[Bf-cycles] A couple of interesting BVH traversal papers

[Matthew Heimlich]

Ah Nathan you're correct, I remembered who said what incorrectly. Thanks
for pointing that out! Either way, 20% is more than a small gain!


On Wed, Apr 2, 2014 at 12:45 PM, Nathan Vegdahl <cessen at cessen.com> wrote:

> Bikker said that?  Are you thinking of this thread:
> http://ompf2.com/viewtopic.php?f=6&t=1894
>
> If so, it was mpeterson who claimed to have an order-of-magnitude faster
> method, not Bikker.  And mpeterson was very vague and not forthcoming about
> it.
>
> Bikker said that his own method is about 20% faster than existing methods,
> which is consistent with the results presented in the paper.
>
> --Nathan
> On Mar 28, 2014 4:02 PM, "Matthew Heimlich" <matt.heimlich at gmail.com>
> wrote:
>
>> Indeed, it doesn't appear to be an order of magnitude to me either,
>> but that's a quote from Bikker himself over on Ompf. Might be worth
>> clarifying with him.
>>
>> On Fri, Mar 28, 2014 at 6:44 PM, Brecht Van Lommel
>> <brechtvanlommel at pandora.be> wrote:
>> > Thanks for the links.
>> >
>> > The stackless BVH doesn't seem so interesting at the moment, because
>> > it's slower on both the CPU and GPU. Maybe it's more for raytracing
>> > hardware.
>> >
>> > The other paper shows about 20% faster raytracing for first bounce
>> > rays, which is neat and it's interesting that it apparently makes this
>> > kind of batching / ray ordering method consistently faster, whereas
>> > previously it was more hit and miss. I wouldn't call it an order of
>> > magnitude faster though, unless I'm missing something.
>> >
>> > Brecht.
>> >
>> > On Tue, Mar 25, 2014 at 3:51 PM, Matthew Heimlich
>> > <matt.heimlich at gmail.com> wrote:
>> >> Thought I'd toss a couple of interesting BVH papers I'd found recently
>> >> onto the mailing list. One is a stackless MBVH method that promises
>> >> nice memory savings on the GPU, the other is a paper for CPU path
>> >> tracing by Jacco Bikker which he claims can be an order of magnitude
>> >> faster than the previous state-of-the-art algorithms. Bikker is the
>> >> mind behind Arauna/Brigade for those who don't know. When he speaks up
>> >> about BVH speed, I tend to take it seriously!
>> >>
>> >> Anyway, thought they'd be interesting to some folks on here.
>> >>
>> >> Cheers!
>> >>
>> >> http://cg.iit.bme.hu/~afra/publications/afra2013cgf_mbvhsl.pdf
>> >>
>> >> http://arauna2.ompf2.com/files/cgf_article.pdf
>> >> _______________________________________________
>> >> Bf-cycles mailing list
>> >> Bf-cycles at blender.org
>> >> http://lists.blender.org/mailman/listinfo/bf-cycles
>> > _______________________________________________
>> > Bf-cycles mailing list
>> > Bf-cycles at blender.org
>> > http://lists.blender.org/mailman/listinfo/bf-cycles
>> _______________________________________________
>> Bf-cycles mailing list
>> Bf-cycles at blender.org
>> http://lists.blender.org/mailman/listinfo/bf-cycles
>>
>
> _______________________________________________
> Bf-cycles mailing list
> Bf-cycles at blender.org
> http://lists.blender.org/mailman/listinfo/bf-cycles
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.blender.org/pipermail/bf-cycles/attachments/20140402/9bb51191/attachment.htm