maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

R600/SI: Move more information into SIProgramInfo struct 

llvm-svn: 223154
This commit is contained in:
Tom Stellard 2014-12-02 21:28:53 +00:00
parent 71ae83c4e1
commit 4df465bd5e
5 changed files with 87 additions and 57 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

104
llvm/lib/Target/R600/AMDGPUAsmPrinter.cpp
View File

@ -240,6 +240,8 @@ void AMDGPUAsmPrinter::EmitProgramInfoR600(const MachineFunction &MF) {
void AMDGPUAsmPrinter::getSIProgramInfo(SIProgramInfo &ProgInfo,
const MachineFunction &MF) const {
const AMDGPUSubtarget &STM = TM.getSubtarget<AMDGPUSubtarget>();
const SIMachineFunctionInfo *MFI = MF.getInfo<SIMachineFunctionInfo>();
uint64_t CodeSize = 0;
unsigned MaxSGPR = 0;
unsigned MaxVGPR = 0;
@ -340,6 +342,8 @@ void AMDGPUAsmPrinter::getSIProgramInfo(SIProgramInfo &ProgInfo,
ProgInfo.NumVGPR = MaxVGPR + 1;
ProgInfo.NumSGPR = MaxSGPR + 1;
ProgInfo.VGPRBlocks = (ProgInfo.NumVGPR - 1) / 4;
ProgInfo.SGPRBlocks = (ProgInfo.NumSGPR - 1) / 8;
// Set the value to initialize FP_ROUND and FP_DENORM parts of the mode
// register.
ProgInfo.FloatMode = getFPMode(MF);
@ -356,6 +360,51 @@ void AMDGPUAsmPrinter::getSIProgramInfo(SIProgramInfo &ProgInfo,
ProgInfo.FlatUsed = FlatUsed;
ProgInfo.VCCUsed = VCCUsed;
ProgInfo.CodeLen = CodeSize;
unsigned LDSAlignShift;
if (STM.getGeneration() < AMDGPUSubtarget::SEA_ISLANDS) {
// LDS is allocated in 64 dword blocks.
LDSAlignShift = 8;
} else {
// LDS is allocated in 128 dword blocks.
LDSAlignShift = 9;
}
unsigned LDSSpillSize = MFI->LDSWaveSpillSize *
MFI->getMaximumWorkGroupSize(MF);
ProgInfo.LDSSize = MFI->LDSSize + LDSSpillSize;
ProgInfo.LDSBlocks =
RoundUpToAlignment(ProgInfo.LDSSize, 1 << LDSAlignShift) >> LDSAlignShift;
// Scratch is allocated in 256 dword blocks.
unsigned ScratchAlignShift = 10;
// We need to program the hardware with the amount of scratch memory that
// is used by the entire wave. ProgInfo.ScratchSize is the amount of
// scratch memory used per thread.
ProgInfo.ScratchBlocks =
RoundUpToAlignment(ProgInfo.ScratchSize * STM.getWavefrontSize(),
1 << ScratchAlignShift) >> ScratchAlignShift;
ProgInfo.ComputePGMRSrc1 =
S_00B848_VGPRS(ProgInfo.VGPRBlocks) |
S_00B848_SGPRS(ProgInfo.SGPRBlocks) |
S_00B848_PRIORITY(ProgInfo.Priority) |
S_00B848_FLOAT_MODE(ProgInfo.FloatMode) |
S_00B848_PRIV(ProgInfo.Priv) |
S_00B848_DX10_CLAMP(ProgInfo.DX10Clamp) |
S_00B848_IEEE_MODE(ProgInfo.DebugMode) |
S_00B848_IEEE_MODE(ProgInfo.IEEEMode);
ProgInfo.ComputePGMRSrc2 =
S_00B84C_SCRATCH_EN(ProgInfo.ScratchBlocks > 0) |
S_00B84C_USER_SGPR(MFI->NumUserSGPRs) |
S_00B84C_TGID_X_EN(1) |
S_00B84C_TGID_Y_EN(1) |
S_00B84C_TGID_Z_EN(1) |
S_00B84C_TG_SIZE_EN(1) |
S_00B84C_TIDIG_COMP_CNT(2) |
S_00B84C_LDS_SIZE(ProgInfo.LDSBlocks);
}
static unsigned getRsrcReg(unsigned ShaderType) {
@ -370,74 +419,31 @@ static unsigned getRsrcReg(unsigned ShaderType) {
void AMDGPUAsmPrinter::EmitProgramInfoSI(const MachineFunction &MF,
const SIProgramInfo &KernelInfo) {
const AMDGPUSubtarget &STM = TM.getSubtarget<AMDGPUSubtarget>();
const SIMachineFunctionInfo *MFI = MF.getInfo<SIMachineFunctionInfo>();
unsigned RsrcReg = getRsrcReg(MFI->getShaderType());
unsigned LDSAlignShift;
if (STM.getGeneration() < AMDGPUSubtarget::SEA_ISLANDS) {
// LDS is allocated in 64 dword blocks.
LDSAlignShift = 8;
} else {
// LDS is allocated in 128 dword blocks.
LDSAlignShift = 9;
}
unsigned LDSSpillSize = MFI->LDSWaveSpillSize *
MFI->getMaximumWorkGroupSize(MF);
unsigned LDSBlocks =
RoundUpToAlignment(MFI->LDSSize + LDSSpillSize,
1 << LDSAlignShift) >> LDSAlignShift;
// Scratch is allocated in 256 dword blocks.
unsigned ScratchAlignShift = 10;
// We need to program the hardware with the amount of scratch memory that
// is used by the entire wave. KernelInfo.ScratchSize is the amount of
// scratch memory used per thread.
unsigned ScratchBlocks =
RoundUpToAlignment(KernelInfo.ScratchSize * STM.getWavefrontSize(),
1 << ScratchAlignShift) >> ScratchAlignShift;
unsigned VGPRBlocks = (KernelInfo.NumVGPR - 1) / 4;
unsigned SGPRBlocks = (KernelInfo.NumSGPR - 1) / 8;
if (MFI->getShaderType() == ShaderType::COMPUTE) {
OutStreamer.EmitIntValue(R_00B848_COMPUTE_PGM_RSRC1, 4);
const uint32_t ComputePGMRSrc1 =
S_00B848_VGPRS(VGPRBlocks) |
S_00B848_SGPRS(SGPRBlocks) |
S_00B848_PRIORITY(KernelInfo.Priority) |
S_00B848_FLOAT_MODE(KernelInfo.FloatMode) |
S_00B848_PRIV(KernelInfo.Priv) |
S_00B848_DX10_CLAMP(KernelInfo.DX10Clamp) |
S_00B848_IEEE_MODE(KernelInfo.DebugMode) |
S_00B848_IEEE_MODE(KernelInfo.IEEEMode);
OutStreamer.EmitIntValue(ComputePGMRSrc1, 4);
OutStreamer.EmitIntValue(KernelInfo.ComputePGMRSrc1, 4);
OutStreamer.EmitIntValue(R_00B84C_COMPUTE_PGM_RSRC2, 4);
const uint32_t ComputePGMRSrc2 =
S_00B84C_LDS_SIZE(LDSBlocks) |
S_00B02C_SCRATCH_EN(ScratchBlocks > 0);
OutStreamer.EmitIntValue(ComputePGMRSrc2, 4);
OutStreamer.EmitIntValue(KernelInfo.ComputePGMRSrc2, 4);
OutStreamer.EmitIntValue(R_00B860_COMPUTE_TMPRING_SIZE, 4);
OutStreamer.EmitIntValue(S_00B860_WAVESIZE(ScratchBlocks), 4);
OutStreamer.EmitIntValue(S_00B860_WAVESIZE(KernelInfo.ScratchBlocks), 4);
// TODO: Should probably note flat usage somewhere. SC emits a "FlatPtr32 =
// 0" comment but I don't see a corresponding field in the register spec.
} else {
OutStreamer.EmitIntValue(RsrcReg, 4);
OutStreamer.EmitIntValue(S_00B028_VGPRS(VGPRBlocks) |
S_00B028_SGPRS(SGPRBlocks), 4);
OutStreamer.EmitIntValue(S_00B028_VGPRS(KernelInfo.VGPRBlocks) |
S_00B028_SGPRS(KernelInfo.SGPRBlocks), 4);
}
if (MFI->getShaderType() == ShaderType::PIXEL) {
OutStreamer.EmitIntValue(R_00B02C_SPI_SHADER_PGM_RSRC2_PS, 4);
OutStreamer.EmitIntValue(S_00B02C_EXTRA_LDS_SIZE(LDSBlocks), 4);
OutStreamer.EmitIntValue(S_00B02C_EXTRA_LDS_SIZE(KernelInfo.LDSBlocks), 4);
OutStreamer.EmitIntValue(R_0286CC_SPI_PS_INPUT_ENA, 4);
OutStreamer.EmitIntValue(MFI->PSInputAddr, 4);
}

25
llvm/lib/Target/R600/AMDGPUAsmPrinter.h
View File

@ -24,8 +24,8 @@ class AMDGPUAsmPrinter : public AsmPrinter {
private:
struct SIProgramInfo {
SIProgramInfo() :
NumVGPR(0),
NumSGPR(0),
VGPRBlocks(0),
SGPRBlocks(0),
Priority(0),
FloatMode(0),
Priv(0),
@ -33,13 +33,19 @@ private:
DebugMode(0),
IEEEMode(0),
ScratchSize(0),
ComputePGMRSrc1(0),
LDSBlocks(0),
ScratchBlocks(0),
ComputePGMRSrc2(0),
NumVGPR(0),
NumSGPR(0),
FlatUsed(false),
VCCUsed(false),
CodeLen(0) {}
// Fields set in PGM_RSRC1 pm4 packet.
uint32_t NumVGPR;
uint32_t NumSGPR;
uint32_t VGPRBlocks;
uint32_t SGPRBlocks;
uint32_t Priority;
uint32_t FloatMode;
uint32_t Priv;
@ -48,6 +54,17 @@ private:
uint32_t IEEEMode;
uint32_t ScratchSize;
uint64_t ComputePGMRSrc1;
// Fields set in PGM_RSRC2 pm4 packet.
uint32_t LDSBlocks;
uint32_t ScratchBlocks;
uint64_t ComputePGMRSrc2;
uint32_t NumVGPR;
uint32_t NumSGPR;
uint32_t LDSSize;
bool FlatUsed;
// Bonus information for debugging.

9
llvm/lib/Target/R600/SIDefines.h
View File

@ -71,7 +71,14 @@ namespace SIOutMods {
#define S_00B028_VGPRS(x) (((x) & 0x3F) << 0)
#define S_00B028_SGPRS(x) (((x) & 0x0F) << 6)
#define R_00B84C_COMPUTE_PGM_RSRC2 0x00B84C
#define S_00B02C_SCRATCH_EN(x) (((x) & 0x1) << 0)
#define S_00B84C_SCRATCH_EN(x) (((x) & 0x1) << 0)
#define S_00B84C_USER_SGPR(x) (((x) & 0x1F) << 1)
#define S_00B84C_TGID_X_EN(x) (((x) & 0x1) << 7)
#define S_00B84C_TGID_Y_EN(x) (((x) & 0x1) << 8)
#define S_00B84C_TGID_Z_EN(x) (((x) & 0x1) << 9)
#define S_00B84C_TG_SIZE_EN(x) (((x) & 0x1) << 10)
#define S_00B84C_TIDIG_COMP_CNT(x) (((x) & 0x03) << 11)
#define S_00B84C_LDS_SIZE(x) (((x) & 0x1FF) << 15)
#define R_0286CC_SPI_PS_INPUT_ENA 0x0286CC

2
llvm/test/CodeGen/R600/local-memory-two-objects.ll
View File

@ -11,7 +11,7 @@
; EG-CHECK: .long 166120
; EG-CHECK-NEXT: .long 8
; SI-CHECK: .long 47180
; SI-CHECK-NEXT: .long 32768
; SI-CHECK-NEXT: .long 38792
; We would like to check the the lds writes are using different
; addresses, but due to variations in the scheduler, we can't do

4
llvm/test/CodeGen/R600/local-memory.ll
View File

@ -10,9 +10,9 @@
; EG: .long 166120
; EG-NEXT: .long 128
; SI: .long 47180
; SI-NEXT: .long 65536
; SI-NEXT: .long 71560
; CI: .long 47180
; CI-NEXT: .long 32768
; CI-NEXT: .long 38792
; EG: LDS_WRITE
; SI-NOT: s_wqm_b64