[Bug 89580] Implement a NIR -> vec4 pass

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=89580

--- Comment #18 from Eduardo Lima Mitev <elima at igalia.com> ---
It has been a few weeks since my last update on the status of this work, so
here it goes:

* Support for all instructions are already in place except texture operations.
We currently have partial support for texture operations and are working hard
to finish the remaining ones and start testing them all.

* We are passing a large set of piglit tests.

$ INTEL_USE_NIR=1 ./piglit-run.py tests/all ./results/nir-vec4/ -x "tex-" -x
texture -x teximage -x texcolor -x texel -x texwrap
[23905/23905] crash: 9, fail: 147, pass: 14755, skip: 8990, warn: 4 /       
Thank you for running Piglit!

Our branch shows only ~81 regressions in piglit, compared to  vec4_visitor. We
are excluding texture related tests to the best possible, but it is very likely
that some of these 81 regressions are due to missing texture ops support.

In dEQP we still have a number of regressions but for different reasons we have
been unable to quantify them (test runs abort). So we are focusing on being
regression-free on piglit first.

As usual, a snapshot of our (work-in-progress) branch can be found at
https://github.com/Igalia/mesa/commits/nir-vec4.    

We hope to start cleaning-up and squashing patches very soon, and I will try to
keep you posted more regularly.

-- 
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/20150602/31b2c87e/attachment-0001.html>