ShuriZma
/
dolphin
mirror of https://github.com/dolphin-emu/dolphin.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

JitArm64: Improve instruction scheduling in fres routine 

We want to have a low number of instructions between the LDP and the
MADD so that the MADD can start immediately after the LDP finishes
even if we're on a lower-end in-order CPU.
This commit is contained in:
JosJuice 2023-06-15 22:02:39 +02:00
parent 5b5124d9ff
commit 9c13436e2b
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
Source/Core/Core/PowerPC/JitArm64/JitAsm.cpp
View File

@ -293,12 +293,12 @@ void JitArm64::GenerateFres()
UBFX(ARM64Reg::X2, ARM64Reg::X1, 47, 5); // Grab upper part of mantissa UBFX(ARM64Reg::X2, ARM64Reg::X1, 47, 5); // Grab upper part of mantissa
MOVP2R(ARM64Reg::X3, &Common::fres_expected); MOVP2R(ARM64Reg::X3, &Common::fres_expected);
ADD(ARM64Reg::X2, ARM64Reg::X3, ARM64Reg::X2, ArithOption(ARM64Reg::X2, ShiftType::LSL, 3)); ADD(ARM64Reg::X2, ARM64Reg::X3, ARM64Reg::X2, ArithOption(ARM64Reg::X2, ShiftType::LSL, 3));
LDP(IndexType::Signed, ARM64Reg::W2, ARM64Reg::W3, ARM64Reg::X2, 0);
UBFX(ARM64Reg::X1, ARM64Reg::X1, 37, 10); // Grab lower part of mantissa UBFX(ARM64Reg::X1, ARM64Reg::X1, 37, 10); // Grab lower part of mantissa
LDP(IndexType::Signed, ARM64Reg::W2, ARM64Reg::W3, ARM64Reg::X2, 0);
MOVI2R(ARM64Reg::W4, 1); MOVI2R(ARM64Reg::W4, 1);
AND(ARM64Reg::X0, ARM64Reg::X0, LogicalImm(Common::DOUBLE_SIGN | Common::DOUBLE_EXP, 64));
MADD(ARM64Reg::W1, ARM64Reg::W3, ARM64Reg::W1, ARM64Reg::W4); MADD(ARM64Reg::W1, ARM64Reg::W3, ARM64Reg::W1, ARM64Reg::W4);
SUB(ARM64Reg::W1, ARM64Reg::W2, ARM64Reg::W1, ArithOption(ARM64Reg::W1, ShiftType::LSR, 1)); SUB(ARM64Reg::W1, ARM64Reg::W2, ARM64Reg::W1, ArithOption(ARM64Reg::W1, ShiftType::LSR, 1));
AND(ARM64Reg::X0, ARM64Reg::X0, LogicalImm(Common::DOUBLE_SIGN | Common::DOUBLE_EXP, 64));
ORR(ARM64Reg::X0, ARM64Reg::X0, ARM64Reg::X1, ArithOption(ARM64Reg::X1, ShiftType::LSL, 29)); ORR(ARM64Reg::X0, ARM64Reg::X0, ARM64Reg::X1, ArithOption(ARM64Reg::X1, ShiftType::LSL, 29));
RET(); RET();