Commit Graph

64 Commits

Author SHA1 Message Date
oahzxl
80efd70c72 improve reorder efficeincy 2022-12-31 13:44:46 +08:00
oahzxl
966e4ea0cb add reorder in mem estimator 2022-12-31 02:20:07 +08:00
oahzxl
e5a5fbb8a9 update source add 2022-12-31 01:00:06 +08:00
oahzxl
f5515e9978 use max_mem to control stratge 2022-12-29 16:55:47 +08:00
oahzxl
1d7ca02301 add benchmark 2022-12-29 14:28:38 +08:00
oahzxl
cb2dd1a106 turn off print mem 2022-12-27 15:01:58 +08:00
oahzxl
a2b4755ce9 code style 2022-12-27 14:49:52 +08:00
oahzxl
6be89a3b82 add chunksize in emit, fix bug in reassgin shape 2022-12-27 14:48:25 +08:00
oahzxl
378a49dc6c code style 2022-12-27 09:48:59 +08:00
oahzxl
8f5a0edfab add chunk select 2022-12-26 23:08:49 +08:00
oahzxl
1b8a066592 add chunk select class 2022-12-26 15:28:01 +08:00
oahzxl
786a398a6b code style 2022-12-23 17:42:51 +08:00
oahzxl
51ef8384c1 finish node reorder 2022-12-23 17:25:36 +08:00
oahzxl
884a228ea6 reorder nodes 2022-12-23 17:06:07 +08:00
oahzxl
e0ae68e736 code style 2022-12-23 15:49:04 +08:00
oahzxl
fa5e6fbf96 code style 2022-12-23 15:38:37 +08:00
oahzxl
4f5e105af3 remove flow tracer 2022-12-23 15:34:41 +08:00
oahzxl
4d89525fc2 remove abandoned function 2022-12-23 14:28:49 +08:00
oahzxl
49ba619085 code style 2022-12-23 14:26:43 +08:00
oahzxl
d309e9338b adapt codegen to prepose node 2022-12-23 14:26:12 +08:00
oahzxl
522f017418 code style 2022-12-23 13:41:51 +08:00
oahzxl
774d34f1aa refactor flow search 2022-12-23 13:41:10 +08:00
oahzxl
ded1005667 format code 2022-12-21 15:03:08 +08:00
oahzxl
d361d533e8 refactor flow tracer 2022-12-21 15:01:03 +08:00
oahzxl
d734529a39 move flow tracer 2022-12-21 15:00:24 +08:00
oahzxl
9d516fa68f fix layernorm 2022-12-18 20:37:55 +08:00
oahzxl
e66a18a0bf optimise search 2022-12-16 15:06:39 +08:00
oahzxl
e83e3c6154 update memory estimate 2022-12-16 11:09:35 +08:00
oahzxl
de65e6c3e8 support output 2022-12-13 11:00:51 +08:00
oahzxl
cda3e8572a support index dupilictae and update loop 2022-12-13 10:02:26 +08:00
oahzxl
1e0fd11bc1 support check_index_duplicate 2022-12-13 10:01:30 +08:00
oahzxl
98f9728e29 code style 2022-12-12 18:15:47 +08:00
oahzxl
8511d900a8 code style 2022-12-12 17:36:17 +08:00
oahzxl
5cdfcfe1d1 code style 2022-12-12 17:29:07 +08:00
oahzxl
b7b67c32ad code style 2022-12-12 17:25:38 +08:00
oahzxl
31a2c5d09f work with outerproductmean and msa 2022-12-12 17:24:06 +08:00
oahzxl
5de9e46381 code format 2022-12-10 17:34:48 +08:00
oahzxl
d31e146687 code format 2022-12-10 17:34:40 +08:00
oahzxl
929445116a pass outproduct mean 2022-12-10 17:29:51 +08:00
oahzxl
979e61db92 redesign index tracer, add source and change compute 2022-12-09 17:39:02 +08:00
oahzxl
2b4ebcc278 finishi codegen on msa 2022-12-08 15:16:10 +08:00
oahzxl
6d99994a7a rename index tracer 2022-12-06 17:35:27 +08:00
oahzxl
a9d64377bb support new op 2022-12-06 17:34:24 +08:00
oahzxl
f24c418bb0 finish chunk define 2022-12-06 16:29:07 +08:00
oahzxl
3b7d671206 finish region search loop 2022-12-06 11:08:39 +08:00
oahzxl
7330d90745 add possible region search 2022-12-04 17:05:28 +08:00
oahzxl
d9ca2f898d polish code 2022-11-15 15:50:50 +08:00
oahzxl
54a34a7e46 update active log 2022-11-15 11:30:43 +08:00
oahzxl
fad3b6d1a6 polish code 2022-11-15 10:46:51 +08:00
oahzxl
7e2bd1e428 polish code 2022-11-15 10:36:02 +08:00