On Tuesday, September 30, 2014 11:37:37 AM Matt Turner wrote: > On Tue, Sep 30, 2014 at 5:22 AM, Samuel Iglesias Gonsálvez > <sigles...@igalia.com> wrote: > > Hello, > > > > I am looking at bug 57439 [0] where it shows an error > > in a piglit test [1] related to textureSize() function happening > > in Intel SandyBridge hardware. > > > > According to SNB's PRM documentation (vol4 part1 page 141), the > > returned value for SURFTYPE_BUFFER (the surface type used in the test) > > is not defined in the 'resinfo' message type. For IvyBridge's doc it is > > defined as the buffer size, which is calculated from combined > > Depth/Height/Width values. > > > > As it is not clear that SNB returns the same value than IVB for that > > kind of message and surface type, I send this email here asking for a > > clarification :-) > > Yes, I can confirm that the internal BSpec says on Sandybridge resinfo > for SURFTYPE_BUFFER (and SURFTYPE_STRBUF, same thing?) returns > undefined results in all channels.
SURFTYPE_STRBUF is "Structured Buffer" - it's a bit different. They have a format, and you access them using the Typed read/write messages. --Ken
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
