Matt Turner <matts...@gmail.com> writes: > On Tue, Aug 8, 2017 at 4:25 PM, Scott D Phillips > <scott.d.phill...@intel.com> wrote: >>> + [BRW_HW_IMM_TYPE_UV] = 2, >>> + [BRW_HW_IMM_TYPE_VF] = 4, >>> + [BRW_HW_IMM_TYPE_V] = 2, >> >> Is this right? I see it was there before, and perhaps I'm being dense, >> but it seems like V and UV should be size 4 from the PRM. > > Yes. The encoded immediates themselves are 4 bytes, but this table > captures the size of the individual components once expanded. That's > admittedly a little weird. > > A V/UV immediate consists of 8x 4-bit integer values. A restriction > documented in "Gen Graphics » BSpec » 3D and Compute Engine » 3D and > GPGPU Programs » EU Overview » Registers and Register Regions » > Immediate" states "When an immediate vector is used in an instruction, > the destination must be 128-bit aligned with destination horizontal > stride equivalent to a word for an immediate integer vector (v) and > equivalent to a DWord for an immediate float vector (vf).". > > So we consider the individual components of the V/UV immediate to > really be words, of size 2. > > Thanks for the good question!
Ah, I see. Thanks for the explanation. Reviewed-by: Scott D Phillips <scott.d.phill...@intel.com> _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev
