github/ROCm
1
1
Fork 0
mirror of https://github.com/ROCm/ROCm.git synced 2026-04-05 03:01:17 -04:00
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'triton-mlir' into ifu-231117

Browse Source
This commit is contained in:
jayfurmanek
2023-12-12 14:24:11 -06:00
committed by GitHub
parent 160dfe838e 605a90c58e
commit a42ac260aa
17 changed files with 2574 additions and 209 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
include/triton/Dialect/TritonGPU/IR/TritonGPUAttrDefs.td
View File

@@ -740,6 +740,24 @@ The data will be distributed between threads as follows:
[ 48 49 50 51 ...... 62 63 ] [ 112 113 114 115 ...... 126 127 ]
[ 48 49 50 51 ...... 62 63 ] [ 112 113 114 115 ...... 126 127 ]
[ 48 49 50 51 ...... 62 63 ] [ 112 113 114 115 ...... 126 127 ]
Example 3:
Suppose we have a tensor with a shape of [8, 8], warpsPerCTA set to [2, 2] and nonKDim set to 4.
The data will be distributed between threads as follows(note that each element is duploicated in 16 threads):
M N -> wave 0 wave 2
| --------------------------/\-------------------------- ------------------------------/\------------------------------
V [ 0,4,8...60 1,5...61 2,6...62 3,7...63 ] [ 128,132...188 129,133...189 130,134...190 131,135...191 ]
[ 0,4,8...60 1,5...61 2,6...62 3,7...63 ] [ 128,132...188 129,133...189 130,134...190 131,135...191 ]
[ 0,4,8...60 1,5...61 2,6...62 3,7...63 ] [ 128,132...188 129,133...189 130,134...190 131,135...191 ]
[ 0,4,8...60 1,5...61 2,6...62 3,7...63 ] [ 128,132...188 129,133...189 130,134...190 131,135...191 ]
wave 1 wave 3
--------------------------/\-------------------------- ------------------------------/\------------------------------
[ 64,68...124 65,69...125 66,70...126 67,71...127 ] [ 192,196...252 193,197...253 194,198...254 195,199...255 ]
[ 64,68...124 65,69...125 66,70...126 67,71...127 ] [ 192,196...252 193,197...253 194,198...254 195,199...255 ]
[ 64,68...124 65,69...125 66,70...126 67,71...127 ] [ 192,196...252 193,197...253 194,198...254 195,199...255 ]
[ 64,68...124 65,69...125 66,70...126 67,71...127 ] [ 192,196...252 193,197...253 194,198...254 195,199...255 ]
}];
let parameters = (