Open CL on the roadmap? When?
Forum rules
Please add your OS and Hardware Configuration in your signature, it makes it easier for us to help you analyze problems. Example: Win 7 64 | Geforce GTX680 | i7 3770 | 16GB
Please add your OS and Hardware Configuration in your signature, it makes it easier for us to help you analyze problems. Example: Win 7 64 | Geforce GTX680 | i7 3770 | 16GB
I realize that a lot of other things are a priority. I'm merely asking because I'm looking into getting a new computer and currently prefer a model that is not nvidia based. And I'm planning for the future. In the short term it's not a big deal if Octane supports the computer, just curious what to expect and what Radiance and company can say officially regarding Open CL/ATI support.
Windows 7 64bit | AMD Phenom 9550 | 6 gigs ram | Blender, Silo, messiah 4.5 |
Well, "the future" in the field of computers in max 1 year - that is when an previously expensive shiny comp can be already labelled as old. So if you are planing to buy an ATI card now for "the future" (that might never come) - that's a bad idea.
I think I can safely assume that the guys will first make a robust and feature-full implementation for CUDA, rather than having a half-functional implementation of both platforms. Which is only logical. So, in terms of computers development relativity, I think an OpenCL implementation is lightyears away.
I think I can safely assume that the guys will first make a robust and feature-full implementation for CUDA, rather than having a half-functional implementation of both platforms. Which is only logical. So, in terms of computers development relativity, I think an OpenCL implementation is lightyears away.
SW: Octane 3.05 | Linux Mint 18.1 64bit | Blender 2.78 HW: EVGA GTX 1070 | i5 2500K | 16GB RAM Drivers: 375.26
cgmo.net
cgmo.net
Hi,
Just get an nvidia card, they are so much better than ATI cards in many ways.
OpenCL is slow, inmature and i'm starting to think that it will fade away.
industry adoption is not good.
Those who have tried to make an equivalent to octane with opencl have 10x less speed results.
Vray was the first to show gpu based path tracing nearly 2 years ago,
yet they have not managed to get a product out using opencl.
Personally, i think OpenCL is vaporware.
it's like HD-DVD against BkuRay,
cuda is bluray for me, it's winning.
it's faster and is developed much more and faster than opencl.
i see the future becoming as such: nvidia cuda compute, ATI directX gaming.
We will port to OpenCL if the need arises but from current results of others trying to use opencl, we need more convincing first.
Radiance
Just get an nvidia card, they are so much better than ATI cards in many ways.
OpenCL is slow, inmature and i'm starting to think that it will fade away.
industry adoption is not good.
Those who have tried to make an equivalent to octane with opencl have 10x less speed results.
Vray was the first to show gpu based path tracing nearly 2 years ago,
yet they have not managed to get a product out using opencl.
Personally, i think OpenCL is vaporware.
it's like HD-DVD against BkuRay,
cuda is bluray for me, it's winning.
it's faster and is developed much more and faster than opencl.
i see the future becoming as such: nvidia cuda compute, ATI directX gaming.
We will port to OpenCL if the need arises but from current results of others trying to use opencl, we need more convincing first.
Radiance
Win 7 x64 & ubuntu | 2x GTX480 | Quad 2.66GHz | 8GB
http://shop.vray.info/siggraph-2010-cha ... demos.html
3rd video down, 15-ish minutes in
I would have to argue after seeing that video that opencl can be just as fast as cuda, imo. But I'm not a programmer, just an end user, and until these guys actually ship, I'm still hangin out with Octane
My only argument FOR opencl is the obvious "double your market share" kinda thing. I mean, once the cuda version is up and running, why not add opencl and get all those ati cards running octane as well. It can only mean more seats in the long run 
until then
cheers!
cp
3rd video down, 15-ish minutes in
I would have to argue after seeing that video that opencl can be just as fast as cuda, imo. But I'm not a programmer, just an end user, and until these guys actually ship, I'm still hangin out with Octane
My only argument FOR opencl is the obvious "double your market share" kinda thing. I mean, once the cuda version is up and running, why not add opencl and get all those ati cards running octane as well. It can only mean more seats in the long run until then
cheers!
cp
win 7 x64 | 460 gtx | q6600 8gb ram
I definitely don't see Open CL as vaporware! Completely the opposite. And I know difficulties exist when it comes to implementing it .... I think it would be worth the time. And its a laptop I'm getting. I'll just stick with one of octanes alternatives while mobile. Thanks for the insight..... pretty clear its not a near or long term possibility.radiance wrote:Hi,
Just get an nvidia card, they are so much better than ATI cards in many ways.
OpenCL is slow, inmature and i'm starting to think that it will fade away.
industry adoption is not good.
Those who have tried to make an equivalent to octane with opencl have 10x less speed results.
Vray was the first to show gpu based path tracing nearly 2 years ago,
yet they have not managed to get a product out using opencl.
Personally, i think OpenCL is vaporware.
it's like HD-DVD against BkuRay,
cuda is bluray for me, it's winning.
it's faster and is developed much more and faster than opencl.
i see the future becoming as such: nvidia cuda compute, ATI directX gaming.
We will port to OpenCL if the need arises but from current results of others trying to use opencl, we need more convincing first.
Radiance
Windows 7 64bit | AMD Phenom 9550 | 6 gigs ram | Blender, Silo, messiah 4.5 |
Don't hold your breathe waiting for vray. Talking with someone behind the scenes of VrayRT at siggraph, they are still a ways away from release. They also said they really aren't happy with opencl, but have spent so much time investing in it that they're stuck with it now.
System 1: EVGA gtx470 1280Mb and MSI gtx470 1280 in Cubix Xpander for Octane, AMD 945, 4Gb Ram
All systems are at stock speeds and settings.
All systems are at stock speeds and settings.
I certainly hope that OpenCL does not fade away, as i'm learning it right now. OpenCL is new and I think that Apple pushed Khronos too fast on the first version. CUDA has, I think, about a 3 or 4 year lead on OpenCL and is more mature because of that. Both Nvidia and AMD are about to release drivers that support OpenCL version 1.1, which is supposed to add more features and fix some issues. I guess we will see how it pans out.
As far as the speed differences between CUDA and OpenCL I have seen benchmarks that go both ways. My feeling is that, in general cases, CUDA is slightly faster then OpenCL. Is some specialized cases OpenCL is slightly faster then CUDA. For me, and I'm coming from a University background, the fact that OpenCL is a non-proprietary solution trumps any speed difference between it and CUDA. If you look at the bigger picture, whether you use CUDA or OpenCL, both solutions are many times faster then a CPU only solution. It's like saying what is better a Ferrari or a Lamborghini? They are both fast, they get you where you want to go in style, and they perform better then any typical vehicle. It's a win - win in my book.
OpenCL is new and I think that Apple pushed Khronos too fast on the first version. CUDA has, I think, about a 3 or 4 year lead on OpenCL and is more mature because of that. Both Nvidia and AMD are about to release drivers that support OpenCL version 1.1, which is supposed to add more features and fix some issues. I guess we will see how it pans out. As far as the speed differences between CUDA and OpenCL I have seen benchmarks that go both ways. My feeling is that, in general cases, CUDA is slightly faster then OpenCL. Is some specialized cases OpenCL is slightly faster then CUDA. For me, and I'm coming from a University background, the fact that OpenCL is a non-proprietary solution trumps any speed difference between it and CUDA. If you look at the bigger picture, whether you use CUDA or OpenCL, both solutions are many times faster then a CPU only solution. It's like saying what is better a Ferrari or a Lamborghini? They are both fast, they get you where you want to go in style, and they perform better then any typical vehicle.
It's a win - win in my book.Linux Mint 21.3 x64 | Nvidia GTX 980 4GB (displays) RTX 2070 8GB| Intel I7 5820K 3.8 Ghz | 32Gb Memory | Nvidia Driver 535.171
Adobe mentioned (when discussing its Premiere CUDA acceleration) that the professional market is heavily dominated by Nvidia with a ratio of 9 to 1.cpember wrote:My only argument FOR opencl is the obvious "double your market share" kinda thing.
As far as users with gaming cards go, Steam's hardware survey suggests a ratio of roughly 1.5 to 1. But I'd consider this an absolute best case scenario for ATI/AMD when it comes to professional graphics artists.
Either way, OpenCL won't nearly double the potential customer base.
that "double" comment is semantics
so what you're saying is ..
YOU have a wicked awesome widget
you sell said widget for 1000 bucks a pop
70% of people can buy it for their grommet in its current state ($70k)
with a few changes, you can sell it to the rest or the population that uses a different grommet ($30k)
you're saying you're willing to toss 30k? just like that? I mean, short of a complete re-engineering, I'm sure the gains would be worth it, market share-wise
cheers
c (devils advocate) p
so what you're saying is ..
YOU have a wicked awesome widget
you sell said widget for 1000 bucks a pop
70% of people can buy it for their grommet in its current state ($70k)
with a few changes, you can sell it to the rest or the population that uses a different grommet ($30k)
you're saying you're willing to toss 30k? just like that? I mean, short of a complete re-engineering, I'm sure the gains would be worth it, market share-wise
cheers
c (devils advocate) p
win 7 x64 | 460 gtx | q6600 8gb ram