Search found 26 matches

by atlas
Fri Jul 22, 2016 7:26 pm
Forum: Visuals
Topic: GPU VCM
Replies: 30
Views: 25774

Re: GPU VCM

Just curious, when do you decide to present a frame? Are you trying to pull a full 8-bounce sample off per pixel before you present, or presenting at every bounce? I always struggle with this because frame display has a bit of overhead, but for interactivity you want to present quite often.
by atlas
Wed Jul 13, 2016 8:17 pm
Forum: Visuals
Topic: GPU VCM
Replies: 30
Views: 25774

Re: GPU VCM

I was actually kind of curious about persistent kernels, so I added it to my pathtracer and got about a 10% slowdown. I also noticed the same behavior, where using about 4 blocks per SM was optimal. I think with modern cards it may just be best to throw everything at it and let the hardware schedule...
by atlas
Thu Jul 07, 2016 11:02 pm
Forum: Visuals
Topic: GPU VCM
Replies: 30
Views: 25774

Re: GPU VCM

A SM can keep more than 1 block 'resident' at a time, as long as resources (register usage, shared memory, etc.) allow for it, and the warp scheduler can actually pull a warp from any of these 'resident' blocks at a given time. So it may be that your kernel permits more than 1 block to be resident o...
by atlas
Thu Jul 07, 2016 8:54 pm
Forum: Visuals
Topic: GPU VCM
Replies: 30
Views: 25774

Re: GPU VCM

Yikes, reading/writing ray states to global memory every bounce sounds scary, but GPUs never manage to surprise me. I suppose you're also in a good state now for doing complex materials as long as you're sorting for coherency like the megakernel paper does. Good work, keep us updated.
by atlas
Sat Jun 11, 2016 8:08 pm
Forum: General Development
Topic: super fast parallel sbvh construction
Replies: 6
Views: 10954

Re: super fast parallel sbvh construction

Don't forget that GPUs have a hard time dealing with branching, and coordinating the memory movement is challenging. Some BVH algorithms like LBVH map really well to GPUs, but I have a feeling this algorithm would be more challenging.
by atlas
Fri Jun 10, 2016 11:43 pm
Forum: Visuals
Topic: GPU VCM
Replies: 30
Views: 25774

Re: GPU VCM

Very fast, great work. Hard to get that caustic in the mirror, those are some difficult paths. You mentioned you use a microfacet BRDF, I've been looking for an efficient parameterized microfacet implementation, something branchless if that's even possible. I'm new to the materials side of things, d...
by atlas
Tue Jun 07, 2016 10:49 pm
Forum: General Development
Topic: super fast parallel sbvh construction
Replies: 6
Views: 10954

Re: super fast parallel sbvh construction

This is interesting, you either find guys with really good theory of spatial subdivision, or really good computer science theory with knowledge about vector units, memory hierarchy, etc. When the two collide, good things happen! Great work
by atlas
Tue Jun 07, 2016 7:19 am
Forum: Visuals
Topic: GPU VCM
Replies: 30
Views: 25774

Re: GPU VCM

That's a good idea, this should be close to what you're talking about (if you haven't seen it yet): http://www-sop.inria.fr/reves/Basilic/2015/PRDD15b/PCBPT.pdf

If you're going to create all those light paths, better to use them for something right? :D
by atlas
Tue Jun 07, 2016 12:59 am
Forum: Visuals
Topic: GPU VCM
Replies: 30
Views: 25774

Re: GPU VCM

Do you mind sharing a detail? How do you create and store light vertices on the GPU - in global memory, and then all blocks and threads select from them randomly? Or in shared memory where each block has its own unique light vertices local to the block that are connected to? P.S. - looks great, appe...
by atlas
Thu Jun 02, 2016 6:09 pm
Forum: Visuals
Topic: GPU VCM
Replies: 30
Views: 25774

Re: GPU VCM

It's interesting seeing the clear distinction between sampling methods, great work and great to hear from you. Time to upgrade that GPU though...

Go to advanced search