Re: [Nouveau] NV50 compute support questions

2015-12-07 Thread Hans de Goede 

Hi

On 04-12-15 09:45, Hans de Goede wrote:


I've ordered a GTX740 (GK107) card, which should arrive soon, and
I'll be using that so I can (hopefully) focus on the llvm tgsi bits
again.


So the card arrived today and I've plugged it in tests/trivial/compute
looks much better with this. But there does seem to be one issue
(other then the atomic bits not working) :

- test_resource_indirect
(1, 0)[0]: got 0x2/0.00, expected 0x3/0.00
(3, 0)[0]: got 0x6/0.00, expected 0x7/0.00
(5, 0)[0]: got 0xa/0.00, expected 0xb/0.00
(7, 0)[0]: got 0xe/0.00, expected 0xf/0.00
(9, 0)[0]: got 0x12/0.00, expected 0x13/0.00
(11, 0)[0]: got 0x16/0.00, expected 0x17/0.00
(13, 0)[0]: got 0x1a/0.00, expected 0x1b/0.00
(15, 0)[0]: got 0x1e/0.00, expected 0x1f/0.00
(17, 0)[0]: got 0x22/0.00, expected 0x23/0.00
(19, 0)[0]: got 0x26/0.00, expected 0x27/0.00
(21, 0)[0]: got 0x2a/0.00, expected 0x2b/0.00
(23, 0)[0]: got 0x2e/0.00, expected 0x2f/0.00
(25, 0)[0]: got 0x32/0.00, expected 0x33/0.00
(27, 0)[0]: got 0x36/0.00, expected 0x37/0.00
(29, 0)[0]: got 0x3a/0.00, expected 0x3b/0.00
(31, 0)[0]: got 0x3e/0.00, expected 0x3f/0.00
(33, 0)[0]: got 0x42/0.00, expected 0x43/0.00
(35, 0)[0]: got 0x46/0.00, expected 0x47/0.00
(37, 0)[0]: got 0x4a/0.00, expected 0x4b/0.00
(39, 0)[0]: got 0x4e/0.00, expected 0x4f/0.00
(64, 1): FAIL (32)

Regards,

Hans
___
Nouveau mailing list
Nouveau@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/nouveau

Re: [Nouveau] NV50 compute support questions

2015-12-07 Thread Samuel Pitoiset 



On 12/07/2015 04:10 PM, Hans de Goede wrote:

Hi



Hi,


On 04-12-15 09:45, Hans de Goede wrote:


I've ordered a GTX740 (GK107) card, which should arrive soon, and
I'll be using that so I can (hopefully) focus on the llvm tgsi bits
again.


So the card arrived today and I've plugged it in tests/trivial/compute
looks much better with this. But there does seem to be one issue
(other then the atomic bits not working) :

- test_resource_indirect


Exactly, two or three test don't work on Kepler < GK110.
It's on my todolist, but with a low priority. :-)

Thanks for reporting this anyway.


(1, 0)[0]: got 0x2/0.00, expected 0x3/0.00
(3, 0)[0]: got 0x6/0.00, expected 0x7/0.00
(5, 0)[0]: got 0xa/0.00, expected 0xb/0.00
(7, 0)[0]: got 0xe/0.00, expected 0xf/0.00
(9, 0)[0]: got 0x12/0.00, expected 0x13/0.00
(11, 0)[0]: got 0x16/0.00, expected 0x17/0.00
(13, 0)[0]: got 0x1a/0.00, expected 0x1b/0.00
(15, 0)[0]: got 0x1e/0.00, expected 0x1f/0.00
(17, 0)[0]: got 0x22/0.00, expected 0x23/0.00
(19, 0)[0]: got 0x26/0.00, expected 0x27/0.00
(21, 0)[0]: got 0x2a/0.00, expected 0x2b/0.00
(23, 0)[0]: got 0x2e/0.00, expected 0x2f/0.00
(25, 0)[0]: got 0x32/0.00, expected 0x33/0.00
(27, 0)[0]: got 0x36/0.00, expected 0x37/0.00
(29, 0)[0]: got 0x3a/0.00, expected 0x3b/0.00
(31, 0)[0]: got 0x3e/0.00, expected 0x3f/0.00
(33, 0)[0]: got 0x42/0.00, expected 0x43/0.00
(35, 0)[0]: got 0x46/0.00, expected 0x47/0.00
(37, 0)[0]: got 0x4a/0.00, expected 0x4b/0.00
(39, 0)[0]: got 0x4e/0.00, expected 0x4f/0.00
(64, 1): FAIL (32)

Regards,

Hans


--
-Samuel
___
Nouveau mailing list
Nouveau@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/nouveau

[Nouveau] [Bug 90887] PhiMovesPass in register allocator broken

2015-12-07 Thread bugzilla-daemon 
https://bugs.freedesktop.org/show_bug.cgi?id=90887

--- Comment #40 from jr  ---
Created attachment 120393
  --> https://bugs.freedesktop.org/attachment.cgi?id=120393=edit
Patch to simplify test case creation

Thanks. I finally had some time to look at the problem again. With the attached
patch it is easier to recreate the problem, because it causes every if without
an else clause to create a critical edge with the right property (i.e. not
being the first incoming edge).

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.
___
Nouveau mailing list
Nouveau@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/nouveau

[Nouveau] [Bug 90887] PhiMovesPass in register allocator broken

2015-12-07 Thread bugzilla-daemon 
https://bugs.freedesktop.org/show_bug.cgi?id=90887

jr  changed:

   What|Removed |Added

 Status|NEW |RESOLVED
 Resolution|--- |FIXED

--- Comment #41 from jr  ---
After doing some tests with this change I'm sure that the reported problem is
fully fixed by Ilja's patch. That splitting more edges seems to help in some
cases most probably has a different cause. I'll try to take another look at the
hearthstone shaders (may not be soon).

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.
___
Nouveau mailing list
Nouveau@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/nouveau

Re: [Nouveau] [Bug 90887] PhiMovesPass in register allocator broken

2015-12-07 Thread Craig Garner 
I'll be done with my current project in a couple days.  I can give you all
the info you want for my quad head setup.  I would like to be able to at
least keep up with Fedora Current :)

Tell me what you need so I'll have it before hand.  I have a laptop I can
still communicate with people and a full workstation backup.


On Mon, Dec 7, 2015 at 2:50 PM,  wrote:

> *Comment # 40  on
> bug 90887  from jr
>  *
>
> Created attachment 120393 
>  [details] 
>  [review] 
> 
> Patch to simplify test case creation
>
> Thanks. I finally had some time to look at the problem again. With the 
> attached
> patch it is easier to recreate the problem, because it causes every if without
> an else clause to create a critical edge with the right property (i.e. not
> being the first incoming edge).
>
> --
> You are receiving this mail because:
>
>- You are the QA Contact for the bug.
>- You are the assignee for the bug.
>
>
> ___
> Nouveau mailing list
> Nouveau@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/nouveau
>
>
___
Nouveau mailing list
Nouveau@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/nouveau

Re: [Nouveau] [Bug 90887] PhiMovesPass in register allocator broken

2015-12-07 Thread Craig Garner 
I'll give you my 2 cents worth when there is an update to the driver, X, or
whatever is causing the issue in fc23.  If I push it, I might be able to
test something tomorrow if there are any updates for fc22->fc23.

Thanks for all your hard work!!!

Craig


On Mon, Dec 7, 2015 at 2:56 PM,  wrote:

> jr  changed bug 90887
> 
> What Removed Added Status NEW RESOLVED Resolution --- FIXED
>
> *Comment # 41  on
> bug 90887  from jr
>  *
>
> After doing some tests with this change I'm sure that the reported problem is
> fully fixed by Ilja's patch. That splitting more edges seems to help in some
> cases most probably has a different cause. I'll try to take another look at 
> the
> hearthstone shaders (may not be soon).
>
> --
> You are receiving this mail because:
>
>- You are the QA Contact for the bug.
>- You are the assignee for the bug.
>
>
> ___
> Nouveau mailing list
> Nouveau@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/nouveau
>
>
___
Nouveau mailing list
Nouveau@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/nouveau