wangtao
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

[mlir][Linalg][Doc] Minor doc fixes

This commit is contained in:
Nicolas Vasilache 2020-04-01 13:41:45 -04:00
parent b8c260c38d
commit c4c20376f7
1 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
mlir/docs/Dialects/Linalg.md
View File

@ -29,7 +29,7 @@ performed on the Linalg IR and that have influenced its design:
1. Tiled Producer-Consumer Fusion with Parametric Tile-And-Fuse.
1. Map to Parallel and Reduction Loops and Hardware.
1. Vectorization: Rewrite in Vector Form.
1. Lower to Loops (Affine and/or Generic).
1. Lower to Loops (Affine, Generic and Parallel).
1. Lower to Library Calls or Special Instructions, Intrinsics or ISA.
1. Partially Lower to Iterations Over a Finer-Grained Linalg Op.
@ -241,7 +241,7 @@ example:
(i, j) -> (i, j),
(i, j) -> (i, j)
}
#attrs = {args_in: 1, args_out: 1, indexings: #indexing_maps}
#attrs = {args_in: 2, args_out: 1, indexings: #indexing_maps}
func @example(%A: memref<?x?xf32>, %B: memref<?x?xf32>, %C: memref<?x?xf32>) {
linalg.generic #attrs (%A, %B, %C) {
^bb0(%a: f32, %b: f32):
@ -295,7 +295,7 @@ example:
(i, j) -> (i, j),
(i, j) -> (i, j)
}
#attrs = {args_in: 1, args_out: 1, indexings: #indexing_maps, fun: #fun_attr}
#attrs = {args_in: 2, args_out: 1, indexings: #indexing_maps, fun: #fun_attr}
func @example(%A: memref<?x?xf32>, %B: memref<?x?xf32>, %C: memref<?x?xf32>) {
linalg.generic #attrs (%A, %B, %C) {
^bb0(%a: f32, %b: f32):