Sanjay Patel 99a5482014 [x86] make SLM extract vector element more expensive than default ... I'm not sure what the effect of this change will be on all of the affected tests or a larger benchmark, but it fixes the horizontal add/sub problems noted here: https://reviews.llvm.org/D59710?vs=227972&id=228095&whitespace=ignore-most#toc The costs are based on reciprocal throughput numbers in Agner's tables for PEXTR*; these appear to be very slow ops on Silvermont. This is a small step towards the larger motivation discussed in PR43605: https://bugs.llvm.org/show_bug.cgi?id=43605 Also, it seems likely that insert/extract is the source of perf regressions on other CPUs (up to 30%) that were cited as part of the reason to revert D59710, so maybe we'll extend the table-based approach to other subtargets. Differential Revision: https://reviews.llvm.org/D70607		2019-11-27 14:08:56 -05:00
..
AArch64	[CostModel] Model all extractvalues as free.	2019-08-29 11:50:30 +00:00
AMDGPU	AMDGPU: Split test functions to avoid dependency on subtarget	2019-11-19 11:12:13 +05:30
ARM
PowerPC	Recommit [PowerPC] Update P9 vector costs for insert/extract	2019-08-26 19:26:08 +00:00
RISCV
SystemZ	[SystemZ] Support z15 processor name	2019-09-20 23:04:45 +00:00
X86	[x86] make SLM extract vector element more expensive than default	2019-11-27 14:08:56 -05:00
no_info.ll