Sjoerd Meijer
cda6f69edf
[AArch64] Cost-model i8 vector loads/stores ...
Loads of <4 x i8> vectors were modeled as extremely expensive. And while we
don't have a load instruction that supports this, it isn't that expensive to
create a vector of i8 elements. The codegen for this was fixed/optimised in
D105110. This now tweaks the cost model and enables SLP vectorisation of my
motivating case loadi8.ll.
Differential Revision: https://reviews.llvm.org/D103629
2021-07-05 11:25:10 +01:00
2021-01-19 20:09:42 +01:00
2021-02-11 12:17:38 -06:00
2021-06-30 20:04:39 +01:00
2020-12-23 22:47:36 +07:00
2021-05-24 19:43:40 +02:00
2021-04-08 11:14:09 -07:00
2020-06-26 20:55:44 -07:00
2021-07-05 11:25:10 +01:00
2021-01-07 19:57:14 -08:00
2021-06-23 10:21:05 +01:00
2021-06-02 10:07:40 -04:00
2021-05-24 19:43:40 +02:00
2021-02-16 10:26:45 +05:30
2019-03-22 18:27:13 +00:00
2020-09-15 11:59:00 -07:00
2020-07-23 12:46:47 -07:00
2020-11-09 00:04:52 -05:00
2021-02-09 12:11:47 -06:00
2020-10-25 19:43:58 +01:00
2020-11-02 13:22:42 -08:00
2020-11-17 23:15:20 -08:00
2021-02-16 10:26:45 +05:30
2020-11-20 20:08:43 -05:00
2021-06-22 11:27:51 +01:00
2020-02-10 13:22:35 -05:00
2020-09-22 11:31:00 -07:00
2021-05-07 16:04:18 +00:00
2021-05-24 19:43:40 +02:00
2021-05-24 19:43:40 +02:00
2020-11-03 21:06:46 -08:00
2018-06-28 14:13:06 +00:00
2020-09-15 11:19:01 -07:00
2018-11-02 17:39:31 +00:00
2020-09-15 11:12:14 -07:00
2021-06-30 08:31:13 -07:00
2020-12-03 09:23:37 -08:00
2021-02-19 08:29:12 -08:00
2021-05-24 19:43:40 +02:00
2021-05-03 16:40:56 -07:00
2020-09-22 17:54:25 -07:00
cda6f69edf