https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123870

--- Comment #29 from Robin Dapp <rdapp at gcc dot gnu.org> ---
(In reply to Ilya Kurdyukov from comment #27)
> > and my results are different
> 
> Your results are no different, all the hashes are the same as mine.
> 
> This is the correct result:
> 
> rdapp@pioneer-1:~/work/pr123870$ ./run.sh -DRVV_PATH1=1 -DRVV_PATH2=1 -O1
> quantsmooth: 1.119ms
> MD5(output.jpg)= 3bd354cc91258b8c7f814cadf85437ef
> -rw-rw-r-- 1 rdapp rdapp 4098 Feb  3 00:51 output.jpg
> 
> And this is the wrong result:
> 
> rdapp@pioneer-1:~/work/pr123870$ ./run.sh -DRVV_PATH1=1 -DRVV_PATH2=1 -O1
> -fschedule-insns
> quantsmooth: 0.935ms
> MD5(output.jpg)= faccc689fe1910dfd364f83dbf2649df
> -rw-rw-r-- 1 rdapp rdapp 4191 Feb  3 00:53 output.jpg
> 
> -DRVV_PATH1=1 -DRVV_PATH2=1 -O1 : correct
> -DRVV_PATH1=1 -DRVV_PATH2=1 -O2 : wrong
> -DRVV_PATH1=1 -DRVV_PATH2=1 -O3 : wrong
> -DRVV_PATH1=1 -DRVV_PATH2=1 -O1 -fschedule-insns : wrong
> ...
> -DRVV_PATH1=1 -DRVV_PATH2=0 -O2 : should be correct
> -DRVV_PATH1=1 -DRVV_PATH2=1 -O2 -fno-schedule-insns : should be correct
> -DRVV_PATH1=1 -DRVV_PATH2=1 -O2 --param=vsetvl-strategy=simple : should be
> correct

Yes, I misunderstood.  Having a look now.

Reply via email to