ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-14 18:17:44 +00:00
Code Issues Packages Projects Releases Wiki Activity
365 Commits 743 Branches 38 Tags
7c9100b53f55e5403015897bb526ec85bb7117be
Commit Graph

66 Commits

Author SHA1 Message Date
Chao Liu
7c9100b53f Added bwd data v3r1 v4r1, tweaking v1 (#10) 
* Added bwd data v3r1: breaking down compute into a series of load balanced GEMM, and launch in a single kernel
* Added bwd data v4r1: like v3r1, but launch GEMMs in multiple kernels
* Tweaked v1r1  and v1r2 (atomic) on AMD GPU

[ROCm/composable_kernel commit: c5da0377fb]
 2020-01-20 10:20:03 -06:00
Chao Liu
24f7d66609 update implicit GEMM forward v4r4 to use gridwise gemm (#9) 
* updated fwd v4r4 to use gridwise gemm
* updated gridwise gemm api calls in bwd-data v1r1 and v2r1

[ROCm/composable_kernel commit: e2b4c5b469]
 2019-12-05 12:36:36 -06:00
Chao Liu
4414e495ed backward data (#7) 
* enabled atomic add in tensor copy
* added gridwise GEMM
* added backward data conv using GEMM + atomic
* added backward data conv using GEMM, no atomic


[ROCm/composable_kernel commit: 8f5f64960e]
 2019-12-03 01:16:12 -06:00
Chao Liu
a9e6c3340c Refactor for MIOpen integration (#4) 
Refactor, so can bring multi-index transformation and padding support into MIOpen

[ROCm/composable_kernel commit: 52c3fe05be]
 2019-10-11 11:37:31 -05:00
Chao Liu
a87fa81015 remove dead code 
[ROCm/composable_kernel commit: 9b280cc50d]
 2019-09-27 02:00:59 -05:00
Chao Liu
c5e4a623f9 clean up 
[ROCm/composable_kernel commit: b12bbceebc]
 2019-09-26 14:59:19 -05:00
Chao Liu
bd8f263b70 removing dependency on old tensor descriptor 
[ROCm/composable_kernel commit: 51a9fa1dbd]
 2019-09-26 11:49:05 -05:00
Chao Liu
313eb881cd removing old implementation of tensor descriptor 
[ROCm/composable_kernel commit: 39d92e7dfd]
 2019-09-25 22:24:06 -05:00
Chao Liu
ca661e1f52 refactor 
[ROCm/composable_kernel commit: 545d930568]
 2019-09-24 18:06:05 -05:00
Chao Liu
c3a1be3865 WIP: explicitly separate offset component into compile-time, block-invariant and per-thread components 
[ROCm/composable_kernel commit: 51884fc214]
 2019-09-21 22:53:03 -05:00
Chao Liu
b9722daae4 refactor 
[ROCm/composable_kernel commit: bf7e7d62a8]
 2019-09-19 23:44:23 -05:00
Chao Liu
056e48a3ac use buffer_load buffer_store intrinsic 
[ROCm/composable_kernel commit: b6e1c52a80]
 2019-09-19 15:39:07 -05:00
Chao Liu
67f93cb50c add global_load and buffer_load inline asm 
[ROCm/composable_kernel commit: 86cc678f18]
 2019-09-18 15:41:55 -05:00
Chao Liu
6c5f82174b experimenting global and buffer load/store 
[ROCm/composable_kernel commit: 5b7a18c506]
 2019-09-18 02:05:42 -05:00
Chao Liu
741a647405 experimenting global and buffer load/store 
[ROCm/composable_kernel commit: c7a6545ec4]
 2019-09-18 01:37:28 -05:00
Chao Liu
03740b5c4a experimenting global and buffer load/store 
[ROCm/composable_kernel commit: 9f46cdf5fa]
 2019-09-18 00:15:57 -05:00
Chao Liu
0342a42f47 enable hip compiler flag: -amdgpu-enable-global-sgpr-addr 
[ROCm/composable_kernel commit: f58bf38445]
 2019-09-17 17:34:39 -05:00
Chao Liu
2b26fb76ca refactor 
[ROCm/composable_kernel commit: f7be86b9e4]
 2019-09-16 22:47:55 -05:00
Chao Liu
a4edaf2ae3 add lds doble buffer to nchw padded v4r1 and v4r4 
[ROCm/composable_kernel commit: bf97542846]
 2019-09-15 16:58:16 -05:00
Chao Liu
e6285a3b55 initial implementation for nchw v4r4 padding 
[ROCm/composable_kernel commit: 2c93b3057d]
 2019-09-15 16:31:54 -05:00
Chao Liu
89dc09f929 clean up 
[ROCm/composable_kernel commit: 53094f7fae]
 2019-09-15 12:13:58 -05:00
Chao Liu
79e13d9ded clean up 
[ROCm/composable_kernel commit: bd7a230006]
 2019-09-12 14:55:46 -05:00
Chao Liu
9a8d9e0e40 enabling padding for chwn format 
[ROCm/composable_kernel commit: 724e984bff]
 2019-09-11 01:13:13 -05:00
Chao Liu
399be319a2 more utility code 
[ROCm/composable_kernel commit: 7a7fe16086]
 2019-09-09 00:29:33 -05:00
Chao Liu
b0f3708397 added tuple 
[ROCm/composable_kernel commit: 625838def0]
 2019-09-06 18:07:56 -05:00
Chao Liu
eb6a36d393 Merge remote-tracking branch 'origin/master' into add_padding 
[ROCm/composable_kernel commit: 86ceded98b]
 2019-08-15 13:48:45 -05:00
Chao Liu
c8e088723f clean up 
[ROCm/composable_kernel commit: 0979fb4af9]
 2019-08-15 13:21:51 -05:00
Chao Liu
3b234309c9 adding padding to implicit gemm v1r3 
[ROCm/composable_kernel commit: 4fb81e008c]
 2019-08-14 10:55:34 -05:00
Chao Liu
15a2fc9029 add back some code 
[ROCm/composable_kernel commit: 40836ab926]
 2019-08-13 12:21:38 -05:00
Chao Liu
18bc57cd93 clean up 
[ROCm/composable_kernel commit: 8bdaba51f8]
 2019-08-13 00:37:23 -05:00
Chao Liu
d54e146f28 clean up 
[ROCm/composable_kernel commit: fab2f10a55]
 2019-08-12 15:48:35 -05:00
Chao Liu
301348f208 cleaning up 
[ROCm/composable_kernel commit: 1c4ef23cff]
 2019-08-09 22:48:28 -05:00
Chao Liu
807d22de0b tweak on amd 
[ROCm/composable_kernel commit: 4908fe3fdc]
 2019-08-08 12:14:06 -05:00
Chao Liu
7b5f9bebc6 added ThreadwiseGenericTensorSliceCopy_v2r1 
[ROCm/composable_kernel commit: a9b2b1dcd7]
 2019-08-08 02:42:52 -05:00
Chao Liu
139b043d07 use ford/for instead of static_ford/static_for in threadwise copy, somehow register spill is greatly reduced on AMD 
[ROCm/composable_kernel commit: bc9ea646f8]
 2019-08-07 19:09:13 -05:00
Chao Liu
c9a4be4b22 bug fix in ford, forgot to reorder lengths 
[ROCm/composable_kernel commit: 5636576f9b]
 2019-08-07 18:27:10 -05:00
Chao Liu
5716cecadc adding ThreadwiseGenericTensorSliceCopy_v1r2 
[ROCm/composable_kernel commit: 9d99a58072]
 2019-08-07 16:51:14 -05:00
Chao Liu
8994adaca4 reworked ThreadwiseGenericTensorSliceCopy_v1 
[ROCm/composable_kernel commit: 1b3c2e4035]
 2019-08-07 00:52:13 -05:00
Chao Liu
fb1953ed86 reimplement threadwise copy 
[ROCm/composable_kernel commit: fdcfae3a62]
 2019-08-06 17:41:58 -05:00
Chao Liu
0319e69e81 tweak 
[ROCm/composable_kernel commit: adc1008836]
 2019-08-03 15:05:25 -05:00
Chao Liu
be8343bb7e added implicit gemm v4r4 and double buffer 
[ROCm/composable_kernel commit: c2d246696f]
 2019-08-03 00:19:19 -05:00
Chao Liu
d0be17cfdf added new tensor copy operator 
[ROCm/composable_kernel commit: c01af89928]
 2019-08-03 00:02:24 -05:00
Chao Liu
1a459a3e63 experimenting TensorCoordinate and new merged tensor copy operator 
[ROCm/composable_kernel commit: a9a392b44d]
 2019-08-01 15:32:40 -05:00
Chao Liu
2c5d1779ff refactor 
[ROCm/composable_kernel commit: 2eeeb1766b]
 2019-07-30 22:50:51 -05:00
Chao Liu
9cb7d14a8a added (1x4)x(2x4) threadwise gemm 
[ROCm/composable_kernel commit: 08cbac98cc]
 2019-07-30 18:20:55 -05:00
Chao Liu
1b43c17bb1 retune implicit gemm v4r1 
[ROCm/composable_kernel commit: c5e5a9307b]
 2019-07-30 12:10:28 -05:00
Chao Liu
472db1ed5a experimenting new merged tensor copy 
[ROCm/composable_kernel commit: cd8de11218]
 2019-07-30 09:35:54 -05:00
Chao Liu
dc2cdacd15 refactored implicit gemm v1r3 
[ROCm/composable_kernel commit: 284e7bb317]
 2019-07-29 15:25:38 -05:00
Chao Liu
360f15cf24 refactored implicit gemm v1r3 
[ROCm/composable_kernel commit: efd419ecbe]
 2019-07-29 15:01:01 -05:00
Chao Liu
cfbbb3a9e4 adding implicit gemm v4r4 
[ROCm/composable_kernel commit: 9ba3b49158]
 2019-07-28 19:39:57 -05:00