[Bug 92760] Add FP64 support to the i965 shader backends

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=92760

--- Comment #11 from Connor Abbott <cwabbott0 at gmail.com> ---
(In reply to Iago Toral from comment #10)
> Hi Jason, Connor:
> 
> we have fixed all the regressions in gen7 and cleaned-up the code, the
> result can be found in this branch:
> 
> https://github.com/Igalia/mesa/commits/i965-fp64-v4-no-regressions-clean
> 
> There are probably a few more things we can squash from the top commits, but
> I don't think it is going to be much smaller than this. Right now there are
> 113 commits in there, so it is a  big series (and it is not even complete!)
> 
> While reviewing the code I noticed that there are some commits that fix
> things in master or are even unrelated to fp64, so we probably want to send
> these for review early. One of these commits is the nir_type_unsigned ->
> nir_type_uint rename from Jason that I guess we want to land soon to ease
> future rebases. These patches are at the beginning of the series but I moved
> them to this branch for easy reference:
> 
> https://github.com/Igalia/mesa/commits/fp64-early-review
> 
> If you don't have objections I'd like to send these for review ASAP.

Sure, go ahead. Note that earlier I sent out a series with some of those
patches:

http://lists.freedesktop.org/archives/mesa-dev/2015-August/091724.html

and some of them got r-b's so you can probably just push them.

One commit I might suggest moving earlier and sending out now is "i965/vec4:
avoid dependency control around Align1 instructions". It uses the
DOUBLE_TO_FLOAT and FLOAT_TO_DOUBLE opcodes that are introduced in the series,
so you'd have to split the patch into two parts: the existing patch minus those
opcodes, and adding those opcodes back in. Then we could get the first part
reviewed now, since the thing its trying to do isn't directly related to fp64,
and it could theoretically fix existing bugs. Also, apparently the PACK opcode
I added could be used for images as well, so we could try landing that ahead of
time too.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/intel-3d-bugs/attachments/20151118/17a34be1/attachment.html>