<div dir="ltr"><br><br><div class="gmail_quote"><div dir="ltr">On Mon, Jan 11, 2016 at 11:23 AM Sarah Sharp <<a href="mailto:sarah.a.sharp@linux.intel.com">sarah.a.sharp@linux.intel.com</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">On Sat, Jan 09, 2016 at 11:57:38AM -0800, Kristian Høgsberg Kristensen wrote:<br>
> Mark Janes <<a href="mailto:mark.a.janes@intel.com" target="_blank">mark.a.janes@intel.com</a>> writes:<br>
><br>
> > Sarah Sharp <<a href="mailto:sarah.a.sharp@intel.com" target="_blank">sarah.a.sharp@intel.com</a>> writes:<br>
> ><br>
> >> On Fri, Jan 08, 2016 at 04:30:20PM -0800, Mark Janes wrote:<br>
> >>> In testing KBL, I found:<br>
> >>><br>
> >>> - urb size was not set for slices gt1.5, gt2, and gt3. The value I<br>
> >>> used for these slices (384) was taken from an earlier patch authored<br>
> >>> by Ben Widawsky.<br>
> >>><br>
> >>> - slice count was missing. This field was added by<br>
> >>> a403ad4f5a034e52a3cd845e91c4aa3e6927b731<br>
> >>><br>
> >>> With this commit, KBL passes piglit at parity with SKL.<br>
> >><br>
> >> Thanks for catching this Mark!<br>
> >><br>
> >> Looking at the docs, I'm not sure the URB size is correct for GT4. The<br>
> >> Device Attributes page shows the GT4 URB size should be 1152K (divided<br>
> >> by the number of slices, that's 384K). However, another table says to<br>
> >> limit the URB size to 1008K for "SKL+" and it doesn't say to exclude<br>
> >> KBL. Yay conflicting documentation!<br>
> ><br>
> > Unfortunately, Ben doesn't get back until Wednesday.<br>
> ><br>
> > Since this patch doesn't modify the GT4 URB size, does it make sense to<br>
> > use it to enable the other skus? GT4 can be addressed later.<br>
><br>
> The patch looks good to me, though urb.size is already set in the<br>
> GEN9_FEATURES macro:<br>
><br>
> ...<br>
> .urb = { \<br>
> .size = 384, \<br>
> ...<br>
><br>
> which looks correct for all SKUs, except GT1 and GT4, which both<br>
> override it to the correct value. As for GT4 urb size, it wont matter<br>
> for a while, but 1088 / 3 is the safe choice.<br>
><br>
> With the .urb.size assignments removed, this patch is<br>
><br>
> Reviewed-by: Kristian Høgsberg Kristensen <<a href="mailto:kristian.h.kristensen@intel.com" target="_blank">kristian.h.kristensen@intel.com</a>><br>
<br>
Thanks for the review, Kristian! I've tested the modified patch (without<br>
the urb.size) on KBL, and it works fine. I'll go push this and mark it<br>
for stable.<br></blockquote><div><br></div><div>oh! my bad. I mislead Mark on the urb size while he caught the real offending part that was num_slices... </div><div><br></div><div>apparently num_slices didn't exist on kbl-po branch so this is why we didn't see this issue earlier... <br></div><div><br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<br>
Sarah Sharp<br>
_______________________________________________<br>
mesa-dev mailing list<br>
<a href="mailto:mesa-dev@lists.freedesktop.org" target="_blank">mesa-dev@lists.freedesktop.org</a><br>
<a href="http://lists.freedesktop.org/mailman/listinfo/mesa-dev" rel="noreferrer" target="_blank">http://lists.freedesktop.org/mailman/listinfo/mesa-dev</a><br>
</blockquote></div></div>