2012-12-12 05:25:42 +08:00
|
|
|
set(LLVM_TARGET_DEFINITIONS AMDGPU.td)
|
|
|
|
|
2018-04-04 20:37:44 +08:00
|
|
|
tablegen(LLVM AMDGPUGenAsmMatcher.inc -gen-asm-matcher)
|
|
|
|
tablegen(LLVM AMDGPUGenAsmWriter.inc -gen-asm-writer)
|
2012-12-12 05:25:42 +08:00
|
|
|
tablegen(LLVM AMDGPUGenCallingConv.inc -gen-callingconv)
|
2018-04-04 20:37:44 +08:00
|
|
|
tablegen(LLVM AMDGPUGenDAGISel.inc -gen-dag-isel)
|
2012-12-12 05:25:42 +08:00
|
|
|
tablegen(LLVM AMDGPUGenDFAPacketizer.inc -gen-dfa-packetizer)
|
2016-02-18 11:42:32 +08:00
|
|
|
tablegen(LLVM AMDGPUGenDisassemblerTables.inc -gen-disassembler)
|
2018-04-04 20:37:44 +08:00
|
|
|
tablegen(LLVM AMDGPUGenInstrInfo.inc -gen-instr-info)
|
|
|
|
tablegen(LLVM AMDGPUGenIntrinsics.inc -gen-tgt-intrinsic)
|
|
|
|
tablegen(LLVM AMDGPUGenMCCodeEmitter.inc -gen-emitter)
|
2016-10-07 01:19:11 +08:00
|
|
|
tablegen(LLVM AMDGPUGenMCPseudoLowering.inc -gen-pseudo-lowering)
|
2017-08-04 05:52:25 +08:00
|
|
|
tablegen(LLVM AMDGPUGenRegisterBank.inc -gen-register-bank)
|
2018-04-04 20:37:44 +08:00
|
|
|
tablegen(LLVM AMDGPUGenRegisterInfo.inc -gen-register-info)
|
AMDGPU: Make getTgtMemIntrinsic table-driven for resource-based intrinsics
Summary:
Avoids having to list all intrinsics manually.
This is in preparation for the new dimension-aware image intrinsics,
which I'd rather not have to list here by hand.
Change-Id: If7ced04998397ef68c4cb8f7de66b5050fb767e5
Reviewers: arsenm, rampitec, b-sumner
Subscribers: kzhuravl, wdng, mgorny, yaxunl, dstuttard, tpr, llvm-commits, t-tye
Differential Revision: https://reviews.llvm.org/D44937
llvm-svn: 328938
2018-04-02 01:09:07 +08:00
|
|
|
tablegen(LLVM AMDGPUGenSearchableTables.inc -gen-searchable-tables)
|
2018-04-04 20:37:44 +08:00
|
|
|
tablegen(LLVM AMDGPUGenSubtargetInfo.inc -gen-subtarget)
|
|
|
|
|
AMDGPU/GlobalISel: Enable TableGen'd instruction selector
Reviewers: arsenm, nhaehnle
Reviewed By: arsenm
Subscribers: kzhuravl, wdng, mgorny, yaxunl, rovka, kristof.beyls, dstuttard, tpr, t-tye, llvm-commits
Differential Revision: https://reviews.llvm.org/D45994
llvm-svn: 332039
2018-05-11 04:53:06 +08:00
|
|
|
set(LLVM_TARGET_DEFINITIONS AMDGPUGISel.td)
|
|
|
|
tablegen(LLVM AMDGPUGenGlobalISel.inc -gen-global-isel)
|
|
|
|
|
2012-12-12 05:25:42 +08:00
|
|
|
add_public_tablegen_target(AMDGPUCommonTableGen)
|
|
|
|
|
2015-06-13 11:28:10 +08:00
|
|
|
add_llvm_target(AMDGPUCodeGen
|
2017-03-18 07:56:58 +08:00
|
|
|
AMDGPUAliasAnalysis.cpp
|
2014-11-04 03:49:05 +08:00
|
|
|
AMDGPUAlwaysInlinePass.cpp
|
2015-11-07 02:01:57 +08:00
|
|
|
AMDGPUAnnotateKernelFeatures.cpp
|
2015-12-16 04:55:55 +08:00
|
|
|
AMDGPUAnnotateUniformValues.cpp
|
2017-08-04 06:30:46 +08:00
|
|
|
AMDGPUArgumentUsageInfo.cpp
|
2012-12-12 05:25:42 +08:00
|
|
|
AMDGPUAsmPrinter.cpp
|
2017-08-04 05:52:25 +08:00
|
|
|
AMDGPUCallLowering.cpp
|
2016-06-24 15:07:55 +08:00
|
|
|
AMDGPUCodeGenPrepare.cpp
|
2013-02-07 01:32:29 +08:00
|
|
|
AMDGPUFrameLowering.cpp
|
2017-08-17 00:23:32 +08:00
|
|
|
AMDGPUInstrInfo.cpp
|
2017-08-04 05:52:25 +08:00
|
|
|
AMDGPUInstructionSelector.cpp
|
2014-06-24 02:00:31 +08:00
|
|
|
AMDGPUIntrinsicInfo.cpp
|
2013-07-23 09:48:29 +08:00
|
|
|
AMDGPUISelDAGToDAG.cpp
|
2017-08-17 00:23:32 +08:00
|
|
|
AMDGPUISelLowering.cpp
|
|
|
|
AMDGPULegalizerInfo.cpp
|
|
|
|
AMDGPULibCalls.cpp
|
|
|
|
AMDGPULibFunc.cpp
|
2017-02-10 06:00:42 +08:00
|
|
|
AMDGPULowerIntrinsics.cpp
|
2018-05-19 05:35:00 +08:00
|
|
|
AMDGPULowerKernelAttributes.cpp
|
2017-05-16 04:18:37 +08:00
|
|
|
AMDGPUMachineCFGStructurizer.cpp
|
2013-04-02 06:05:58 +08:00
|
|
|
AMDGPUMachineFunction.cpp
|
2017-07-22 05:19:23 +08:00
|
|
|
AMDGPUMachineModuleInfo.cpp
|
2017-08-17 00:23:32 +08:00
|
|
|
AMDGPUMacroFusion.cpp
|
|
|
|
AMDGPUMCInstLower.cpp
|
2017-10-11 03:39:48 +08:00
|
|
|
AMDGPUOpenCLEnqueuedBlockLowering.cpp
|
2014-06-18 00:53:14 +08:00
|
|
|
AMDGPUPromoteAlloca.cpp
|
2017-06-09 03:25:21 +08:00
|
|
|
AMDGPURegAsmNames.inc.cpp
|
2017-08-04 05:52:25 +08:00
|
|
|
AMDGPURegisterBankInfo.cpp
|
2012-12-12 05:25:42 +08:00
|
|
|
AMDGPURegisterInfo.cpp
|
2017-07-29 02:40:05 +08:00
|
|
|
AMDGPURewriteOutArguments.cpp
|
2017-08-17 00:23:32 +08:00
|
|
|
AMDGPUSubtarget.cpp
|
|
|
|
AMDGPUTargetMachine.cpp
|
|
|
|
AMDGPUTargetObjectFile.cpp
|
|
|
|
AMDGPUTargetTransformInfo.cpp
|
2017-03-25 03:52:05 +08:00
|
|
|
AMDGPUUnifyDivergentExitNodes.cpp
|
2017-08-17 00:23:32 +08:00
|
|
|
AMDGPUUnifyMetadata.cpp
|
2017-09-20 12:25:58 +08:00
|
|
|
AMDGPUInline.cpp
|
2018-05-26 01:25:12 +08:00
|
|
|
AMDGPUPerfHintAnalysis.cpp
|
2017-08-17 00:23:32 +08:00
|
|
|
AMDILCFGStructurizer.cpp
|
2016-04-30 08:23:06 +08:00
|
|
|
GCNHazardRecognizer.cpp
|
2017-08-17 00:23:32 +08:00
|
|
|
GCNIterativeScheduler.cpp
|
|
|
|
GCNMinRegStrategy.cpp
|
|
|
|
GCNRegPressure.cpp
|
2016-08-30 03:42:52 +08:00
|
|
|
GCNSchedStrategy.cpp
|
2018-05-25 04:02:01 +08:00
|
|
|
R600AsmPrinter.cpp
|
2013-10-02 03:32:58 +08:00
|
|
|
R600ClauseMergePass.cpp
|
2013-04-02 06:05:58 +08:00
|
|
|
R600ControlFlowFinalizer.cpp
|
|
|
|
R600EmitClauseMarkers.cpp
|
2012-12-12 05:25:42 +08:00
|
|
|
R600ExpandSpecialInstrs.cpp
|
2016-06-24 14:30:11 +08:00
|
|
|
R600FrameLowering.cpp
|
2012-12-12 05:25:42 +08:00
|
|
|
R600InstrInfo.cpp
|
|
|
|
R600ISelLowering.cpp
|
|
|
|
R600MachineFunctionInfo.cpp
|
2013-03-06 02:54:05 +08:00
|
|
|
R600MachineScheduler.cpp
|
2018-05-13 18:04:48 +08:00
|
|
|
R600OpenCLImageTypeLoweringPass.cpp
|
2013-06-06 05:38:04 +08:00
|
|
|
R600OptimizeVectorRegisters.cpp
|
2013-04-30 08:14:27 +08:00
|
|
|
R600Packetizer.cpp
|
2012-12-12 05:25:42 +08:00
|
|
|
R600RegisterInfo.cpp
|
2012-12-20 06:10:31 +08:00
|
|
|
SIAnnotateControlFlow.cpp
|
2016-05-11 02:33:41 +08:00
|
|
|
SIDebuggerInsertNops.cpp
|
2013-08-07 07:12:34 +08:00
|
|
|
SIFixSGPRCopies.cpp
|
2017-01-25 01:46:17 +08:00
|
|
|
SIFixVGPRCopies.cpp
|
[AMDGPU] Add support for Whole Wavefront Mode
Summary:
Whole Wavefront Wode (WWM) is similar to WQM, except that all of the
lanes are always enabled, regardless of control flow. This is required
for implementing wavefront reductions in non-uniform control flow, where
we need to use the inactive lanes to propagate intermediate results, so
they need to be enabled. We need to propagate WWM to uses (unless
they're explicitly marked as exact) so that they also propagate
intermediate results correctly. We do the analysis and exec mask munging
during the WQM pass, since there are interactions with WQM for things
that require both WQM and WWM. For simplicity, WWM is entirely
block-local -- blocks are never WWM on entry or exit of a block, and WWM
is not propagated to the block level. This means that computations
involving WWM cannot involve control flow, but we only ever plan to use
WWM for a few limited purposes (none of which involve control flow)
anyways.
Shaders can ask for WWM using the @llvm.amdgcn.wwm intrinsic. There
isn't yet a way to turn WWM off -- that will be added in a future
change.
Finally, it turns out that turning on inactive lanes causes a number of
problems with register allocation. While the best long-term solution
seems like teaching LLVM's register allocator about predication, for now
we need to add some hacks to prevent ourselves from getting into trouble
due to constraints that aren't currently expressed in LLVM. For the gory
details, see the comments at the top of SIFixWWMLiveness.cpp.
Reviewers: arsenm, nhaehnle, tpr
Subscribers: kzhuravl, wdng, mgorny, yaxunl, dstuttard, t-tye, llvm-commits
Differential Revision: https://reviews.llvm.org/D35524
llvm-svn: 310087
2017-08-05 02:36:52 +08:00
|
|
|
SIFixWWMLiveness.cpp
|
2014-11-22 06:06:37 +08:00
|
|
|
SIFoldOperands.cpp
|
2018-06-01 04:13:51 +08:00
|
|
|
SIFormMemoryClauses.cpp
|
2015-11-07 02:17:45 +08:00
|
|
|
SIFrameLowering.cpp
|
2016-08-23 03:33:16 +08:00
|
|
|
SIInsertSkips.cpp
|
2017-04-12 11:25:12 +08:00
|
|
|
SIInsertWaitcnts.cpp
|
2012-12-12 05:25:42 +08:00
|
|
|
SIInstrInfo.cpp
|
|
|
|
SIISelLowering.cpp
|
2014-10-11 06:01:59 +08:00
|
|
|
SILoadStoreOptimizer.cpp
|
2012-12-12 05:25:42 +08:00
|
|
|
SILowerControlFlow.cpp
|
2014-04-30 23:31:33 +08:00
|
|
|
SILowerI1Copies.cpp
|
2012-12-12 05:25:42 +08:00
|
|
|
SIMachineFunctionInfo.cpp
|
2016-01-14 00:10:10 +08:00
|
|
|
SIMachineScheduler.cpp
|
2017-07-22 05:19:23 +08:00
|
|
|
SIMemoryLegalizer.cpp
|
2016-09-29 09:44:16 +08:00
|
|
|
SIOptimizeExecMasking.cpp
|
2017-08-02 07:14:32 +08:00
|
|
|
SIOptimizeExecMaskingPreRA.cpp
|
[ADMGPU] SDWA peephole optimization pass.
Summary:
First iteration of SDWA peephole.
This pass tries to combine several instruction into one SDWA instruction. E.g. it converts:
'''
V_LSHRREV_B32_e32 %vreg0, 16, %vreg1
V_ADD_I32_e32 %vreg2, %vreg0, %vreg3
V_LSHLREV_B32_e32 %vreg4, 16, %vreg2
'''
Into:
'''
V_ADD_I32_sdwa %vreg4, %vreg1, %vreg3 dst_sel:WORD_1 dst_unused:UNUSED_PAD src0_sel:WORD_1 src1_sel:DWORD
'''
Pass structure:
1. Iterate over machine instruction in basic block and try to apply "SDWA patterns" to each of them. SDWA patterns match machine instruction into either source or destination SDWA operand. E.g. ''' V_LSHRREV_B32_e32 %vreg0, 16, %vreg1''' is matched to source SDWA operand '''%vreg1 src_sel:WORD_1'''.
2. Iterate over found SDWA operands and find instruction that could be potentially coverted into SDWA. E.g. for source SDWA operand potential instruction are all instruction in this basic block that uses '''%vreg0'''
3. Iterate over all potential instructions and check if they can be converted into SDWA.
4. Convert instructions to SDWA.
This review contains basic implementation of SDWA peephole pass. This pass requires additional testing fot both correctness and performance (no performance testing done).
There are several ways this pass can be improved:
1. Make this pass work on whole function not only basic block. As I can see this can be done right now without changes to pass.
2. Introduce more SDWA patterns
3. Introduce mnemonics to limit when SDWA patterns should apply
Reviewers: vpykhtin, alex-t, arsenm, rampitec
Subscribers: wdng, nhaehnle, mgorny
Differential Revision: https://reviews.llvm.org/D30038
llvm-svn: 298365
2017-03-21 20:51:34 +08:00
|
|
|
SIPeepholeSDWA.cpp
|
2012-12-12 05:25:42 +08:00
|
|
|
SIRegisterInfo.cpp
|
2014-07-22 00:55:33 +08:00
|
|
|
SIShrinkInstructions.cpp
|
2016-03-22 04:28:33 +08:00
|
|
|
SIWholeQuadMode.cpp
|
2017-11-20 22:35:53 +08:00
|
|
|
GCNILPSched.cpp
|
2012-12-12 05:25:42 +08:00
|
|
|
)
|
|
|
|
|
2014-11-14 22:08:00 +08:00
|
|
|
add_subdirectory(AsmParser)
|
2016-02-18 11:42:32 +08:00
|
|
|
add_subdirectory(Disassembler)
|
2018-04-23 20:49:34 +08:00
|
|
|
add_subdirectory(InstPrinter)
|
2012-12-12 05:25:42 +08:00
|
|
|
add_subdirectory(MCTargetDesc)
|
2018-04-23 20:49:34 +08:00
|
|
|
add_subdirectory(TargetInfo)
|
2015-06-27 05:15:07 +08:00
|
|
|
add_subdirectory(Utils)
|