On 16/10/14 21:43, Andrew Pinski wrote:
On Thu, Oct 16, 2014 at 1:38 PM, Sebastian Pop <seb...@gmail.com> wrote:
Richard Biener wrote:
I have posted 5 patches as part of a larger series to merge
(parts) from the match-and-simplify branch. While I think
there was overall consensus that the idea behind the project
is sound there are technical questions left for how the
thing should look in the end. I've raised them in 3/n
which is the only patch of the series that contains any
patterns sofar.
To re-iterate here (as I expect most people will only look
at [0/n] patches ;)), the question is whether we are fine
with making fold-const (thus fold_{unary,binary,ternary})
not handle some cases it handles currently.
I have tested on aarch64 all the code in the match-and-simplify against trunk as
of the last merge at r216315:
2014-10-16 Richard Biener <rguent...@suse.de>
Merge from trunk r216235 through r216315.
Overall, I see a lot of perf regressions (about 2/3 of the tests) than
improvements (1/3 of the tests). I will try to reduce tests.
For instance, saxpy regresses at -O3 on aarch64:
void saxpy(double* x, double* y, double* z) {
int i=0;
for (i = 0 ; i < ARRAY_SIZE; i++) {
z[i] = x[i] + scalar*y[i];
}
}
This looks like a scheduling issue rather than anything else. The
scheduler for a57 is not complete and does not model some things like
the fusion of the compares and branch which is most likely what you
are seeing.
Huh !! how is that related to the code generation shown by Seb ?
See the replacement of subs by cmp and sub. Folding cmp into other flag
setting instructions is a very useful optimization on ARM and AArch64
and that's what appears missing in fold-const. That maybe what's causing
the slowdown. I've never known that to be caused by any scheduler
vagaries !
regards
Ramana
Thanks,
Andrew Pinski
$ diff -u base.s mas.s
--- base.s 2014-10-16 15:30:15.351430000 -0500
+++ mas.s 2014-10-16 15:30:16.183035000 -0500
@@ -2,12 +2,14 @@
add x1, x2, 800
ldr q0, [x0, x2]
add x3, x2, 1600
+ cmp x0, 784
ldr q1, [x0, x1]
+ add x1, x0, 16
fmla v0.2d, v1.2d, v2.2d
str q0, [x0, x3]
- add x0, x0, 16
- cmp x0, 800
+ mov x0, x1
bne .L140
.LBE179:
- subs w4, w4, #1
+ cmp w4, 1
+ sub w4, w4, #1
bne .L139
Thanks,
Sebastian
