64">;
-def DS_RSUB_RTN_U64 : DS_1A1D_RET <0x62, "ds_rsub_rtn_u64", VReg_64, "ds_rsub_u64">;
-def DS_INC_RTN_U64 : DS_1A1D_RET <0x63, "ds_inc_rtn_u64", VReg_64, "ds_inc_u64">;
-def DS_DEC_RTN_U64 : DS_1A1D_RET <0x64, "ds_dec_rtn_u64", VReg_64, "ds_dec_u64">;
-def DS_MIN_RTN_I64 : DS_1A1D_RET <0x65, "ds_min_rtn_i64", VReg_64, "ds_min_i64">;
-def DS_MAX_RTN_I64 : DS_1A1D_RET <0x66, "ds_max_rtn_i64", VReg_64, "ds_max_i64">;
-def DS_MIN_RTN_U64 : DS_1A1D_RET <0x67, "ds_min_rtn_u64", VReg_64, "ds_min_u64">;
-def DS_MAX_RTN_U64 : DS_1A1D_RET <0x68, "ds_max_rtn_u64", VReg_64, "ds_max_u64">;
-def DS_AND_RTN_B64 : DS_1A1D_RET <0x69, "ds_and_rtn_b64", VReg_64, "ds_and_b64">;
-def DS_OR_RTN_B64 : DS_1A1D_RET <0x6a, "ds_or_rtn_b64", VReg_64, "ds_or_b64">;
-def DS_XOR_RTN_B64 : DS_1A1D_RET <0x6b, "ds_xor_rtn_b64", VReg_64, "ds_xor_b64">;
-def DS_MSKOR_RTN_B64 : DS_1A1D_RET <0x6c, "ds_mskor_rtn_b64", VReg_64, "ds_mskor_b64">;
-def DS_WRXCHG_RTN_B64 : DS_1A1D_RET <0x6d, "ds_wrxchg_rtn_b64", VReg_64, "ds_wrxchg_b64">;
-//def DS_WRXCHG2_RTN_B64 : DS_2A0D_RET <0x6e, "ds_wrxchg2_rtn_b64", VReg_64, "ds_wrxchg2_b64">;
-//def DS_WRXCHG2ST64_RTN_B64 : DS_2A0D_RET <0x6f, "ds_wrxchg2_rtn_b64", VReg_64, "ds_wrxchg2st64_b64">;
-def DS_CMPST_RTN_B64 : DS_1A2D_RET <0x70, "ds_cmpst_rtn_b64", VReg_64, "ds_cmpst_b64">;
-def DS_CMPST_RTN_F64 : DS_1A2D_RET <0x71, "ds_cmpst_rtn_f64", VReg_64, "ds_cmpst_f64">;
-def DS_MIN_RTN_F64 : DS_1A1D_RET <0x72, "ds_min_f64", VReg_64, "ds_min_f64">;
-def DS_MAX_RTN_F64 : DS_1A1D_RET <0x73, "ds_max_f64", VReg_64, "ds_max_f64">;
+defm DS_ADD_SRC2_U32 : DS_1A <0x80, "ds_add_src2_u32">;
+defm DS_SUB_SRC2_U32 : DS_1A <0x81, "ds_sub_src2_u32">;
+defm DS_RSUB_SRC2_U32 : DS_1A <0x82, "ds_rsub_src2_u32">;
+defm DS_INC_SRC2_U32 : DS_1A <0x83, "ds_inc_src2_u32">;
+defm DS_DEC_SRC2_U32 : DS_1A <0x84, "ds_dec_src2_u32">;
+defm DS_MIN_SRC2_I32 : DS_1A <0x85, "ds_min_src2_i32">;
+defm DS_MAX_SRC2_I32 : DS_1A <0x86, "ds_max_src2_i32">;
+defm DS_MIN_SRC2_U32 : DS_1A <0x87, "ds_min_src2_u32">;
+defm DS_MAX_SRC2_U32 : DS_1A <0x88, "ds_max_src2_u32">;
+defm DS_AND_SRC2_B32 : DS_1A <0x89, "ds_and_src_b32">;
+defm DS_OR_SRC2_B32 : DS_1A <0x8a, "ds_or_src2_b32">;
+defm DS_XOR_SRC2_B32 : DS_1A <0x8b, "ds_xor_src2_b32">;
+defm DS_WRITE_SRC2_B32 : DS_1A <0x8c, "ds_write_src2_b32">;
+
+defm DS_MIN_SRC2_F32 : DS_1A <0x92, "ds_min_src2_f32">;
+defm DS_MAX_SRC2_F32 : DS_1A <0x93, "ds_max_src2_f32">;
+
+defm DS_ADD_SRC2_U64 : DS_1A <0xc0, "ds_add_src2_u64">;
+defm DS_SUB_SRC2_U64 : DS_1A <0xc1, "ds_sub_src2_u64">;
+defm DS_RSUB_SRC2_U64 : DS_1A <0xc2, "ds_rsub_src2_u64">;
+defm DS_INC_SRC2_U64 : DS_1A <0xc3, "ds_inc_src2_u64">;
+defm DS_DEC_SRC2_U64 : DS_1A <0xc4, "ds_dec_src2_u64">;
+defm DS_MIN_SRC2_I64 : DS_1A <0xc5, "ds_min_src2_i64">;
+defm DS_MAX_SRC2_I64 : DS_1A <0xc6, "ds_max_src2_i64">;
+defm DS_MIN_SRC2_U64 : DS_1A <0xc7, "ds_min_src2_u64">;
+defm DS_MAX_SRC2_U64 : DS_1A <0xc8, "ds_max_src2_u64">;
+defm DS_AND_SRC2_B64 : DS_1A <0xc9, "ds_and_src2_b64">;
+defm DS_OR_SRC2_B64 : DS_1A <0xca, "ds_or_src2_b64">;
+defm DS_XOR_SRC2_B64 : DS_1A <0xcb, "ds_xor_src2_b64">;
+defm DS_WRITE_SRC2_B64 : DS_1A <0xcc, "ds_write_src2_b64">;
+
+defm DS_MIN_SRC2_F64 : DS_1A <0xd2, "ds_min_src2_f64">;
+defm DS_MAX_SRC2_F64 : DS_1A <0xd3, "ds_max_src2_f64">;
//let SubtargetPredicate = isCI in {
// DS_CONDXCHG32_RTN_B64
// DS_CONDXCHG32_RTN_B128
//} // End isCI
-// TODO: _SRC2_* forms
-
-defm DS_WRITE_B32 : DS_Store_Helper <0x0000000d, "ds_write_b32", VGPR_32>;
-defm DS_WRITE_B8 : DS_Store_Helper <0x00000001e, "ds_write_b8", VGPR_32>;
-defm DS_WRITE_B16 : DS_Store_Helper <0x00000001f, "ds_write_b16", VGPR_32>;
-defm DS_WRITE_B64 : DS_Store_Helper <0x00000004d, "ds_write_b64", VReg_64>;
-
-defm DS_READ_B32 : DS_Load_Helper <0x00000036, "ds_read_b32", VGPR_32>;
-defm DS_READ_I8 : DS_Load_Helper <0x00000039, "ds_read_i8", VGPR_32>;
-defm DS_READ_U8 : DS_Load_Helper <0x0000003a, "ds_read_u8", VGPR_32>;
-defm DS_READ_I16 : DS_Load_Helper <0x0000003b, "ds_read_i16", VGPR_32>;
-defm DS_READ_U16 : DS_Load_Helper <0x0000003c, "ds_read_u16", VGPR_32>;
-defm DS_READ_B64 : DS_Load_Helper <0x00000076, "ds_read_b64", VReg_64>;
-
-// 2 forms.
-defm DS_WRITE2_B32 : DS_Store2_Helper <0x0000000E, "ds_write2_b32", VGPR_32>;
-defm DS_WRITE2ST64_B32 : DS_Store2_Helper <0x0000000F, "ds_write2st64_b32", VGPR_32>;
-defm DS_WRITE2_B64 : DS_Store2_Helper <0x0000004E, "ds_write2_b64", VReg_64>;
-defm DS_WRITE2ST64_B64 : DS_Store2_Helper <0x0000004F, "ds_write2st64_b64", VReg_64>;
-
-defm DS_READ2_B32 : DS_Load2_Helper <0x00000037, "ds_read2_b32", VReg_64>;
-defm DS_READ2ST64_B32 : DS_Load2_Helper <0x00000038, "ds_read2st64_b32", VReg_64>;
-defm DS_READ2_B64 : DS_Load2_Helper <0x00000075, "ds_read2_b64", VReg_128>;
-defm DS_READ2ST64_B64 : DS_Load2_Helper <0x00000076, "ds_read2st64_b64", VReg_128>;
-
//===----------------------------------------------------------------------===//
// MUBUF Instructions
//===----------------------------------------------------------------------===//
-//def BUFFER_LOAD_FORMAT_X : MUBUF_ <mubuf<0x00>, "buffer_load_format_x", []>;
-//def BUFFER_LOAD_FORMAT_XY : MUBUF_ <mubuf<0x01>, "buffer_load_format_xy", []>;
-//def BUFFER_LOAD_FORMAT_XYZ : MUBUF_ <mubuf<0x02>, "buffer_load_format_xyz", []>;
-defm BUFFER_LOAD_FORMAT_XYZW : MUBUF_Load_Helper <mubuf<0x03>, "buffer_load_format_xyzw", VReg_128>;
-//def BUFFER_STORE_FORMAT_X : MUBUF_ <mubuf<0x04>, "buffer_store_format_x", []>;
-//def BUFFER_STORE_FORMAT_XY : MUBUF_ <mubuf<0x05>, "buffer_store_format_xy", []>;
-//def BUFFER_STORE_FORMAT_XYZ : MUBUF_ <mubuf<0x06>, "buffer_store_format_xyz", []>;
-//def BUFFER_STORE_FORMAT_XYZW : MUBUF_ <mubuf<0x07>, "buffer_store_format_xyzw", []>;
+defm BUFFER_LOAD_FORMAT_X : MUBUF_Load_Helper <
+ mubuf<0x00>, "buffer_load_format_x", VGPR_32
+>;
+defm BUFFER_LOAD_FORMAT_XY : MUBUF_Load_Helper <
+ mubuf<0x01>, "buffer_load_format_xy", VReg_64
+>;
+defm BUFFER_LOAD_FORMAT_XYZ : MUBUF_Load_Helper <
+ mubuf<0x02>, "buffer_load_format_xyz", VReg_96
+>;
+defm BUFFER_LOAD_FORMAT_XYZW : MUBUF_Load_Helper <
+ mubuf<0x03>, "buffer_load_format_xyzw", VReg_128
+>;
+defm BUFFER_STORE_FORMAT_X : MUBUF_Store_Helper <
+ mubuf<0x04>, "buffer_store_format_x", VGPR_32
+>;
+defm BUFFER_STORE_FORMAT_XY : MUBUF_Store_Helper <
+ mubuf<0x05>, "buffer_store_format_xy", VReg_64
+>;
+defm BUFFER_STORE_FORMAT_XYZ : MUBUF_Store_Helper <
+ mubuf<0x06>, "buffer_store_format_xyz", VReg_96
+>;
+defm BUFFER_STORE_FORMAT_XYZW : MUBUF_Store_Helper <
+ mubuf<0x07>, "buffer_store_format_xyzw", VReg_128
+>;
defm BUFFER_LOAD_UBYTE : MUBUF_Load_Helper <
mubuf<0x08, 0x10>, "buffer_load_ubyte", VGPR_32, i32, az_extloadi8_global
>;
mubuf<0x1e, 0x1f>, "buffer_store_dwordx4", VReg_128, v4i32, global_store
>;
-let SubtargetPredicate = isSICI in {
-
-//def BUFFER_ATOMIC_SWAP : MUBUF_ <0x00000030, "buffer_atomic_swap", []>;
defm BUFFER_ATOMIC_SWAP : MUBUF_Atomic <
- 0x00000030, "buffer_atomic_swap", VGPR_32, i32, atomic_swap_global
+ mubuf<0x30, 0x40>, "buffer_atomic_swap", VGPR_32, i32, atomic_swap_global
>;
-//def BUFFER_ATOMIC_CMPSWAP : MUBUF_ <0x00000031, "buffer_atomic_cmpswap", []>;
+//def BUFFER_ATOMIC_CMPSWAP : MUBUF_ <mubuf<0x31, 0x41>, "buffer_atomic_cmpswap", []>;
defm BUFFER_ATOMIC_ADD : MUBUF_Atomic <
- 0x00000032, "buffer_atomic_add", VGPR_32, i32, atomic_add_global
+ mubuf<0x32, 0x42>, "buffer_atomic_add", VGPR_32, i32, atomic_add_global
>;
defm BUFFER_ATOMIC_SUB : MUBUF_Atomic <
- 0x00000033, "buffer_atomic_sub", VGPR_32, i32, atomic_sub_global
+ mubuf<0x33, 0x43>, "buffer_atomic_sub", VGPR_32, i32, atomic_sub_global
>;
-//def BUFFER_ATOMIC_RSUB : MUBUF_ <0x00000034, "buffer_atomic_rsub", []>;
+//def BUFFER_ATOMIC_RSUB : MUBUF_ <mubuf<0x34>, "buffer_atomic_rsub", []>; // isn't on CI & VI
defm BUFFER_ATOMIC_SMIN : MUBUF_Atomic <
- 0x00000035, "buffer_atomic_smin", VGPR_32, i32, atomic_min_global
+ mubuf<0x35, 0x44>, "buffer_atomic_smin", VGPR_32, i32, atomic_min_global
>;
defm BUFFER_ATOMIC_UMIN : MUBUF_Atomic <
- 0x00000036, "buffer_atomic_umin", VGPR_32, i32, atomic_umin_global
+ mubuf<0x36, 0x45>, "buffer_atomic_umin", VGPR_32, i32, atomic_umin_global
>;
defm BUFFER_ATOMIC_SMAX : MUBUF_Atomic <
- 0x00000037, "buffer_atomic_smax", VGPR_32, i32, atomic_max_global
+ mubuf<0x37, 0x46>, "buffer_atomic_smax", VGPR_32, i32, atomic_max_global
>;
defm BUFFER_ATOMIC_UMAX : MUBUF_Atomic <
- 0x00000038, "buffer_atomic_umax", VGPR_32, i32, atomic_umax_global
+ mubuf<0x38, 0x47>, "buffer_atomic_umax", VGPR_32, i32, atomic_umax_global
>;
defm BUFFER_ATOMIC_AND : MUBUF_Atomic <
- 0x00000039, "buffer_atomic_and", VGPR_32, i32, atomic_and_global
+ mubuf<0x39, 0x48>, "buffer_atomic_and", VGPR_32, i32, atomic_and_global
>;
defm BUFFER_ATOMIC_OR : MUBUF_Atomic <
- 0x0000003a, "buffer_atomic_or", VGPR_32, i32, atomic_or_global
+ mubuf<0x3a, 0x49>, "buffer_atomic_or", VGPR_32, i32, atomic_or_global
>;
defm BUFFER_ATOMIC_XOR : MUBUF_Atomic <
- 0x0000003b, "buffer_atomic_xor", VGPR_32, i32, atomic_xor_global
->;
-//def BUFFER_ATOMIC_INC : MUBUF_ <0x0000003c, "buffer_atomic_inc", []>;
-//def BUFFER_ATOMIC_DEC : MUBUF_ <0x0000003d, "buffer_atomic_dec", []>;
-//def BUFFER_ATOMIC_FCMPSWAP : MUBUF_ <0x0000003e, "buffer_atomic_fcmpswap", []>;
-//def BUFFER_ATOMIC_FMIN : MUBUF_ <0x0000003f, "buffer_atomic_fmin", []>;
-//def BUFFER_ATOMIC_FMAX : MUBUF_ <0x00000040, "buffer_atomic_fmax", []>;
-//def BUFFER_ATOMIC_SWAP_X2 : MUBUF_X2 <0x00000050, "buffer_atomic_swap_x2", []>;
-//def BUFFER_ATOMIC_CMPSWAP_X2 : MUBUF_X2 <0x00000051, "buffer_atomic_cmpswap_x2", []>;
-//def BUFFER_ATOMIC_ADD_X2 : MUBUF_X2 <0x00000052, "buffer_atomic_add_x2", []>;
-//def BUFFER_ATOMIC_SUB_X2 : MUBUF_X2 <0x00000053, "buffer_atomic_sub_x2", []>;
-//def BUFFER_ATOMIC_RSUB_X2 : MUBUF_X2 <0x00000054, "buffer_atomic_rsub_x2", []>;
-//def BUFFER_ATOMIC_SMIN_X2 : MUBUF_X2 <0x00000055, "buffer_atomic_smin_x2", []>;
-//def BUFFER_ATOMIC_UMIN_X2 : MUBUF_X2 <0x00000056, "buffer_atomic_umin_x2", []>;
-//def BUFFER_ATOMIC_SMAX_X2 : MUBUF_X2 <0x00000057, "buffer_atomic_smax_x2", []>;
-//def BUFFER_ATOMIC_UMAX_X2 : MUBUF_X2 <0x00000058, "buffer_atomic_umax_x2", []>;
-//def BUFFER_ATOMIC_AND_X2 : MUBUF_X2 <0x00000059, "buffer_atomic_and_x2", []>;
-//def BUFFER_ATOMIC_OR_X2 : MUBUF_X2 <0x0000005a, "buffer_atomic_or_x2", []>;
-//def BUFFER_ATOMIC_XOR_X2 : MUBUF_X2 <0x0000005b, "buffer_atomic_xor_x2", []>;
-//def BUFFER_ATOMIC_INC_X2 : MUBUF_X2 <0x0000005c, "buffer_atomic_inc_x2", []>;
-//def BUFFER_ATOMIC_DEC_X2 : MUBUF_X2 <0x0000005d, "buffer_atomic_dec_x2", []>;
-//def BUFFER_ATOMIC_FCMPSWAP_X2 : MUBUF_X2 <0x0000005e, "buffer_atomic_fcmpswap_x2", []>;
-//def BUFFER_ATOMIC_FMIN_X2 : MUBUF_X2 <0x0000005f, "buffer_atomic_fmin_x2", []>;
-//def BUFFER_ATOMIC_FMAX_X2 : MUBUF_X2 <0x00000060, "buffer_atomic_fmax_x2", []>;
-//def BUFFER_WBINVL1_SC : MUBUF_WBINVL1 <0x00000070, "buffer_wbinvl1_sc", []>;
-//def BUFFER_WBINVL1 : MUBUF_WBINVL1 <0x00000071, "buffer_wbinvl1", []>;
-
-} // End SubtargetPredicate = isSICI
+ mubuf<0x3b, 0x4a>, "buffer_atomic_xor", VGPR_32, i32, atomic_xor_global
+>;
+//def BUFFER_ATOMIC_INC : MUBUF_ <mubuf<0x3c, 0x4b>, "buffer_atomic_inc", []>;
+//def BUFFER_ATOMIC_DEC : MUBUF_ <mubuf<0x3d, 0x4c>, "buffer_atomic_dec", []>;
+//def BUFFER_ATOMIC_FCMPSWAP : MUBUF_ <mubuf<0x3e>, "buffer_atomic_fcmpswap", []>; // isn't on VI
+//def BUFFER_ATOMIC_FMIN : MUBUF_ <mubuf<0x3f>, "buffer_atomic_fmin", []>; // isn't on VI
+//def BUFFER_ATOMIC_FMAX : MUBUF_ <mubuf<0x40>, "buffer_atomic_fmax", []>; // isn't on VI
+//def BUFFER_ATOMIC_SWAP_X2 : MUBUF_X2 <mubuf<0x50, 0x60>, "buffer_atomic_swap_x2", []>;
+//def BUFFER_ATOMIC_CMPSWAP_X2 : MUBUF_X2 <mubuf<0x51, 0x61>, "buffer_atomic_cmpswap_x2", []>;
+//def BUFFER_ATOMIC_ADD_X2 : MUBUF_X2 <mubuf<0x52, 0x62>, "buffer_atomic_add_x2", []>;
+//def BUFFER_ATOMIC_SUB_X2 : MUBUF_X2 <mubuf<0x53, 0x63>, "buffer_atomic_sub_x2", []>;
+//def BUFFER_ATOMIC_RSUB_X2 : MUBUF_X2 <mubuf<0x54>, "buffer_atomic_rsub_x2", []>; // isn't on CI & VI
+//def BUFFER_ATOMIC_SMIN_X2 : MUBUF_X2 <mubuf<0x55, 0x64>, "buffer_atomic_smin_x2", []>;
+//def BUFFER_ATOMIC_UMIN_X2 : MUBUF_X2 <mubuf<0x56, 0x65>, "buffer_atomic_umin_x2", []>;
+//def BUFFER_ATOMIC_SMAX_X2 : MUBUF_X2 <mubuf<0x57, 0x66>, "buffer_atomic_smax_x2", []>;
+//def BUFFER_ATOMIC_UMAX_X2 : MUBUF_X2 <mubuf<0x58, 0x67>, "buffer_atomic_umax_x2", []>;
+//def BUFFER_ATOMIC_AND_X2 : MUBUF_X2 <mubuf<0x59, 0x68>, "buffer_atomic_and_x2", []>;
+//def BUFFER_ATOMIC_OR_X2 : MUBUF_X2 <mubuf<0x5a, 0x69>, "buffer_atomic_or_x2", []>;
+//def BUFFER_ATOMIC_XOR_X2 : MUBUF_X2 <mubuf<0x5b, 0x6a>, "buffer_atomic_xor_x2", []>;
+//def BUFFER_ATOMIC_INC_X2 : MUBUF_X2 <mubuf<0x5c, 0x6b>, "buffer_atomic_inc_x2", []>;
+//def BUFFER_ATOMIC_DEC_X2 : MUBUF_X2 <mubuf<0x5d, 0x6c>, "buffer_atomic_dec_x2", []>;
+//def BUFFER_ATOMIC_FCMPSWAP_X2 : MUBUF_X2 <mubuf<0x5e>, "buffer_atomic_fcmpswap_x2", []>; // isn't on VI
+//def BUFFER_ATOMIC_FMIN_X2 : MUBUF_X2 <mubuf<0x5f>, "buffer_atomic_fmin_x2", []>; // isn't on VI
+//def BUFFER_ATOMIC_FMAX_X2 : MUBUF_X2 <mubuf<0x60>, "buffer_atomic_fmax_x2", []>; // isn't on VI
+//def BUFFER_WBINVL1_SC : MUBUF_WBINVL1 <mubuf<0x70>, "buffer_wbinvl1_sc", []>; // isn't on CI & VI
+//def BUFFER_WBINVL1_VOL : MUBUF_WBINVL1 <mubuf<0x70, 0x3f>, "buffer_wbinvl1_vol", []>; // isn't on SI
+//def BUFFER_WBINVL1 : MUBUF_WBINVL1 <mubuf<0x71, 0x3e>, "buffer_wbinvl1", []>;
//===----------------------------------------------------------------------===//
// MTBUF Instructions
//def IMAGE_ATOMIC_FCMPSWAP : MIMG_NoPattern_ <"image_atomic_fcmpswap", 0x0000001d>;
//def IMAGE_ATOMIC_FMIN : MIMG_NoPattern_ <"image_atomic_fmin", 0x0000001e>;
//def IMAGE_ATOMIC_FMAX : MIMG_NoPattern_ <"image_atomic_fmax", 0x0000001f>;
-defm IMAGE_SAMPLE : MIMG_Sampler <0x00000020, "image_sample">;
-defm IMAGE_SAMPLE_CL : MIMG_Sampler <0x00000021, "image_sample_cl">;
+defm IMAGE_SAMPLE : MIMG_Sampler_WQM <0x00000020, "image_sample">;
+defm IMAGE_SAMPLE_CL : MIMG_Sampler_WQM <0x00000021, "image_sample_cl">;
defm IMAGE_SAMPLE_D : MIMG_Sampler <0x00000022, "image_sample_d">;
defm IMAGE_SAMPLE_D_CL : MIMG_Sampler <0x00000023, "image_sample_d_cl">;
defm IMAGE_SAMPLE_L : MIMG_Sampler <0x00000024, "image_sample_l">;
-defm IMAGE_SAMPLE_B : MIMG_Sampler <0x00000025, "image_sample_b">;
-defm IMAGE_SAMPLE_B_CL : MIMG_Sampler <0x00000026, "image_sample_b_cl">;
+defm IMAGE_SAMPLE_B : MIMG_Sampler_WQM <0x00000025, "image_sample_b">;
+defm IMAGE_SAMPLE_B_CL : MIMG_Sampler_WQM <0x00000026, "image_sample_b_cl">;
defm IMAGE_SAMPLE_LZ : MIMG_Sampler <0x00000027, "image_sample_lz">;
-defm IMAGE_SAMPLE_C : MIMG_Sampler <0x00000028, "image_sample_c">;
-defm IMAGE_SAMPLE_C_CL : MIMG_Sampler <0x00000029, "image_sample_c_cl">;
+defm IMAGE_SAMPLE_C : MIMG_Sampler_WQM <0x00000028, "image_sample_c">;
+defm IMAGE_SAMPLE_C_CL : MIMG_Sampler_WQM <0x00000029, "image_sample_c_cl">;
defm IMAGE_SAMPLE_C_D : MIMG_Sampler <0x0000002a, "image_sample_c_d">;
defm IMAGE_SAMPLE_C_D_CL : MIMG_Sampler <0x0000002b, "image_sample_c_d_cl">;
defm IMAGE_SAMPLE_C_L : MIMG_Sampler <0x0000002c, "image_sample_c_l">;
-defm IMAGE_SAMPLE_C_B : MIMG_Sampler <0x0000002d, "image_sample_c_b">;
-defm IMAGE_SAMPLE_C_B_CL : MIMG_Sampler <0x0000002e, "image_sample_c_b_cl">;
+defm IMAGE_SAMPLE_C_B : MIMG_Sampler_WQM <0x0000002d, "image_sample_c_b">;
+defm IMAGE_SAMPLE_C_B_CL : MIMG_Sampler_WQM <0x0000002e, "image_sample_c_b_cl">;
defm IMAGE_SAMPLE_C_LZ : MIMG_Sampler <0x0000002f, "image_sample_c_lz">;
-defm IMAGE_SAMPLE_O : MIMG_Sampler <0x00000030, "image_sample_o">;
-defm IMAGE_SAMPLE_CL_O : MIMG_Sampler <0x00000031, "image_sample_cl_o">;
+defm IMAGE_SAMPLE_O : MIMG_Sampler_WQM <0x00000030, "image_sample_o">;
+defm IMAGE_SAMPLE_CL_O : MIMG_Sampler_WQM <0x00000031, "image_sample_cl_o">;
defm IMAGE_SAMPLE_D_O : MIMG_Sampler <0x00000032, "image_sample_d_o">;
defm IMAGE_SAMPLE_D_CL_O : MIMG_Sampler <0x00000033, "image_sample_d_cl_o">;
defm IMAGE_SAMPLE_L_O : MIMG_Sampler <0x00000034, "image_sample_l_o">;
-defm IMAGE_SAMPLE_B_O : MIMG_Sampler <0x00000035, "image_sample_b_o">;
-defm IMAGE_SAMPLE_B_CL_O : MIMG_Sampler <0x00000036, "image_sample_b_cl_o">;
+defm IMAGE_SAMPLE_B_O : MIMG_Sampler_WQM <0x00000035, "image_sample_b_o">;
+defm IMAGE_SAMPLE_B_CL_O : MIMG_Sampler_WQM <0x00000036, "image_sample_b_cl_o">;
defm IMAGE_SAMPLE_LZ_O : MIMG_Sampler <0x00000037, "image_sample_lz_o">;
-defm IMAGE_SAMPLE_C_O : MIMG_Sampler <0x00000038, "image_sample_c_o">;
-defm IMAGE_SAMPLE_C_CL_O : MIMG_Sampler <0x00000039, "image_sample_c_cl_o">;
+defm IMAGE_SAMPLE_C_O : MIMG_Sampler_WQM <0x00000038, "image_sample_c_o">;
+defm IMAGE_SAMPLE_C_CL_O : MIMG_Sampler_WQM <0x00000039, "image_sample_c_cl_o">;
defm IMAGE_SAMPLE_C_D_O : MIMG_Sampler <0x0000003a, "image_sample_c_d_o">;
defm IMAGE_SAMPLE_C_D_CL_O : MIMG_Sampler <0x0000003b, "image_sample_c_d_cl_o">;
defm IMAGE_SAMPLE_C_L_O : MIMG_Sampler <0x0000003c, "image_sample_c_l_o">;
-defm IMAGE_SAMPLE_C_B_O : MIMG_Sampler <0x0000003d, "image_sample_c_b_o">;
-defm IMAGE_SAMPLE_C_B_CL_O : MIMG_Sampler <0x0000003e, "image_sample_c_b_cl_o">;
+defm IMAGE_SAMPLE_C_B_O : MIMG_Sampler_WQM <0x0000003d, "image_sample_c_b_o">;
+defm IMAGE_SAMPLE_C_B_CL_O : MIMG_Sampler_WQM <0x0000003e, "image_sample_c_b_cl_o">;
defm IMAGE_SAMPLE_C_LZ_O : MIMG_Sampler <0x0000003f, "image_sample_c_lz_o">;
-defm IMAGE_GATHER4 : MIMG_Gather <0x00000040, "image_gather4">;
-defm IMAGE_GATHER4_CL : MIMG_Gather <0x00000041, "image_gather4_cl">;
+defm IMAGE_GATHER4 : MIMG_Gather_WQM <0x00000040, "image_gather4">;
+defm IMAGE_GATHER4_CL : MIMG_Gather_WQM <0x00000041, "image_gather4_cl">;
defm IMAGE_GATHER4_L : MIMG_Gather <0x00000044, "image_gather4_l">;
-defm IMAGE_GATHER4_B : MIMG_Gather <0x00000045, "image_gather4_b">;
-defm IMAGE_GATHER4_B_CL : MIMG_Gather <0x00000046, "image_gather4_b_cl">;
+defm IMAGE_GATHER4_B : MIMG_Gather_WQM <0x00000045, "image_gather4_b">;
+defm IMAGE_GATHER4_B_CL : MIMG_Gather_WQM <0x00000046, "image_gather4_b_cl">;
defm IMAGE_GATHER4_LZ : MIMG_Gather <0x00000047, "image_gather4_lz">;
-defm IMAGE_GATHER4_C : MIMG_Gather <0x00000048, "image_gather4_c">;
-defm IMAGE_GATHER4_C_CL : MIMG_Gather <0x00000049, "image_gather4_c_cl">;
+defm IMAGE_GATHER4_C : MIMG_Gather_WQM <0x00000048, "image_gather4_c">;
+defm IMAGE_GATHER4_C_CL : MIMG_Gather_WQM <0x00000049, "image_gather4_c_cl">;
defm IMAGE_GATHER4_C_L : MIMG_Gather <0x0000004c, "image_gather4_c_l">;
-defm IMAGE_GATHER4_C_B : MIMG_Gather <0x0000004d, "image_gather4_c_b">;
-defm IMAGE_GATHER4_C_B_CL : MIMG_Gather <0x0000004e, "image_gather4_c_b_cl">;
+defm IMAGE_GATHER4_C_B : MIMG_Gather_WQM <0x0000004d, "image_gather4_c_b">;
+defm IMAGE_GATHER4_C_B_CL : MIMG_Gather_WQM <0x0000004e, "image_gather4_c_b_cl">;
defm IMAGE_GATHER4_C_LZ : MIMG_Gather <0x0000004f, "image_gather4_c_lz">;
-defm IMAGE_GATHER4_O : MIMG_Gather <0x00000050, "image_gather4_o">;
-defm IMAGE_GATHER4_CL_O : MIMG_Gather <0x00000051, "image_gather4_cl_o">;
+defm IMAGE_GATHER4_O : MIMG_Gather_WQM <0x00000050, "image_gather4_o">;
+defm IMAGE_GATHER4_CL_O : MIMG_Gather_WQM <0x00000051, "image_gather4_cl_o">;
defm IMAGE_GATHER4_L_O : MIMG_Gather <0x00000054, "image_gather4_l_o">;
-defm IMAGE_GATHER4_B_O : MIMG_Gather <0x00000055, "image_gather4_b_o">;
+defm IMAGE_GATHER4_B_O : MIMG_Gather_WQM <0x00000055, "image_gather4_b_o">;
defm IMAGE_GATHER4_B_CL_O : MIMG_Gather <0x00000056, "image_gather4_b_cl_o">;
defm IMAGE_GATHER4_LZ_O : MIMG_Gather <0x00000057, "image_gather4_lz_o">;
-defm IMAGE_GATHER4_C_O : MIMG_Gather <0x00000058, "image_gather4_c_o">;
-defm IMAGE_GATHER4_C_CL_O : MIMG_Gather <0x00000059, "image_gather4_c_cl_o">;
+defm IMAGE_GATHER4_C_O : MIMG_Gather_WQM <0x00000058, "image_gather4_c_o">;
+defm IMAGE_GATHER4_C_CL_O : MIMG_Gather_WQM <0x00000059, "image_gather4_c_cl_o">;
defm IMAGE_GATHER4_C_L_O : MIMG_Gather <0x0000005c, "image_gather4_c_l_o">;
-defm IMAGE_GATHER4_C_B_O : MIMG_Gather <0x0000005d, "image_gather4_c_b_o">;
-defm IMAGE_GATHER4_C_B_CL_O : MIMG_Gather <0x0000005e, "image_gather4_c_b_cl_o">;
+defm IMAGE_GATHER4_C_B_O : MIMG_Gather_WQM <0x0000005d, "image_gather4_c_b_o">;
+defm IMAGE_GATHER4_C_B_CL_O : MIMG_Gather_WQM <0x0000005e, "image_gather4_c_b_cl_o">;
defm IMAGE_GATHER4_C_LZ_O : MIMG_Gather <0x0000005f, "image_gather4_c_lz_o">;
-defm IMAGE_GET_LOD : MIMG_Sampler <0x00000060, "image_get_lod">;
+defm IMAGE_GET_LOD : MIMG_Sampler_WQM <0x00000060, "image_get_lod">;
defm IMAGE_SAMPLE_CD : MIMG_Sampler <0x00000068, "image_sample_cd">;
defm IMAGE_SAMPLE_CD_CL : MIMG_Sampler <0x00000069, "image_sample_cd_cl">;
defm IMAGE_SAMPLE_C_CD : MIMG_Sampler <0x0000006a, "image_sample_c_cd">;
// VOP1 Instructions
//===----------------------------------------------------------------------===//
-//def V_NOP : VOP1_ <0x00000000, "v_nop", []>;
+let vdst = 0, src0 = 0 in {
+defm V_NOP : VOP1_m <vop1<0x0>, (outs), (ins), "v_nop", [], "v_nop">;
+}
let isMoveImm = 1 in {
defm V_MOV_B32 : VOP1Inst <vop1<0x1>, "v_mov_b32", VOP_I32_I32>;
VOP_I32_F32, cvt_rpi_i32_f32>;
defm V_CVT_FLR_I32_F32 : VOP1Inst <vop1<0xd>, "v_cvt_flr_i32_f32",
VOP_I32_F32, cvt_flr_i32_f32>;
-//defm V_CVT_OFF_F32_I4 : VOP1_32 <0x0000000e, "v_cvt_off_f32_i4", []>;
+defm V_CVT_OFF_F32_I4 : VOP1Inst <vop1<0x0e>, "v_cvt_off_f32_i4", VOP_F32_I32>;
defm V_CVT_F32_F64 : VOP1Inst <vop1<0xf>, "v_cvt_f32_f64",
VOP_F32_F64, fround
>;
defm V_FFBH_U32 : VOP1Inst <vop1<0x39, 0x2d>, "v_ffbh_u32", VOP_I32_I32>;
defm V_FFBL_B32 : VOP1Inst <vop1<0x3a, 0x2e>, "v_ffbl_b32", VOP_I32_I32>;
defm V_FFBH_I32 : VOP1Inst <vop1<0x3b, 0x2f>, "v_ffbh_i32", VOP_I32_I32>;
-//defm V_FREXP_EXP_I32_F64 : VOPInst <0x0000003c, "v_frexp_exp_i32_f64", VOP_I32_F32>;
+defm V_FREXP_EXP_I32_F64 : VOP1Inst <vop1<0x3c,0x30>, "v_frexp_exp_i32_f64",
+ VOP_I32_F64
+>;
defm V_FREXP_MANT_F64 : VOP1Inst <vop1<0x3d, 0x31>, "v_frexp_mant_f64",
VOP_F64_F64
>;
defm V_FRACT_F64 : VOP1Inst <vop1<0x3e, 0x32>, "v_fract_f64", VOP_F64_F64>;
-//defm V_FREXP_EXP_I32_F32 : VOPInst <0x0000003f, "v_frexp_exp_i32_f32", VOP_I32_F32>;
+defm V_FREXP_EXP_I32_F32 : VOP1Inst <vop1<0x3f, 0x33>, "v_frexp_exp_i32_f32",
+ VOP_I32_F32
+>;
defm V_FREXP_MANT_F32 : VOP1Inst <vop1<0x40, 0x34>, "v_frexp_mant_f32",
VOP_F32_F32
>;
-//def V_CLREXCP : VOP1_ <0x00000041, "v_clrexcp", []>;
+let vdst = 0, src0 = 0 in {
+defm V_CLREXCP : VOP1_m <vop1<0x41,0x35>, (outs), (ins), "v_clrexcp", [],
+ "v_clrexcp"
+>;
+}
defm V_MOVRELD_B32 : VOP1Inst <vop1<0x42, 0x36>, "v_movreld_b32", VOP_I32_I32>;
defm V_MOVRELS_B32 : VOP1Inst <vop1<0x43, 0x37>, "v_movrels_b32", VOP_I32_I32>;
defm V_MOVRELSD_B32 : VOP1Inst <vop1<0x44, 0x38>, "v_movrelsd_b32", VOP_I32_I32>;
// VOP2 Instructions
//===----------------------------------------------------------------------===//
-defm V_CNDMASK_B32_e64 : VOP3_m_nosrcmod <vop3<0x100>, (outs VGPR_32:$dst),
+defm V_CNDMASK_B32_e64 : VOP3_m_nomods <vop3<0x100>, (outs VGPR_32:$dst),
(ins VSrc_32:$src0, VSrc_32:$src1, SSrc_64:$src2),
"v_cndmask_b32_e64 $dst, $src0, $src1, $src2",
[(set i32:$dst, (select i1:$src2, i32:$src1, i32:$src0))],
defm V_MUL_I32_I24 : VOP2Inst <vop2<0x9, 0x6>, "v_mul_i32_i24",
VOP_I32_I32_I32, AMDGPUmul_i24
>;
-//defm V_MUL_HI_I32_I24 : VOP2_32 <0x0000000a, "v_mul_hi_i32_i24", []>;
+
+defm V_MUL_HI_I32_I24 : VOP2Inst <vop2<0xa,0x7>, "v_mul_hi_i32_i24",
+ VOP_I32_I32_I32
+>;
+
defm V_MUL_U32_U24 : VOP2Inst <vop2<0xb, 0x8>, "v_mul_u32_u24",
VOP_I32_I32_I32, AMDGPUmul_u24
>;
-//defm V_MUL_HI_U32_U24 : VOP2_32 <0x0000000c, "v_mul_hi_u32_u24", []>;
+
+defm V_MUL_HI_U32_U24 : VOP2Inst <vop2<0xc,0x9>, "v_mul_hi_u32_u24",
+ VOP_I32_I32_I32
+>;
defm V_MIN_F32 : VOP2Inst <vop2<0xf, 0xa>, "v_min_f32", VOP_F32_F32_F32,
fminnum>;
defm V_MAX_F32 : VOP2Inst <vop2<0x10, 0xb>, "v_max_f32", VOP_F32_F32_F32,
fmaxnum>;
-defm V_MIN_I32 : VOP2Inst <vop2<0x11, 0xc>, "v_min_i32", VOP_I32_I32_I32,
- AMDGPUsmin
->;
-defm V_MAX_I32 : VOP2Inst <vop2<0x12, 0xd>, "v_max_i32", VOP_I32_I32_I32,
- AMDGPUsmax
->;
-defm V_MIN_U32 : VOP2Inst <vop2<0x13, 0xe>, "v_min_u32", VOP_I32_I32_I32,
- AMDGPUumin
->;
-defm V_MAX_U32 : VOP2Inst <vop2<0x14, 0xf>, "v_max_u32", VOP_I32_I32_I32,
- AMDGPUumax
->;
+defm V_MIN_I32 : VOP2Inst <vop2<0x11, 0xc>, "v_min_i32", VOP_I32_I32_I32>;
+defm V_MAX_I32 : VOP2Inst <vop2<0x12, 0xd>, "v_max_i32", VOP_I32_I32_I32>;
+defm V_MIN_U32 : VOP2Inst <vop2<0x13, 0xe>, "v_min_u32", VOP_I32_I32_I32>;
+defm V_MAX_U32 : VOP2Inst <vop2<0x14, 0xf>, "v_max_u32", VOP_I32_I32_I32>;
-// No non-Rev Op on VI
defm V_LSHRREV_B32 : VOP2Inst <
vop2<0x16, 0x10>, "v_lshrrev_b32", VOP_I32_I32_I32, null_frag,
- "v_lshr_b32", "v_lshrrev_b32"
+ "v_lshr_b32"
>;
-// No non-Rev OP on VI
defm V_ASHRREV_I32 : VOP2Inst <
vop2<0x18, 0x11>, "v_ashrrev_i32", VOP_I32_I32_I32, null_frag,
- "v_ashr_i32", "v_ashrrev_i32"
+ "v_ashr_i32"
>;
-// No non-Rev OP on VI
defm V_LSHLREV_B32 : VOP2Inst <
vop2<0x1a, 0x12>, "v_lshlrev_b32", VOP_I32_I32_I32, null_frag,
- "v_lshl_b32", "v_lshlrev_b32"
+ "v_lshl_b32"
>;
-defm V_AND_B32 : VOP2Inst <vop2<0x1b, 0x13>, "v_and_b32",
- VOP_I32_I32_I32, and>;
-defm V_OR_B32 : VOP2Inst <vop2<0x1c, 0x14>, "v_or_b32",
- VOP_I32_I32_I32, or
->;
-defm V_XOR_B32 : VOP2Inst <vop2<0x1d, 0x15>, "v_xor_b32",
- VOP_I32_I32_I32, xor
->;
+defm V_AND_B32 : VOP2Inst <vop2<0x1b, 0x13>, "v_and_b32", VOP_I32_I32_I32>;
+defm V_OR_B32 : VOP2Inst <vop2<0x1c, 0x14>, "v_or_b32", VOP_I32_I32_I32>;
+defm V_XOR_B32 : VOP2Inst <vop2<0x1d, 0x15>, "v_xor_b32", VOP_I32_I32_I32>;
defm V_MAC_F32 : VOP2Inst <vop2<0x1f, 0x16>, "v_mac_f32", VOP_F32_F32_F32>;
} // End isCommutable = 1
-defm V_MADMK_F32 : VOP2Inst <vop2<0x20, 0x17>, "v_madmk_f32", VOP_F32_F32_F32>;
+defm V_MADMK_F32 : VOP2MADK <vop2<0x20, 0x17>, "v_madmk_f32">;
let isCommutable = 1 in {
-defm V_MADAK_F32 : VOP2Inst <vop2<0x21, 0x18>, "v_madak_f32", VOP_F32_F32_F32>;
+defm V_MADAK_F32 : VOP2MADK <vop2<0x21, 0x18>, "v_madak_f32">;
} // End isCommutable = 1
let isCommutable = 1, Defs = [VCC] in { // Carry-out goes to VCC
defm V_ADD_I32 : VOP2bInst <vop2<0x25, 0x19>, "v_add_i32",
VOP_I32_I32_I32, add
>;
-defm V_SUB_I32 : VOP2bInst <vop2<0x26, 0x1a>, "v_sub_i32",
- VOP_I32_I32_I32, sub
->;
+defm V_SUB_I32 : VOP2bInst <vop2<0x26, 0x1a>, "v_sub_i32", VOP_I32_I32_I32>;
defm V_SUBREV_I32 : VOP2bInst <vop2<0x27, 0x1b>, "v_subrev_i32",
VOP_I32_I32_I32, null_frag, "v_sub_i32"
let Uses = [VCC] in { // Carry-in comes from VCC
defm V_ADDC_U32 : VOP2bInst <vop2<0x28, 0x1c>, "v_addc_u32",
- VOP_I32_I32_I32_VCC, adde
+ VOP_I32_I32_I32_VCC
>;
defm V_SUBB_U32 : VOP2bInst <vop2<0x29, 0x1d>, "v_subb_u32",
- VOP_I32_I32_I32_VCC, sube
+ VOP_I32_I32_I32_VCC
>;
defm V_SUBBREV_U32 : VOP2bInst <vop2<0x2a, 0x1e>, "v_subbrev_u32",
VOP_I32_I32_I32_VCC, null_frag, "v_subb_u32"
vop3 <0x001, 0x289>,
"v_readlane_b32",
(outs SReg_32:$vdst),
- (ins VGPR_32:$src0, SSrc_32:$vsrc1),
- "v_readlane_b32 $vdst, $src0, $vsrc1"
+ (ins VGPR_32:$src0, SCSrc_32:$src1),
+ "v_readlane_b32 $vdst, $src0, $src1"
>;
defm V_WRITELANE_B32 : VOP2SI_3VI_m <
vop3 <0x002, 0x28a>,
"v_writelane_b32",
(outs VGPR_32:$vdst),
- (ins SReg_32:$src0, SSrc_32:$vsrc1),
- "v_writelane_b32 $vdst, $src0, $vsrc1"
+ (ins SReg_32:$src0, SCSrc_32:$src1),
+ "v_writelane_b32 $vdst, $src0, $src1"
>;
// These instructions only exist on SI and CI
let SubtargetPredicate = isSICI in {
-let isCommutable = 1 in {
-defm V_MAC_LEGACY_F32 : VOP2Inst <vop2<0x6>, "v_mac_legacy_f32",
- VOP_F32_F32_F32
->;
-} // End isCommutable = 1
-
-defm V_MIN_LEGACY_F32 : VOP2Inst <vop2<0xd>, "v_min_legacy_f32",
+defm V_MIN_LEGACY_F32 : VOP2InstSI <vop2<0xd>, "v_min_legacy_f32",
VOP_F32_F32_F32, AMDGPUfmin_legacy
>;
-defm V_MAX_LEGACY_F32 : VOP2Inst <vop2<0xe>, "v_max_legacy_f32",
+defm V_MAX_LEGACY_F32 : VOP2InstSI <vop2<0xe>, "v_max_legacy_f32",
VOP_F32_F32_F32, AMDGPUfmax_legacy
>;
let isCommutable = 1 in {
-defm V_LSHR_B32 : VOP2Inst <vop2<0x15>, "v_lshr_b32", VOP_I32_I32_I32, srl>;
-defm V_ASHR_I32 : VOP2Inst <vop2<0x17>, "v_ashr_i32",
- VOP_I32_I32_I32, sra
->;
-
-let hasPostISelHook = 1 in {
-defm V_LSHL_B32 : VOP2Inst <vop2<0x19>, "v_lshl_b32", VOP_I32_I32_I32, shl>;
-}
-
+defm V_LSHR_B32 : VOP2InstSI <vop2<0x15>, "v_lshr_b32", VOP_I32_I32_I32>;
+defm V_ASHR_I32 : VOP2InstSI <vop2<0x17>, "v_ashr_i32", VOP_I32_I32_I32>;
+defm V_LSHL_B32 : VOP2InstSI <vop2<0x19>, "v_lshl_b32", VOP_I32_I32_I32>;
} // End isCommutable = 1
} // End let SubtargetPredicate = SICI
+let isCommutable = 1 in {
+defm V_MAC_LEGACY_F32 : VOP2_VI3_Inst <vop23<0x6, 0x28e>, "v_mac_legacy_f32",
+ VOP_F32_F32_F32
+>;
+} // End isCommutable = 1
+
defm V_BFM_B32 : VOP2_VI3_Inst <vop23<0x1e, 0x293>, "v_bfm_b32", VOP_I32_I32_I32,
AMDGPUbfm
>;
VOP_F32_F32_I32, AMDGPUldexp
>;
-////def V_CVT_PKACCUM_U8_F32 : VOP2_U8 <0x0000002c, "v_cvt_pkaccum_u8_f32", []>;
-////def V_CVT_PKNORM_I16_F32 : VOP2_I16 <0x0000002d, "v_cvt_pknorm_i16_f32", []>;
-////def V_CVT_PKNORM_U16_F32 : VOP2_U16 <0x0000002e, "v_cvt_pknorm_u16_f32", []>;
+
+defm V_CVT_PKACCUM_U8_F32 : VOP2_VI3_Inst <vop23<0x2c, 0x1f0>, "v_cvt_pkaccum_u8_f32",
+ VOP_I32_F32_I32>; // TODO: set "Uses = dst"
+
+defm V_CVT_PKNORM_I16_F32 : VOP2_VI3_Inst <vop23<0x2d, 0x294>, "v_cvt_pknorm_i16_f32",
+ VOP_I32_F32_F32
+>;
+defm V_CVT_PKNORM_U16_F32 : VOP2_VI3_Inst <vop23<0x2e, 0x295>, "v_cvt_pknorm_u16_f32",
+ VOP_I32_F32_F32
+>;
defm V_CVT_PKRTZ_F16_F32 : VOP2_VI3_Inst <vop23<0x2f, 0x296>, "v_cvt_pkrtz_f16_f32",
- VOP_I32_F32_F32, int_SI_packf16
+ VOP_I32_F32_F32, int_SI_packf16
+>;
+defm V_CVT_PK_U16_U32 : VOP2_VI3_Inst <vop23<0x30, 0x297>, "v_cvt_pk_u16_u32",
+ VOP_I32_I32_I32
+>;
+defm V_CVT_PK_I16_I32 : VOP2_VI3_Inst <vop23<0x31, 0x298>, "v_cvt_pk_i16_i32",
+ VOP_I32_I32_I32
>;
-////def V_CVT_PK_U16_U32 : VOP2_U16 <0x00000030, "v_cvt_pk_u16_u32", []>;
-////def V_CVT_PK_I16_I32 : VOP2_I16 <0x00000031, "v_cvt_pk_i16_i32", []>;
//===----------------------------------------------------------------------===//
// VOP3 Instructions
VOP_I32_I32_I32_I32
>;
-defm V_MIN3_F32 : VOP3Inst <vop3<0x151>, "v_min3_f32",
+defm V_MIN3_F32 : VOP3Inst <vop3<0x151, 0x1d0>, "v_min3_f32",
VOP_F32_F32_F32_F32, AMDGPUfmin3>;
-defm V_MIN3_I32 : VOP3Inst <vop3<0x152>, "v_min3_i32",
+defm V_MIN3_I32 : VOP3Inst <vop3<0x152, 0x1d1>, "v_min3_i32",
VOP_I32_I32_I32_I32, AMDGPUsmin3
>;
-defm V_MIN3_U32 : VOP3Inst <vop3<0x153>, "v_min3_u32",
+defm V_MIN3_U32 : VOP3Inst <vop3<0x153, 0x1d2>, "v_min3_u32",
VOP_I32_I32_I32_I32, AMDGPUumin3
>;
-defm V_MAX3_F32 : VOP3Inst <vop3<0x154>, "v_max3_f32",
+defm V_MAX3_F32 : VOP3Inst <vop3<0x154, 0x1d3>, "v_max3_f32",
VOP_F32_F32_F32_F32, AMDGPUfmax3
>;
-defm V_MAX3_I32 : VOP3Inst <vop3<0x155>, "v_max3_i32",
+defm V_MAX3_I32 : VOP3Inst <vop3<0x155, 0x1d4>, "v_max3_i32",
VOP_I32_I32_I32_I32, AMDGPUsmax3
>;
-defm V_MAX3_U32 : VOP3Inst <vop3<0x156>, "v_max3_u32",
+defm V_MAX3_U32 : VOP3Inst <vop3<0x156, 0x1d5>, "v_max3_u32",
VOP_I32_I32_I32_I32, AMDGPUumax3
>;
-//def V_MED3_F32 : VOP3_MED3 <0x00000157, "v_med3_f32", []>;
-//def V_MED3_I32 : VOP3_MED3 <0x00000158, "v_med3_i32", []>;
-//def V_MED3_U32 : VOP3_MED3 <0x00000159, "v_med3_u32", []>;
+defm V_MED3_F32 : VOP3Inst <vop3<0x157, 0x1d6>, "v_med3_f32",
+ VOP_F32_F32_F32_F32
+>;
+defm V_MED3_I32 : VOP3Inst <vop3<0x158, 0x1d7>, "v_med3_i32",
+ VOP_I32_I32_I32_I32
+>;
+defm V_MED3_U32 : VOP3Inst <vop3<0x159, 0x1d8>, "v_med3_u32",
+ VOP_I32_I32_I32_I32
+>;
+
//def V_SAD_U8 : VOP3_U8 <0x0000015a, "v_sad_u8", []>;
//def V_SAD_HI_U8 : VOP3_U8 <0x0000015b, "v_sad_hi_u8", []>;
//def V_SAD_U16 : VOP3_U16 <0x0000015c, "v_sad_u16", []>;
} // isCommutable = 1, SchedRW = [WriteQuarterRate32]
+let SchedRW = [WriteFloatFMA, WriteSALU] in {
defm V_DIV_SCALE_F32 : VOP3b_32 <vop3<0x16d, 0x1e0>, "v_div_scale_f32", []>;
+}
-let SchedRW = [WriteDouble] in {
+let SchedRW = [WriteDouble, WriteSALU] in {
// Double precision division pre-scale.
defm V_DIV_SCALE_F64 : VOP3b_64 <vop3<0x16e, 0x1e1>, "v_div_scale_f64", []>;
} // let SchedRW = [WriteDouble]
-let isCommutable = 1 in {
-defm V_DIV_FMAS_F32 : VOP3Inst <vop3<0x16f, 0x1e2>, "v_div_fmas_f32",
+let isCommutable = 1, Uses = [VCC] in {
+
+// v_div_fmas_f32:
+// result = src0 * src1 + src2
+// if (vcc)
+// result *= 2^32
+//
+defm V_DIV_FMAS_F32 : VOP3_VCC_Inst <vop3<0x16f, 0x1e2>, "v_div_fmas_f32",
VOP_F32_F32_F32_F32, AMDGPUdiv_fmas
>;
+
let SchedRW = [WriteDouble] in {
-defm V_DIV_FMAS_F64 : VOP3Inst <vop3<0x170, 0x1e3>, "v_div_fmas_f64",
+// v_div_fmas_f64:
+// result = src0 * src1 + src2
+// if (vcc)
+// result *= 2^64
+//
+defm V_DIV_FMAS_F64 : VOP3_VCC_Inst <vop3<0x170, 0x1e3>, "v_div_fmas_f64",
VOP_F64_F64_F64_F64, AMDGPUdiv_fmas
>;
+
} // End SchedRW = [WriteDouble]
} // End isCommutable = 1
// These instructions only exist on SI and CI
let SubtargetPredicate = isSICI in {
-defm V_LSHL_B64 : VOP3Inst <vop3<0x161>, "v_lshl_b64",
- VOP_I64_I64_I32, shl
->;
-
-defm V_LSHR_B64 : VOP3Inst <vop3<0x162>, "v_lshr_b64",
- VOP_I64_I64_I32, srl
->;
-
-defm V_ASHR_I64 : VOP3Inst <vop3<0x163>, "v_ashr_i64",
- VOP_I64_I64_I32, sra
->;
+defm V_LSHL_B64 : VOP3Inst <vop3<0x161>, "v_lshl_b64", VOP_I64_I64_I32>;
+defm V_LSHR_B64 : VOP3Inst <vop3<0x162>, "v_lshr_b64", VOP_I64_I64_I32>;
+defm V_ASHR_I64 : VOP3Inst <vop3<0x163>, "v_ashr_i64", VOP_I64_I64_I32>;
defm V_MULLIT_F32 : VOP3Inst <vop3<0x150>, "v_mullit_f32",
VOP_F32_F32_F32_F32>;
} // End SubtargetPredicate = isSICI
+let SubtargetPredicate = isVI in {
+
+defm V_LSHLREV_B64 : VOP3Inst <vop3<0, 0x28f>, "v_lshlrev_b64",
+ VOP_I64_I32_I64
+>;
+defm V_LSHRREV_B64 : VOP3Inst <vop3<0, 0x290>, "v_lshrrev_b64",
+ VOP_I64_I32_I64
+>;
+defm V_ASHRREV_I64 : VOP3Inst <vop3<0, 0x291>, "v_ashrrev_i64",
+ VOP_I64_I32_I64
+>;
+
+} // End SubtargetPredicate = isVI
+
//===----------------------------------------------------------------------===//
// Pseudo Instructions
//===----------------------------------------------------------------------===//
} // Uses = [EXEC,VCC,M0], Defs = [EXEC,VCC,M0]
-let usesCustomInserter = 1 in {
-
-def V_SUB_F64 : InstSI <
- (outs VReg_64:$dst),
- (ins VReg_64:$src0, VReg_64:$src1),
- "v_sub_f64 $dst, $src0, $src1",
- [(set f64:$dst, (fsub f64:$src0, f64:$src1))]
->;
-
-} // end usesCustomInserter
-
multiclass SI_SPILL_SGPR <RegisterClass sgpr_class> {
let UseNamedOperandTable = 1 in {
/* int_SI_vs_load_input */
def : Pat<
(SIload_input v4i32:$tlst, imm:$attr_offset, i32:$buf_idx_vgpr),
- (BUFFER_LOAD_FORMAT_XYZW_IDXEN $tlst, $buf_idx_vgpr, imm:$attr_offset, 0, 0, 0, 0)
+ (BUFFER_LOAD_FORMAT_XYZW_IDXEN $tlst, $buf_idx_vgpr, 0, imm:$attr_offset, 0, 0, 0)
>;
/* int_SI_export */
//===----------------------------------------------------------------------===//
let Predicates = [UnsafeFPMath] in {
-def : RcpPat<V_RCP_F64_e32, f64>;
-defm : RsqPat<V_RSQ_F64_e32, f64>;
-defm : RsqPat<V_RSQ_F32_e32, f32>;
+
+//def : RcpPat<V_RCP_F64_e32, f64>;
+//defm : RsqPat<V_RSQ_F64_e32, f64>;
+//defm : RsqPat<V_RSQ_F32_e32, f32>;
+
+def : RsqPat<V_RSQ_F32_e32, f32>;
+def : RsqPat<V_RSQ_F64_e32, f64>;
}
//===----------------------------------------------------------------------===//
(V_MUL_LEGACY_F32_e32 $src0, (V_RCP_LEGACY_F32_e32 $src1))
>;
-def : Pat<
- (fdiv f64:$src0, f64:$src1),
- (V_MUL_F64 0 /* src0_modifiers */, $src0,
- 0 /* src1_modifiers */, (V_RCP_F64_e32 $src1),
- 0 /* clamp */, 0 /* omod */)
->;
-
def : Pat <
(int_AMDGPU_cube v4f32:$src),
(REG_SEQUENCE VReg_128,
(V_MUL_HI_I32 $src0, $src1)
>;
-def : Vop3ModPat<V_MAD_F32, VOP_F32_F32_F32_F32, AMDGPUmad>;
-
-
defm : BFIPatterns <V_BFI_B32, S_MOV_B32, SReg_64>;
def : ROTRPattern <V_ALIGNBIT_B32>;
multiclass MUBUFLoad_Pattern <MUBUF Instr_ADDR64, ValueType vt,
PatFrag constant_ld> {
def : Pat <
- (vt (constant_ld (MUBUFAddr64 v4i32:$srsrc, i64:$vaddr, i16:$offset))),
- (Instr_ADDR64 $srsrc, $vaddr, $offset)
+ (vt (constant_ld (MUBUFAddr64 v4i32:$srsrc, i64:$vaddr, i32:$soffset,
+ i16:$offset, i1:$glc, i1:$slc, i1:$tfe))),
+ (Instr_ADDR64 $srsrc, $vaddr, $soffset, $offset, $glc, $slc, $tfe)
>;
}
(vt (int_SI_buffer_load_dword v4i32:$rsrc, (i32 imm), i32:$soffset,
imm:$offset, 0, 0, imm:$glc, imm:$slc,
imm:$tfe)),
- (offset $rsrc, (as_i16imm $offset), $soffset, (as_i1imm $glc),
+ (offset $rsrc, $soffset, (as_i16imm $offset), (as_i1imm $glc),
(as_i1imm $slc), (as_i1imm $tfe))
>;
(vt (int_SI_buffer_load_dword v4i32:$rsrc, i32:$vaddr, i32:$soffset,
imm:$offset, 0, 1, imm:$glc, imm:$slc,
imm:$tfe)),
- (idxen $rsrc, $vaddr, (as_i16imm $offset), $soffset, (as_i1imm $glc),
+ (idxen $rsrc, $vaddr, $soffset, (as_i16imm $offset), (as_i1imm $glc),
(as_i1imm $slc), (as_i1imm $tfe))
>;
def : Pat <
(vt (int_SI_buffer_load_dword v4i32:$rsrc, v2i32:$vaddr, i32:$soffset,
- imm, 1, 1, imm:$glc, imm:$slc,
+ imm:$offset, 1, 1, imm:$glc, imm:$slc,
imm:$tfe)),
- (bothen $rsrc, $vaddr, $soffset, (as_i1imm $glc), (as_i1imm $slc),
+ (bothen $rsrc, $vaddr, $soffset, (as_i16imm $offset), (as_i1imm $glc), (as_i1imm $slc),
(as_i1imm $tfe))
>;
}
(V_CMP_EQ_I32_e64 (V_AND_B32_e64 (i32 1), $a), 1)
>;
+def : Pat <
+ (i1 (trunc i64:$a)),
+ (V_CMP_EQ_I32_e64 (V_AND_B32_e64 (i32 1),
+ (EXTRACT_SUBREG $a, sub0)), 1)
+>;
+
def : Pat <
(i32 (bswap i32:$a)),
(V_BFI_B32 (S_MOV_B32 0x00ff00ff),
projects / oota-llvm.git / blobdiff