Rob Jenkins (robj++at++quid.csd.sgi.com)
Thu, 4 Jun 1998 13:58:04 -0700
I've been meaning to respond to this thread earlier because Uwe's mention of
the problem getting worse made me think of the problem Ran is now describing. I
saw a case a while back where on an O2, polygons flashed in a random fashion
when
an NFS filesysyem was being concurrently accessed.
This was SGI bug 560431 but
turned out to be a duplicate of 'BUG 545276 : crime empty wait timeouts, fifo
not draining'. The problem seems to have been one of handshaking between
kernel, the crime chip and the gfx pipe, sometimes the pipe gets locked when it
shouldn't. Both the bugs I mentioned were given priority 1 ( highest ) and
545276 is fixed in Irix 6.5. I don't think the fix has been put in any Irix 6.3
patches and with Irix 6.5 being so close to shipping then that might be OK.
*Ran* you might be referring to a different problem, let me know if you think
so.
Without being able to confirm my guess then I would suggest that anyone with a
problem that sounds anything like Ran describes: "the O2 drops parts of
polygons (mostly clipped polygons), when the machine is stressed." should log a
support call and have the support group run your test case on an O2 with Irix
6.5 installed ( or try it yourself if your a Beta test site ). If the problem
goes away then it might well have been bug 545276, if not then it's new and the
support call should take care of it - if the problem doesn't get the priority
that you think it should then let support know that, they have mechanisms for
setting priority of stuff, if it impacts your business and/or SGI sales or
customer sat. then SGI obviously wants to take care of it.
Cheers
Rob
On Jun 4, 9:40pm, Ran Yakir wrote:
> Subject: Re: Triangle Troubles
> This problem is an O2 bug (don't blame Uwe Pachler). I've had this problem a
long
> time ago where the O2 drops parts of polygons (mostly clipped polygons), when
the
> machine is stressed.
> Yes, I've managed to write a test case. The way I could simulate the stress
was by
> causing many memory swaps and disk accesses while drawing one big polygon.
The
> answer from SGI was something along these lines : " If the user has dropped
> polygons while the disk is swapping, then the user should avoid swapping when
he
> is drawing". This bug has a number (which I don't remember right now but I
can
> reproduce), and it was given a very low priority at SGI, which means it will
never
> be fixed.
> I think this is quite surprising that a company that made its reputation on
> drawing polygons should ignore a bug that causes polygons to be drawn
incorrectly.
>
> Ran
>
> Rob Jenkins wrote:
>
> > On Jun 3, 4:20pm, Uwe Pachler wrote:
> > > Subject: Triangle Troubles
> > > Hello!
> > >
> > > We run an O2 with CRM graphics and a R10000 with Performer 2.0 and IRIX
> > > 6.3. We run a performer application that simulates an industry
environment
> > > where the user is able to put on helmets and gas masks.
> > > The problem is that performer sometimes doesn't draw the whole area of a
> > > triangle, if the triangle seems to be bigger than the screen (maybe a
> > > clipping problem?)
> > > For example, if the user wears a gasmask, the mask, consisting of
> > > transparent polygons, covers the whole screen. The scene behind it is
> > > visible of course because the mask is drawn transparently. Sometimes,
> > > parts of the mask are not drawn. But not a whole triangle is missing, but
> > > only triangular parts of a triangle, where one side of that missing
> > > triangle is the screen's border (that's why I assume it is clipping bug
in
> > > the graphics library).
> > > Has onybody heard before from this bug, or has even fixed it?
> > >
> > > Thanks in advance,
> > > Uwe Pachler
> > >
> > > =======================================================================
> > > List Archives, FAQ, FTP: http://www.sgi.com/Technology/Performer/
> > > Submissions: info-performer++at++sgi.com
> > > Admin. requests: info-performer-request++at++sgi.com
> > >-- End of excerpt from Uwe Pachler
> >
> > If Svend's suggestion doesn't help then there are a couple of ( mostly
> > old/fixed ) bugs that might be related to this. It sounds to me like you're
> > describing triangular sections dropping in/out rather than something where
the
> > problem is constant. Without a test case it would be hard to narrow down
what
> > problem you have, if it is a known thing so I suggest that you first make
sure
> > you have the latest O2 gfx patch ( actually, get the latest patchset from
> > www.sgi.com/support ), any recent fixes will be in there and I also
remember
> > that there was an O2 gfx patch that was found to be bad and got yanked,
some
> > people might have it though. That was patch 1604. the latest O2 gfx patch
is:
> > 2943 although I'm not sure that is in a patchset yet, the latest in a
patchset
> > is probably patch 2862 and you should get at least that. If the patches and
> > examination of your viewing frustum setup don't help then you ought to log
a
> > support call, ideally with some testcase they can use.
> >
> > Cheers
> > Rob
> >
> > --
> > ________________________________________________________________
> > Rob Jenkins mailto:robj++at++sgi.com
> > Silicon Graphics, Mtn View, California, USA
> > =======================================================================
> > List Archives, FAQ, FTP: http://www.sgi.com/Technology/Performer/
> > Submissions: info-performer++at++sgi.com
> > Admin. requests: info-performer-request++at++sgi.com
>
>
>
> --
> __ | Ran Yakir
> /_) _ __ \ / _ / o __ | 28 Ben Gurion St.
> / )_ (_(_) ) \/ (_(_/<_(_)( | Hod Hasharon 54200
> _/ | Israel
> -------------------------------------+--------------------------------
> At Home : | At Work :
> | RT-SET
> Voice : +972-9-7489974 | Voice : +972-9-9552236
> Fax : +972-9-7422149 | Fax : +972-9-9552239
> E-mail : rany++at++netvision.net.il | E-mail : rany++at++rtset.co.il
> http://rtset.co.il/rany
>
>-- End of excerpt from Ran Yakir
--
________________________________________________________________
Rob Jenkins mailto:robj++at++sgi.com
Silicon Graphics, Mtn View, California, USA
=======================================================================
List Archives, FAQ, FTP: http://www.sgi.com/Technology/Performer/
Submissions: info-performer++at++sgi.com
Admin. requests: info-performer-request++at++sgi.com
This archive was generated by hypermail 2.0b2 on Mon Aug 10 1998 - 17:57:30 PDT