llvm-6502/test/CodeGen
Chandler Carruth 1d02acb7a0 [x86] Remove the 2-addr-to-3-addr "optimization" from shufps to pshufd.
This trades a (register-renamer-friendly) movaps for a floating point
/ integer domain cross. That is a very bad trade, even on architectures
where domain crossing is relatively fast. On any chip where there is
even a cycle stall, this is a Very Bad Idea. It doesn't even seem likely
to cause a spill to be introduced because the reason for the copy is to
destructively shuffle in place.

Thanks to Ben Kramer for fixing a bug in this code that my new shuffle
lowering exposed and highlighting that perhaps it should just go away.
=]

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@219090 91177308-0d34-0410-b5e6-96231b3b80d8
2014-10-05 22:57:31 +00:00
..
AArch64 Revert "Revert "DI: Fold constant arguments into a single MDString"" 2014-10-03 20:01:09 +00:00
ARM Revert "Revert "DI: Fold constant arguments into a single MDString"" 2014-10-03 20:01:09 +00:00
CPP
Generic Revert "Revert "DI: Fold constant arguments into a single MDString"" 2014-10-03 20:01:09 +00:00
Hexagon Revert "Revert "DI: Fold constant arguments into a single MDString"" 2014-10-03 20:01:09 +00:00
Inputs Revert "Revert "DI: Fold constant arguments into a single MDString"" 2014-10-03 20:01:09 +00:00
Mips Add fptrunc to mips fast-sel 2014-10-01 18:47:02 +00:00
MSP430
NVPTX Revert r216862 due to a performance regression 2014-10-01 15:22:13 +00:00
PowerPC Revert "Revert "DI: Fold constant arguments into a single MDString"" 2014-10-03 20:01:09 +00:00
R600 R600/SI: Custom lower f64 -> i64 conversions 2014-10-03 23:54:56 +00:00
SPARC
SystemZ
Thumb Revert "Revert "DI: Fold constant arguments into a single MDString"" 2014-10-03 20:01:09 +00:00
Thumb2
X86 [x86] Remove the 2-addr-to-3-addr "optimization" from shufps to pshufd. 2014-10-05 22:57:31 +00:00
XCore Revert "Revert "DI: Fold constant arguments into a single MDString"" 2014-10-03 20:01:09 +00:00