jmolloy 548ef37194 [DFAPacketizer] Allow up to 64 functional units ... Summary: To drive the automaton we used a uint64_t as an action type. This contained the transition's resource requirements as a conjunction: (a OR b) AND (b OR c) We encoded this conjunction as a sequence of four 16-bit bitmasks. This limited the number of addressable functional units to 16, which is quite low and has bitten many people in the past. Instead, the DFAEmitter now generates a lookup table from InstrItinerary class (index of the ItinData inside the ProcItineraries) to an internal action index which is essentially a dense embedding of the conjunctive form. Because we never materialize the conjunctive form, we no longer have the 16 FU restriction. In this patch we limit to 64 functional units due to using a uint64_t bitmask in the DFAEmitter. Now that we've decoupled these representations we can increase this in future. Reviewers: ThomasRaoux, kparzysz, majnemer Reviewed By: ThomasRaoux Subscribers: hiraditya, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D69110		2019-11-05 15:41:42 +00:00
..
GlobalISel	[gicombiner] Add a CodeExpander to handle C++ fragments with variable expansion	2019-10-03 19:13:39 +00:00
AsmMatcherEmitter.cpp	Cosmetic; don't use the magic constant 35 when HASH is more readable. This matches other MCK__<THING>_* usage better.	2019-09-23 12:52:42 +00:00
AsmWriterEmitter.cpp	[tblgen] Add getOperatorAsDef() to Record	2019-10-08 18:41:32 +00:00
AsmWriterInst.cpp	[tablegen] Add locations to many PrintFatalError() calls	2019-02-12 17:36:57 +00:00
AsmWriterInst.h
Attributes.cpp
CallingConvEmitter.cpp	AArch64: support arm64_32, an ILP32 slice for watchOS.	2019-09-12 10:22:23 +00:00
CMakeLists.txt	[TableGen] Introduce a generic automaton (DFA) backend	2019-10-04 09:03:36 +00:00
CodeEmitterGen.cpp	[TableGen] Fix crash when using HwModes in CodeEmitterGen	2019-10-09 09:15:34 +00:00
CodeGenDAGPatterns.cpp	Stop static analyzer warnings about using bitwise operators on booleans. NFCI.	2019-11-02 22:40:04 +00:00
CodeGenDAGPatterns.h	TableGen: Add MinAlignment predicate	2019-07-31 00:14:43 +00:00
CodeGenHwModes.cpp
CodeGenHwModes.h
CodeGenInstruction.cpp	GlobalISel: Add target pre-isel instructions	2019-10-07 18:43:29 +00:00
CodeGenInstruction.h	GlobalISel: Add target pre-isel instructions	2019-10-07 18:43:29 +00:00
CodeGenIntrinsics.h	Teach GlobalISelEmitter to treat used iPTRAny operands as pointer operands	2019-08-20 22:04:10 +00:00
CodeGenMapTable.cpp	[TableGen] CodeGenMapTable - Don't dereference a dyn_cast result. NFCI.	2019-09-17 17:32:15 +00:00
CodeGenRegisters.cpp	GlobalISel: Support physical register inputs in patterns	2019-09-06 20:32:37 +00:00
CodeGenRegisters.h	GlobalISel: Support physical register inputs in patterns	2019-09-06 20:32:37 +00:00
CodeGenSchedule.cpp	[TableGen] Improve error reporting of overlapping definitions (NFC)	2019-10-02 19:44:53 +00:00
CodeGenSchedule.h
CodeGenTarget.cpp	[TableGen] Support encoding per-HwMode	2019-09-19 13:39:54 +00:00
CodeGenTarget.h	Recommit "[GlobalISel] Import patterns containing INSERT_SUBREG"	2019-08-27 17:47:06 +00:00
CTagsEmitter.cpp
DAGISelEmitter.cpp	[llvm] Migrate llvm::make_unique to std::make_unique	2019-08-15 15:54:37 +00:00
DAGISelMatcher.cpp	[TableGen] Replace a dyn_cast with isa to avoid an unused variable warning introduced in r355785. NFC	2019-03-11 16:51:37 +00:00
DAGISelMatcher.h	TableGen: Support physical register inputs > 255	2019-07-22 15:02:34 +00:00
DAGISelMatcherEmitter.cpp	TableGen: Support physical register inputs > 255	2019-07-22 15:02:34 +00:00
DAGISelMatcherGen.cpp	[TableGen] Correct comments for end of namespace. NFC	2019-08-25 10:47:30 +00:00
DAGISelMatcherOpt.cpp	DAGISelMatcherOpt - TGParser::ParseOperation - silence static analyzer cast_or_null<CheckTypeMatcher> null dereference warning. NFCI.	2019-09-26 17:38:47 +00:00
DFAEmitter.cpp	[TableGen] Pacify gcc-5.4 more	2019-10-07 08:23:20 +00:00
DFAEmitter.h	[TableGen] Introduce a generic automaton (DFA) backend	2019-10-04 09:03:36 +00:00
DFAPacketizerEmitter.cpp	[DFAPacketizer] Allow up to 64 functional units	2019-11-05 15:41:42 +00:00
DisassemblerEmitter.cpp	[TableGen] Correct comments for end of namespace. NFC	2019-08-25 10:47:30 +00:00
ExegesisEmitter.cpp
FastISelEmitter.cpp
FixedLenDecoderEmitter.cpp	[TableGen] Support encoding per-HwMode	2019-09-19 13:39:54 +00:00
GICombinerEmitter.cpp	[gicombiner] Add parse failure tests for defs/match	2019-10-25 12:56:49 -07:00
GlobalISelEmitter.cpp	MatchTableRecord::emit - fix boolean operator precedence warnings from PVS Studio. NFCI.	2019-11-02 21:04:07 +00:00
InfoByHwMode.cpp	[TableGen] Support encoding per-HwMode	2019-09-19 13:39:54 +00:00
InfoByHwMode.h	[TableGen] Support encoding per-HwMode	2019-09-19 13:39:54 +00:00
InstrDocsEmitter.cpp	[TableGen] Correct comments for end of namespace. NFC	2019-08-25 10:47:30 +00:00
InstrInfoEmitter.cpp	GlobalISel: Add target pre-isel instructions	2019-10-07 18:43:29 +00:00
IntrinsicEmitter.cpp	[IntrinsicEmitter] Add overloaded type VecOfBitcastsToInt for SVE intrinsics	2019-10-02 09:25:02 +00:00
LLVMBuild.txt
OptParserEmitter.cpp
PredicateExpander.cpp
PredicateExpander.h
PseudoLoweringEmitter.cpp
RegisterBankEmitter.cpp
RegisterInfoEmitter.cpp	TableGen: Use enum names in composeSubRegIndices table	2019-10-27 21:26:56 -07:00
RISCVCompressInstEmitter.cpp	[tblgen] Add getOperatorAsDef() to Record	2019-10-08 18:41:32 +00:00
SDNodeProperties.cpp	[tablegen] Add locations to many PrintFatalError() calls	2019-02-12 17:36:57 +00:00
SDNodeProperties.h
SearchableTableEmitter.cpp	Use a bit of relaxed constexpr to make FeatureBitset costant intializable	2019-08-24 15:02:44 +00:00
SequenceToOffsetTable.h	Relax assertions when there's really no entries. [NFC]	2019-10-22 15:26:30 -04:00
SubtargetEmitter.cpp	[TableGen] Fix a bug that MCSchedClassDesc is interfered between different SchedModel	2019-10-11 08:36:54 +00:00
SubtargetFeatureInfo.cpp	Use a bit of relaxed constexpr to make FeatureBitset costant intializable	2019-08-24 15:02:44 +00:00
SubtargetFeatureInfo.h	Use bitset for assembler predicates	2019-03-11 17:04:35 +00:00
TableGen.cpp	[TableGen] Introduce a generic automaton (DFA) backend	2019-10-04 09:03:36 +00:00
TableGenBackends.h	[TableGen] Introduce a generic automaton (DFA) backend	2019-10-04 09:03:36 +00:00
tdtags
Types.cpp
Types.h
WebAssemblyDisassemblerEmitter.cpp	[TableGen] Correct comments for end of namespace. NFC	2019-08-25 10:47:30 +00:00
WebAssemblyDisassemblerEmitter.h
X86DisassemblerShared.h
X86DisassemblerTables.cpp	[llvm] Migrate llvm::make_unique to std::make_unique	2019-08-15 15:54:37 +00:00
X86DisassemblerTables.h
X86EVEX2VEXTablesEmitter.cpp	[X86] Consider isCodeGenOnly in the EVEX2VEX pass to make VMAXPD/PS map to the non-commutable VEX instruction. Use EVEX2VEX override to fix the scalar instructions.	2019-10-01 07:10:09 +00:00
X86FoldTablesEmitter.cpp	[X86] Make the X86FoldTablesEmitter functional again. Fix the spacing in the output to make it easier to diff.	2019-06-01 06:20:55 +00:00
X86ModRMFilters.cpp
X86ModRMFilters.h
X86RecognizableInstr.cpp	[llvm] Migrate llvm::make_unique to std::make_unique	2019-08-15 15:54:37 +00:00
X86RecognizableInstr.h	[X86] Split the VEX_WPrefix in X86Inst tablegen class into 3 separate fields with clear meanings.	2019-04-09 07:40:06 +00:00