Anton Korobeynikov 0f885eb7fd Materialize global addresses via movt/movw pair, this is always better ... than doing the same via constpool: 1. Load from constpool costs 3 cycles on A9, movt/movw pair - just 2. 2. Load from constpool might stall up to 300 cycles due to cache miss. 3. Movt/movw does not use load/store unit. 4. Less constpool entries => better compiler performance. This is only enabled on ELF systems, since darwin does not have needed relocations (yet). llvm-svn: 89720		2009-11-24 00:44:37 +00:00
..
Alpha
ARM	Materialize global addresses via movt/movw pair, this is always better	2009-11-24 00:44:37 +00:00
Blackfin
CBackend
CellSPU	Revert the main portion of r31856. It was causing BranchFolding	2009-10-22 00:03:58 +00:00
CPP	fix PR5295 where the .ll parser didn't reject a function after a global	2009-10-25 23:22:50 +00:00
Generic	Convert test to FileCheck.	2009-11-22 13:16:36 +00:00
Mips	Unbreak test, Bruno please check.	2009-11-19 07:18:49 +00:00
MSP430	Add and-not (bic) patterns. Based heavily on patch by Brian Lucas!	2009-11-08 15:33:12 +00:00
PIC16	Convert a few tests to FileCheck for PR5307.	2009-11-22 11:45:44 +00:00
PowerPC	Fix for bad FileCheck converts in revision 89584.	2009-11-22 12:50:05 +00:00
SPARC
SystemZ
Thumb	More consistent thumb1 asm printing.	2009-11-19 06:57:41 +00:00
Thumb2	move fconst[sd] to UAL. <rdar://7414913>	2009-11-23 21:08:25 +00:00
X86	Turns out stuff gets allocated to different registers depending on the subtarget.	2009-11-22 15:15:52 +00:00
XCore	Add XCore support for indirectbr / blockaddress.	2009-11-18 23:20:42 +00:00