Add a late SSEDomainFix pass that twiddles SSE instructions to avoid domain crossings.
On Nehalem and newer CPUs there is a 2 cycle latency penalty on using a register
in a different domain than where it was defined. Some instructions have
equvivalents for different domains, like por/orps/orpd.
The SSEDomainFix pass tries to minimize the number of domain crossings by
changing between equvivalent opcodes where possible.
This is a work in progress, in particular the pass doesn't do anything yet. SSE
instructions are tagged with their execution domain in TableGen using the last
two bits of TSFlags. Note that not all instructions are tagged correctly. Life
just isn't that simple.
The SSE execution domain issue is very similar to the ARM NEON/VFP pipeline
issue handled by NEONMoveFixPass. This pass may become target independent to
handle both.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@99524 91177308-0d34-0410-b5e6-96231b3b80d8
diff --git a/lib/Target/X86/X86InstrInfo.h b/lib/Target/X86/X86InstrInfo.h
index 5111719..965740d 100644
--- a/lib/Target/X86/X86InstrInfo.h
+++ b/lib/Target/X86/X86InstrInfo.h
@@ -398,7 +398,10 @@
FS = 1 << SegOvrShift,
GS = 2 << SegOvrShift,
- // Bits 22 -> 23 are unused
+ // Execution domain for SSE instructions in bits 22, 23.
+ // 0 in bits 22-23 means normal, non-SSE instruction. See SSEDomain below.
+ SSEDomainShift = 22,
+
OpcodeShift = 24,
OpcodeMask = 0xFF << OpcodeShift
};
@@ -486,7 +489,7 @@
/// MemOp2RegOpTable - Load / store unfolding opcode map.
///
DenseMap<unsigned*, std::pair<unsigned, unsigned> > MemOp2RegOpTable;
-
+
public:
explicit X86InstrInfo(X86TargetMachine &tm);
@@ -716,6 +719,15 @@
///
unsigned getGlobalBaseReg(MachineFunction *MF) const;
+ /// Some SSE instructions come in variants for three domains.
+ enum SSEDomain { NotSSEDomain, PackedInt, PackedSingle, PackedDouble };
+
+ /// GetSSEDomain - Return the SSE execution domain of MI, or NotSSEDomain for
+ /// unknown instructions. If the instruction has equivalents for other
+ /// domains, equiv points to a list of opcodes for [PackedInt, PackedSingle,
+ /// PackedDouble].
+ SSEDomain GetSSEDomain(const MachineInstr *MI, const unsigned *&equiv) const;
+
private:
MachineInstr * convertToThreeAddressWithLEA(unsigned MIOpc,
MachineFunction::iterator &MFI,