[Beignet] *** SPAM LEVEL 4.053 *** compiler_local_memory_barrier_wg64 fails in current HEAD

Zhigang Gong zhigang.gong at linux.intel.com
Sun Jun 23 19:35:26 PDT 2013


Thanks for the testing. Then this seems a known issue and we don't know the
root cause currently.
Maybe we need to add a fence to the local memory barrier as a temporary work
around. Any comments?

> -----Original Message-----
> From: beignet-bounces+zhigang.gong=linux.intel.com at lists.freedesktop.org
>
[mailto:beignet-bounces+zhigang.gong=linux.intel.com at lists.freedesktop.org]
> On Behalf Of Xing, Homer
> Sent: Monday, June 24, 2013 9:46 AM
> To: beignet at lists.freedesktop.org
> Subject: Re: [Beignet] *** SPAM LEVEL 4.053 ***
> compiler_local_memory_barrier_wg64 fails in current HEAD
> 
> I can reproduce this.
> I ran this case twenty-five times, and got two times of failure,
twenty-three
> times of success.
> After changing to CLK_GLOBAL_MEM_FENCE, I ran this case fifty times, and
> fifty times of success.
> 
> Homer
> 
> -----Original Message-----
> From: beignet-bounces+homer.xing=intel.com at lists.freedesktop.org
> [mailto:beignet-bounces+homer.xing=intel.com at lists.freedesktop.org] On
> Behalf Of Zhigang Gong
> Sent: Monday, June 24, 2013 9:35 AM
> To: 'Simon Richter'; beignet at lists.freedesktop.org
> Subject: Re: [Beignet] *** SPAM LEVEL 4.053 ***
> compiler_local_memory_barrier_wg64 fails in current HEAD
> 
> I can't reproduce this on my machine. But I do remember it really happened
> one or two times about one month ago.
> Does it reproduce easily on your machine? If it is, could you try to
change the
> local memory barrier to global memory Barrier and see what's the
difference.
> 
> I mean change the CLK_LOCAL_MEM_FENCE to CLK_GLOBAL_MEM_FENCE at
> kernels/compiler_local_memory_barrier_wg64.cl
> 
> Thanks.
> 
> > -----Original Message-----
> > From:
> > beignet-bounces+zhigang.gong=linux.intel.com at lists.freedesktop.org
> >
>
[mailto:beignet-bounces+zhigang.gong=linux.intel.com at lists.freedesktop.org]
> > On Behalf Of Simon Richter
> > Sent: Friday, June 21, 2013 6:06 PM
> > To: beignet at lists.freedesktop.org
> > Subject: [Beignet] *** SPAM LEVEL 4.053 ***
> > compiler_local_memory_barrier_wg64 fails in current HEAD
> >
> > -----BEGIN PGP SIGNED MESSAGE-----
> > Hash: SHA1
> >
> > Hi,
> >
> > I'm getting a testcase failure on IVB2 in 64 bit mode:
> >
> >   compiler_local_memory_barrier_wg64()    [FAILED]
> >     Error: dst[i+j] == 63-j
> >
> > Is that the barrier issue I've seen mentioned earlier, or something
else?
> >
> >    Simon
> > -----BEGIN PGP SIGNATURE-----
> > Version: GnuPG v1.4.12 (GNU/Linux)
> > Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
> >
> > iJwEAQECAAYFAlHEJXYACgkQ0sfeulffv7ukAgQAso6OLcld2kLiRh0MNoTw6v/y
> >
> hV/gxk5d03jO46dxDZLUe6Nl6O5KIyao6cLgzv5jAuxu1BoJx3Nk+5K1/MQhOMyQ
> >
> 46H8hAwZ2MOnZmmBPVgG65WH9BaWNtBse01OtyCiA71xuetb9L9rz7KkQBip
> > XyaV
> > IPCzZ5zmxMrXY/yMp6A=
> > =Qppu
> > -----END PGP SIGNATURE-----
> > _______________________________________________
> > Beignet mailing list
> > Beignet at lists.freedesktop.org
> > http://lists.freedesktop.org/mailman/listinfo/beignet
> 
> _______________________________________________
> Beignet mailing list
> Beignet at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/beignet
> _______________________________________________
> Beignet mailing list
> Beignet at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/beignet



More information about the Beignet mailing list