mirror of
https://github.com/c64scene-ar/llvm-6502.git
synced 2024-08-30 21:29:28 +00:00
It's completely unnecessary and can be replace with proper SReg_64 handling instead. This actually fixes a piglit test on SI. v2: use correct register class in addRegisterClass, set special classes as not allocatable v3: revert setting special classes as not allocateable This is a candidate for the stable branch. Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Tom Stellard <thomas.stellard@amd.com> git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@175355 91177308-0d34-0410-b5e6-96231b3b80d8
191 lines
7.4 KiB
TableGen
191 lines
7.4 KiB
TableGen
|
|
class SIReg <string n, bits<16> encoding = 0> : Register<n> {
|
|
let Namespace = "AMDGPU";
|
|
let HWEncoding = encoding;
|
|
}
|
|
|
|
class SI_64 <string n, list<Register> subregs, bits<16> encoding> : RegisterWithSubRegs<n, subregs> {
|
|
let Namespace = "AMDGPU";
|
|
let SubRegIndices = [sub0, sub1];
|
|
let HWEncoding = encoding;
|
|
}
|
|
|
|
class SGPR_32 <bits<16> num, string name> : SIReg<name, num>;
|
|
|
|
class VGPR_32 <bits<16> num, string name> : SIReg<name, num> {
|
|
let HWEncoding{8} = 1;
|
|
}
|
|
|
|
// Special Registers
|
|
def VCC : SIReg<"VCC", 106>;
|
|
def EXEC_LO : SIReg <"EXEC LO", 126>;
|
|
def EXEC_HI : SIReg <"EXEC HI", 127>;
|
|
def EXEC : SI_64<"EXEC", [EXEC_LO, EXEC_HI], 126>;
|
|
def SCC : SIReg<"SCC", 253>;
|
|
def M0 : SIReg <"M0", 124>;
|
|
|
|
//Interpolation registers
|
|
def PERSP_SAMPLE_I : SIReg <"PERSP_SAMPLE_I">;
|
|
def PERSP_SAMPLE_J : SIReg <"PERSP_SAMPLE_J">;
|
|
def PERSP_CENTER_I : SIReg <"PERSP_CENTER_I">;
|
|
def PERSP_CENTER_J : SIReg <"PERSP_CENTER_J">;
|
|
def PERSP_CENTROID_I : SIReg <"PERSP_CENTROID_I">;
|
|
def PERSP_CENTROID_J : SIReg <"PERP_CENTROID_J">;
|
|
def PERSP_I_W : SIReg <"PERSP_I_W">;
|
|
def PERSP_J_W : SIReg <"PERSP_J_W">;
|
|
def PERSP_1_W : SIReg <"PERSP_1_W">;
|
|
def LINEAR_SAMPLE_I : SIReg <"LINEAR_SAMPLE_I">;
|
|
def LINEAR_SAMPLE_J : SIReg <"LINEAR_SAMPLE_J">;
|
|
def LINEAR_CENTER_I : SIReg <"LINEAR_CENTER_I">;
|
|
def LINEAR_CENTER_J : SIReg <"LINEAR_CENTER_J">;
|
|
def LINEAR_CENTROID_I : SIReg <"LINEAR_CENTROID_I">;
|
|
def LINEAR_CENTROID_J : SIReg <"LINEAR_CENTROID_J">;
|
|
def LINE_STIPPLE_TEX_COORD : SIReg <"LINE_STIPPLE_TEX_COORD">;
|
|
def POS_X_FLOAT : SIReg <"POS_X_FLOAT">;
|
|
def POS_Y_FLOAT : SIReg <"POS_Y_FLOAT">;
|
|
def POS_Z_FLOAT : SIReg <"POS_Z_FLOAT">;
|
|
def POS_W_FLOAT : SIReg <"POS_W_FLOAT">;
|
|
def FRONT_FACE : SIReg <"FRONT_FACE">;
|
|
def ANCILLARY : SIReg <"ANCILLARY">;
|
|
def SAMPLE_COVERAGE : SIReg <"SAMPLE_COVERAGE">;
|
|
def POS_FIXED_PT : SIReg <"POS_FIXED_PT">;
|
|
|
|
// SGPR 32-bit registers
|
|
foreach Index = 0-101 in {
|
|
def SGPR#Index : SGPR_32 <Index, "SGPR"#Index>;
|
|
}
|
|
|
|
def SGPR_32 : RegisterClass<"AMDGPU", [f32, i32], 32,
|
|
(add (sequence "SGPR%u", 0, 101))>;
|
|
|
|
// SGPR 64-bit registers
|
|
def SGPR_64 : RegisterTuples<[sub0, sub1],
|
|
[(add (decimate SGPR_32, 2)),
|
|
(add(decimate (rotl SGPR_32, 1), 2))]>;
|
|
|
|
// SGPR 128-bit registers
|
|
def SGPR_128 : RegisterTuples<[sub0, sub1, sub2, sub3],
|
|
[(add (decimate SGPR_32, 4)),
|
|
(add (decimate (rotl SGPR_32, 1), 4)),
|
|
(add (decimate (rotl SGPR_32, 2), 4)),
|
|
(add (decimate (rotl SGPR_32, 3), 4))]>;
|
|
|
|
// SGPR 256-bit registers
|
|
def SGPR_256 : RegisterTuples<[sub0, sub1, sub2, sub3, sub4, sub5, sub6, sub7],
|
|
[(add (decimate SGPR_32, 8)),
|
|
(add (decimate (rotl SGPR_32, 1), 8)),
|
|
(add (decimate (rotl SGPR_32, 2), 8)),
|
|
(add (decimate (rotl SGPR_32, 3), 8)),
|
|
(add (decimate (rotl SGPR_32, 4), 8)),
|
|
(add (decimate (rotl SGPR_32, 5), 8)),
|
|
(add (decimate (rotl SGPR_32, 6), 8)),
|
|
(add (decimate (rotl SGPR_32, 7), 8))]>;
|
|
|
|
// VGPR 32-bit registers
|
|
foreach Index = 0-255 in {
|
|
def VGPR#Index : VGPR_32 <Index, "VGPR"#Index>;
|
|
}
|
|
|
|
def VGPR_32 : RegisterClass<"AMDGPU", [f32, i32], 32,
|
|
(add (sequence "VGPR%u", 0, 255))>;
|
|
|
|
// VGPR 64-bit registers
|
|
def VGPR_64 : RegisterTuples<[sub0, sub1],
|
|
[(add VGPR_32),
|
|
(add (rotl VGPR_32, 1))]>;
|
|
|
|
// VGPR 128-bit registers
|
|
def VGPR_128 : RegisterTuples<[sub0, sub1, sub2, sub3],
|
|
[(add VGPR_32),
|
|
(add (rotl VGPR_32, 1)),
|
|
(add (rotl VGPR_32, 2)),
|
|
(add (rotl VGPR_32, 3))]>;
|
|
|
|
// VGPR 256-bit registers
|
|
def VGPR_256 : RegisterTuples<[sub0, sub1, sub2, sub3, sub4, sub5, sub6, sub7],
|
|
[(add VGPR_32),
|
|
(add (rotl VGPR_32, 1)),
|
|
(add (rotl VGPR_32, 2)),
|
|
(add (rotl VGPR_32, 3)),
|
|
(add (rotl VGPR_32, 4)),
|
|
(add (rotl VGPR_32, 5)),
|
|
(add (rotl VGPR_32, 6)),
|
|
(add (rotl VGPR_32, 7))]>;
|
|
|
|
// VGPR 512-bit registers
|
|
def VGPR_512 : RegisterTuples<[sub0, sub1, sub2, sub3, sub4, sub5, sub6, sub7,
|
|
sub8, sub9, sub10, sub11, sub12, sub13, sub14, sub15],
|
|
[(add VGPR_32),
|
|
(add (rotl VGPR_32, 1)),
|
|
(add (rotl VGPR_32, 2)),
|
|
(add (rotl VGPR_32, 3)),
|
|
(add (rotl VGPR_32, 4)),
|
|
(add (rotl VGPR_32, 5)),
|
|
(add (rotl VGPR_32, 6)),
|
|
(add (rotl VGPR_32, 7)),
|
|
(add (rotl VGPR_32, 8)),
|
|
(add (rotl VGPR_32, 9)),
|
|
(add (rotl VGPR_32, 10)),
|
|
(add (rotl VGPR_32, 11)),
|
|
(add (rotl VGPR_32, 12)),
|
|
(add (rotl VGPR_32, 13)),
|
|
(add (rotl VGPR_32, 14)),
|
|
(add (rotl VGPR_32, 15))]>;
|
|
|
|
// Register class for all scalar registers (SGPRs + Special Registers)
|
|
def SReg_32 : RegisterClass<"AMDGPU", [f32, i32], 32,
|
|
(add SGPR_32, M0, EXEC_LO, EXEC_HI)
|
|
>;
|
|
|
|
def SReg_64 : RegisterClass<"AMDGPU", [i1, i64], 64, (add SGPR_64, VCC, EXEC)>;
|
|
|
|
def SReg_128 : RegisterClass<"AMDGPU", [v4f32, v4i32], 128, (add SGPR_128)>;
|
|
|
|
def SReg_256 : RegisterClass<"AMDGPU", [v8i32], 256, (add SGPR_256)>;
|
|
|
|
// Register class for all vector registers (VGPRs + Interploation Registers)
|
|
def VReg_32 : RegisterClass<"AMDGPU", [f32, i32, v1i32], 32,
|
|
(add VGPR_32,
|
|
PERSP_SAMPLE_I, PERSP_SAMPLE_J,
|
|
PERSP_CENTER_I, PERSP_CENTER_J,
|
|
PERSP_CENTROID_I, PERSP_CENTROID_J,
|
|
PERSP_I_W, PERSP_J_W, PERSP_1_W,
|
|
LINEAR_SAMPLE_I, LINEAR_SAMPLE_J,
|
|
LINEAR_CENTER_I, LINEAR_CENTER_J,
|
|
LINEAR_CENTROID_I, LINEAR_CENTROID_J,
|
|
LINE_STIPPLE_TEX_COORD,
|
|
POS_X_FLOAT,
|
|
POS_Y_FLOAT,
|
|
POS_Z_FLOAT,
|
|
POS_W_FLOAT,
|
|
FRONT_FACE,
|
|
ANCILLARY,
|
|
SAMPLE_COVERAGE,
|
|
POS_FIXED_PT
|
|
)
|
|
>;
|
|
|
|
def VReg_64 : RegisterClass<"AMDGPU", [i64, v2i32], 64, (add VGPR_64)>;
|
|
|
|
def VReg_128 : RegisterClass<"AMDGPU", [v4f32, v4i32], 128, (add VGPR_128)>;
|
|
|
|
def VReg_256 : RegisterClass<"AMDGPU", [v8i32], 256, (add VGPR_256)>;
|
|
|
|
def VReg_512 : RegisterClass<"AMDGPU", [v16i32], 512, (add VGPR_512)>;
|
|
|
|
// [SV]Src_* operands can have either an immediate or an register
|
|
def SSrc_32 : RegisterClass<"AMDGPU", [i32, f32], 32, (add SReg_32)>;
|
|
|
|
def SSrc_64 : RegisterClass<"AMDGPU", [i1, i64], 64, (add SReg_64)>;
|
|
|
|
def VSrc_32 : RegisterClass<"AMDGPU", [i32, f32], 32, (add VReg_32, SReg_32)>;
|
|
|
|
def VSrc_64 : RegisterClass<"AMDGPU", [i64], 64, (add SReg_64, VReg_64)>;
|
|
|
|
// Special register classes for predicates and the M0 register
|
|
def SCCReg : RegisterClass<"AMDGPU", [i1], 1, (add SCC)>;
|
|
def VCCReg : RegisterClass<"AMDGPU", [i1], 1, (add VCC)>;
|
|
def EXECReg : RegisterClass<"AMDGPU", [i1], 1, (add EXEC)>;
|
|
def M0Reg : RegisterClass<"AMDGPU", [i32], 32, (add M0)>;
|
|
|