Skip to content
Snippets Groups Projects
Commit 7a03a821 authored by Russel Arbore's avatar Russel Arbore
Browse files

some opts

parent 2727439c
No related branches found
No related tags found
No related merge requests found
Pipeline #202690 failed
......@@ -6,6 +6,7 @@ macro simpl!(X) {
phi-elim(X);
crc(X);
slf(X);
const-inline[false](X);
dce(X);
infer-schedules(X);
}
......@@ -22,9 +23,25 @@ reduce-slf(*);
simpl!(*);
if feature("cuda") {
// Base
let func = outline(matmul@outer);
const-inline[false](matmul, func);
simpl!(matmul, func);
no-memset(matmul@res);
gpu(func);
// Opt #1
let outer = func@outer \ func@inner;
fork-tile[32, 1, false, true](outer);
fork-tile[32, 0, false, true](outer);
let (outer, inner) = fork-reshape[[0, 2], [1, 3]](outer);
// Opt #2
fork-tile[32, 0, false, true](func@inner);
fork-split(func@inner);
clean-monoid-reduces(func);
simpl!(func);
xdot[true](*);
} else {
unforkify(*);
fork-split(*);
......
0% Loading or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment