Skip to content

Use movups instead of movdqu in block op codegen #1367

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wants to merge 1 commit into from
Closed

Use movups instead of movdqu in block op codegen #1367

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 6 additions & 6 deletions src/coreclr/src/jit/codegenxarch.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3022,11 +3022,11 @@ void CodeGen::genCodeForInitBlkUnroll(GenTreeBlk* node)
{
if (dstLclNum != BAD_VAR_NUM)
{
emit->emitIns_S_R(INS_movdqu, EA_ATTR(regSize), srcXmmReg, dstLclNum, dstOffset);
emit->emitIns_S_R(INS_movups, EA_ATTR(regSize), srcXmmReg, dstLclNum, dstOffset);
}
else
{
emit->emitIns_ARX_R(INS_movdqu, EA_ATTR(regSize), srcXmmReg, dstAddrBaseReg, dstAddrIndexReg,
emit->emitIns_ARX_R(INS_movups, EA_ATTR(regSize), srcXmmReg, dstAddrBaseReg, dstAddrIndexReg,
dstAddrIndexScale, dstOffset);
}
}
Expand Down Expand Up @@ -3221,21 +3221,21 @@ void CodeGen::genCodeForCpBlkUnroll(GenTreeBlk* node)
{
if (srcLclNum != BAD_VAR_NUM)
{
emit->emitIns_R_S(INS_movdqu, EA_ATTR(regSize), tempReg, srcLclNum, srcOffset);
emit->emitIns_R_S(INS_movups, EA_ATTR(regSize), tempReg, srcLclNum, srcOffset);
}
else
{
emit->emitIns_R_ARX(INS_movdqu, EA_ATTR(regSize), tempReg, srcAddrBaseReg, srcAddrIndexReg,
emit->emitIns_R_ARX(INS_movups, EA_ATTR(regSize), tempReg, srcAddrBaseReg, srcAddrIndexReg,
srcAddrIndexScale, srcOffset);
}

if (dstLclNum != BAD_VAR_NUM)
{
emit->emitIns_S_R(INS_movdqu, EA_ATTR(regSize), tempReg, dstLclNum, dstOffset);
emit->emitIns_S_R(INS_movups, EA_ATTR(regSize), tempReg, dstLclNum, dstOffset);
}
else
{
emit->emitIns_ARX_R(INS_movdqu, EA_ATTR(regSize), tempReg, dstAddrBaseReg, dstAddrIndexReg,
emit->emitIns_ARX_R(INS_movups, EA_ATTR(regSize), tempReg, dstAddrBaseReg, dstAddrIndexReg,
dstAddrIndexScale, dstOffset);
}
}
Expand Down
26 changes: 4 additions & 22 deletions src/coreclr/src/jit/emitxarch.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -14353,6 +14353,10 @@ emitter::insExecutionCharacteristics emitter::getInsExecutionCharacteristics(ins

case INS_movdqa:
case INS_movdqu:
case INS_movaps:
case INS_movups:
case INS_movapd:
case INS_movupd:
if (memAccessKind == PERFSCORE_MEMORY_NONE)
{
// ins reg, reg
Expand Down Expand Up @@ -14537,28 +14541,6 @@ emitter::insExecutionCharacteristics emitter::getInsExecutionCharacteristics(ins
result.insLatency += PERFSCORE_LATENCY_5C;
break;

case INS_movaps:
case INS_movups:
case INS_movapd:
case INS_movupd:
if (memAccessKind == PERFSCORE_MEMORY_NONE)
{
result.insThroughput = PERFSCORE_THROUGHPUT_4X;
result.insLatency = PERFSCORE_LATENCY_1C;
}
else if (memAccessKind == PERFSCORE_MEMORY_READ)
{
result.insThroughput = PERFSCORE_THROUGHPUT_2X;
result.insLatency += PERFSCORE_LATENCY_2C;
}
else
{
assert(memAccessKind == PERFSCORE_MEMORY_WRITE);
result.insThroughput = PERFSCORE_THROUGHPUT_1C;
result.insLatency += PERFSCORE_LATENCY_3C;
}
break;

case INS_paddb:
case INS_psubb:
case INS_paddw:
Expand Down