<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#FFFFFF">
Hi David,<br>
yes I already got some reports from people reporting issues.<br>
<br>
That's why I want to build the kernels for non Kepler cards with the
proven to work 4.0 toolkit and only build the 3.0 kernel with the
toolkit 4.2. :)<br>
4.2 toolkit is still an RC, and it probably has some issues still. <br>
<br>
Regards,<br>
Thomas<br>
<br>
Am 05.04.2012 14:24, schrieb David Black:
<blockquote cite="mid:4F7D8EEC.7000602@yahoo.co.uk" type="cite">
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
<font face="Calibri">Hi Thomas,<br>
<br>
I know your mail is meant for Brecht so hope you won't mind my
little bit of information.<br>
<br>
While researching instancing I tried the "19 Billions of
Polygons, not a ridiculous number anymore" compared your recent
CUDA 4.2 test build designed for GTX 680, compared a recent
build from GraphicAll.org containing CUDA 4.0.<br>
<br>
With the CUDA 4.2 build I get some graphical glitches compared
to CUDA 4.0, </font><font face="Calibri">please see attached
cropped images, this </font><font face="Calibri">m</font><font
face="Calibri">ight be because I am using a GTX 460 but thought
best to let you know in case it's caused by CUDA 4.2</font><font
face="Calibri">.<br>
<br>
Thank you,<br>
<br>
David<br>
</font>
<div class="moz-signature">--
<p style="font-size:small;"><a moz-do-not-send="true"
href="http://www.3d-designs-davidblack.blogspot.com">3d-designs-davidblack.blogspot.com</a></p>
</div>
<br>
On 05/04/2012 11:04, Thomas Dinges wrote:
<blockquote cite="mid:4F7D6E15.3070502@dingto.org" type="cite">
<pre wrap="">Hi Brecht,
for the upcoming 2.63 release I propose the following actions:
CUDA:
The toolkit 4.2 is still in RC phase. Although I don't have any issues
or slowdown here (Geforce 540), I'd stick to 4.0 toolkit for now.
So I propose to build the sm_13, sm_20 and sm_21 kernels with toolkit 4.0
and the sm_30 with the toolkit 4.2 for the Kepler gpus.
I am not sure if the kernels are different for Linux or Windows, if they
aren't I can build the Kepler kernels, upload, so release builders can
manually copy the sm_30 kernel into addons/cycles/lib.
Otherwise I will create the sm_30 kernels for windows only and upload
for jesterKing.
OpenCL:
As it has been confirmed by several people in the forums, I think it's
ok to enable _KERNEL_SHADING_ for OpenCL on AMD cards.
There are some reports about not working textures, so we could enable it
without textures support for now.
But I think it's good to have this tested by a wider audience, rather
than asking people to modify on their own builds.
Should be pretty save, OpenCL is marked as Experimental feature in
Cycles anyways.
Best regards,
Thomas
</pre>
</blockquote>
<br>
<fieldset class="mimeAttachmentHeader"></fieldset>
<br>
<pre wrap="">_______________________________________________
Bf-cycles mailing list
<a class="moz-txt-link-abbreviated" href="mailto:Bf-cycles@blender.org">Bf-cycles@blender.org</a>
<a class="moz-txt-link-freetext" href="http://lists.blender.org/mailman/listinfo/bf-cycles">http://lists.blender.org/mailman/listinfo/bf-cycles</a>
</pre>
</blockquote>
<br>
<br>
<pre class="moz-signature" cols="72">--
Thomas Dinges
Blender Developer, Artist and Musician
<a class="moz-txt-link-abbreviated" href="http://www.dingto.org">www.dingto.org</a></pre>
</body>
</html>