<div dir="ltr"><div class="gmail_extra"><div class="gmail_quote">On Mon, May 23, 2016 at 10:41 AM, Nanley Chery <span dir="ltr"><<a href="mailto:nanleychery@gmail.com" target="_blank">nanleychery@gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><span class="">On Fri, May 20, 2016 at 06:12:34PM -0700, Jason Ekstrand wrote:<br>
> This little series effectively moves the surface format table from<br>
> brw_surface_formats.c into ISL. Previously, it got built into<br>
> <a href="http://libi965_compiler.la" rel="noreferrer" target="_blank">libi965_compiler.la</a> because we needed to share it between drivers and<br>
> didn't have a better place to put it. Now it can live in ISL where it<br>
> belongs.<br>
><br>
> When we pull it into ISL, we also clean up the API for querying a bit.<br>
> Instead of simply having a table with gen numbers in it, everything uses<br>
> isl_format_supports_* queries. This allows us to special-case things in<br>
> cases where the table just doesn't quite work. For instance, there are a<br>
> number of formats that become avilable for vertex fetch on Haswell and Bay<br>
> Trail but don't exist on Ivy Bridge. This isn't something the current<br>
> table-based approach can handle properly.<br>
><br>
<br>
</span>Pulling the table into ISL seems like a great idea and helpers shouldn't<br>
hurt. It seems like the problem helpers would solve is the case when a<br>
format loses its support in a future generation - has this happened with a<br>
Gen introduction? For the example you stated, I think we could fix it by<br>
giving Baytrail a gen number between 70 and 75. This would require scaling<br>
up all brw_device_info::gen fields by 10 of course.<br></blockquote><div><br></div><div>Yes, we could call Bay Trail 7.1 or something like that but we can't really guarantee that it will be monotonic.<br></div><div>--Jason<br></div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
- Nanley<br>
<div><div class="h5"><br>
> Jason Ekstrand (6):<br>
> i965/surface_formats: Update the VB column for new formats added on<br>
> BYT<br>
> isl: Add support for quering the string name of a format<br>
> isl: Add the ISL_FORMAT_R32G32_FLOAT_LD format<br>
> isl: Add per-gen format introspection<br>
> anv/formats: Use isl_format_supports* for format introspection<br>
> i965: Use ISL for surface format introspection<br>
><br>
> Nanley Chery (1):<br>
> i965: Unset alpha blend for R10G10B10_SNORM_A2_UNORM<br>
><br>
> src/intel/isl/isl.h | 21 ++<br>
> src/intel/isl/isl_format.c | 386 ++++++++++++++++++++++++<br>
> src/intel/isl/isl_format_layout.csv | 1 +<br>
> src/intel/isl/isl_format_layout_gen.bash | 3 +-<br>
> src/intel/vulkan/anv_formats.c | 41 ++-<br>
> src/mesa/drivers/dri/i965/Makefile.sources | 3 +-<br>
> src/mesa/drivers/dri/i965/brw_context.h | 2 -<br>
> src/mesa/drivers/dri/i965/brw_state.h | 1 -<br>
> src/mesa/drivers/dri/i965/brw_state_dump.c | 8 +-<br>
> src/mesa/drivers/dri/i965/brw_surface_formats.c | 344 +--------------------<br>
> src/mesa/drivers/dri/i965/brw_surface_formats.h | 43 ---<br>
> src/mesa/drivers/dri/i965/intel_mipmap_tree.c | 5 +-<br>
> 12 files changed, 448 insertions(+), 410 deletions(-)<br>
> delete mode 100644 src/mesa/drivers/dri/i965/brw_surface_formats.h<br>
><br>
> --<br>
> 2.5.0.400.gff86faf<br>
><br>
</div></div>> _______________________________________________<br>
> mesa-dev mailing list<br>
> <a href="mailto:mesa-dev@lists.freedesktop.org">mesa-dev@lists.freedesktop.org</a><br>
> <a href="https://lists.freedesktop.org/mailman/listinfo/mesa-dev" rel="noreferrer" target="_blank">https://lists.freedesktop.org/mailman/listinfo/mesa-dev</a><br>
</blockquote></div><br></div></div>