Redshift does indeed use CUDA and is therefore currently limited to NVidia hardware. However OpenCL support is definitely in our plan, though not in the immediate term (in other words it's not something we're working on at the moment).
On Wed, Mar 13, 2013 at 12:32 PM, Mirko Jankovic <[email protected]>wrote: > grats on progress. > I guess that renderer is tightly bind to CUDA and OpenCL versions like for > AMD cards are not in plan? > > > On Wed, Mar 13, 2013 at 8:23 PM, Daryl Dunlap <[email protected]>wrote: > >> Congrats on getting this up and going. I signed up for access. Do you >> have any ballpark for where you think your price point will come in at? >> >> >> On Wed, Mar 13, 2013 at 3:18 PM, Nicolas Burtnyk >> <[email protected]>wrote: >> >>> Thanks Alok! >>> >>> That'a an excellent question. >>> We don't currently have any official support for farm rendering. That >>> being said, with a suitable render manager or some hand rolled scripts, I >>> don't see why you couldn't have multiple machines working on different >>> frames of an animation. And of course, we do plan on having more complete >>> render farm support in the future. >>> >>> Your concern regarding the hardware configuration of render nodes is >>> very valid. Redshift does require a GPU and we do not fall back on CPU in >>> the absence of a suitable GPU. The performance gap is so large that it >>> didn't really make sense for us at this stage to invest the time in making >>> this kind of thing work (though we could entertain this possibility). >>> Unfortunately, this does mean that some farms will not be good candidates >>> for Redshift, particularly those in which adding a GPU to some or all of >>> the nodes is not possible (e.g blades with no space in them). >>> >>> Keep in mind though that with the performance improvement factor >>> provided by Redshift, you can get the performance of a farm with a dozen >>> nodes from a single GPU-equipped node. We don't want to make any hard >>> performance claims yet, but our internal tests are often showing 10x to 20x >>> the performance of MR and VRay, sometimes more (on suitably complex scenes >>> - super simple 10 seconds to render in MR stuff doesn't get 10x faster). >>> So maybe you keep your existing farm for CPU renders and add a few GPU >>> equipped nodes for Redshift. >>> >>> >>> On Wed, Mar 13, 2013 at 11:45 AM, Alok <[email protected]> wrote: >>> >>>> Congrats on your achievement, looking forward to see more of it. >>>> >>>> Just a quick question. For most of the render farm setups, usually the >>>> machines do not have high-end GPU's on them for the obvious reason that >>>> farm nodes never get user interaction and hence no need for realtime >>>> graphic intensive processing operations. Do you see that as a problem ? If >>>> yes then how would you address it. Does Red Shift turn to CPU for rendering >>>> in absence of the required GPU. >>>> >>>> ALOK >>>> >>>> GANDHI >>>> >>>> / chef directeur technique - lead technical director >>>> >>>> [email protected] >>>> >>>> T: >>>> *450 430-0010 x225 >>>> >>>> F: >>>> 450 430-0009 >>>> www.modusfx.com >>>> >>>> >>>> --------------------------------------------------------------------------------------------- >>>> >>>> MODUS >>>> >>>> FX >>>> >>>> 120 Rue Turgeon, >>>> >>>> Sainte-Therese (Quebec) CANADA J7E 3J1 >>>> >>>> Follow us on >>>> Facebook <http://www.facebook.com/ModusFX> >>>> >>>> & >>>> Twitter <https://twitter.com/Modusfx> >>>> * >>>> On 13/03/2013 2:33 PM, Nicolas Burtnyk wrote: >>>> >>>> Yes - I'll try to make a video of that if I can get set up correctly >>>> for it. >>>> >>>> Note that this is 2 mins on a GTX 470 which is nothing special in >>>> terms of GPUs. You can expect significantly better times with a GTX 580 >>>> for example. I don't have official times for that card, but I'd guess >>>> under 1.5 minutes. >>>> >>>> These kinds of times really underscore the power of biased rendering. >>>> When you need to reduce noise, you have a lot more options than "let's >>>> just throw a ton more samples at the whole thing". >>>> >>>> >>>> >>>> >>>> On Wed, Mar 13, 2013 at 11:28 AM, olivier jeannel < >>>> [email protected]> wrote: >>>> >>>>> The classroom is *really* 2min render ? >>>>> >>>>> Congrats to you, sending a request :) >>>>> >>>>> Le 13/03/2013 19:18, Steven Caron a écrit : >>>>> >>>>> congrats to you and your team! i was wondering when we would see/hear >>>>> about your work. >>>>> >>>>> it would be great to see a video demonstration of redshift in >>>>> softimage. >>>>> >>>>> >>>>> On Wed, Mar 13, 2013 at 11:12 AM, Nicolas Burtnyk < >>>>> [email protected]> wrote: >>>>> >>>>>> Hello folks, >>>>>> >>>>>> In March of last year, 2 colleagues and I left our jobs as software >>>>>> developers in the games industry to form our own company - Redshift. >>>>>> Our goal was to apply our experience with graphics hardware to the >>>>>> problem of "offline" rendering. >>>>>> Artists friends had been asking us for years why Mental Ray and other >>>>>> renderers were not taking advantage of the GPU. >>>>>> As the ideas bounced around in our heads, we figured we'd take a >>>>>> crack at it. As it turns out, it's really freakin' hard, but not >>>>>> impossible! >>>>>> >>>>>> Today, we're very excited to announce the official launch of >>>>>> Redshift v0.1 alpha, to our knowledge, the world's first fully >>>>>> GPU-accelerated biased renderer. >>>>>> Redshift supports multiple GI solutions: Brute-Force GI, Irradiance >>>>>> Caching (aka Final Gather), Irradiance Point Cloud (aka Light Cache) and >>>>>> Photon Mapping (GI and Caustics). >>>>>> All are fully GPU-accelerated and perform many times faster than >>>>>> similar CPU-based offerings. >>>>>> >>>>>> A problem that plagues many GPU renderers on the market is that >>>>>> they are limited by the available VRAM on the graphics card (and most >>>>>> systems have significantly less VRAM than main memory). >>>>>> Redshift addresses this by using an out-of-core architecture for >>>>>> geometry >>>>>> and textures allowing you to render scenes with tens of millions of >>>>>> polygons and gigabytes of textures with off-the-shelf, inexpensive >>>>>> hardware. >>>>>> >>>>>> Redshift currently integrates directly with Softimage 2011 through >>>>>> 2013 and Maya 2011 through 2013 on Windows XP or higher. 3ds Max support >>>>>> is in development. To run Redshift, you'll need an NVidia graphics card >>>>>> supporting compute 1.2 or higher with 1GB VRAM or more. >>>>>> >>>>>> You can check out our website http://www.redshift3d.com for more >>>>>> info. >>>>>> >>>>>> We're currently in closed alpha, and looking to find some >>>>>> interested alpha testers on this list. >>>>>> Our goals for alpha are to shake out bugs and gather feedback from >>>>>> users to help focus our development efforts. >>>>>> >>>>>> If you're interested in taking Redshift for a spin, drop me an >>>>>> email at [email protected]. >>>>>> >>>>>> Some sample renders: >>>>>> http://s3.redshift3d.com/jp_studio_icp_1280.png >>>>>> http://s3.redshift3d.com/mazda_1024.png >>>>>> http://s3.redshift3d.com/AI_V8_S10_1200.png >>>>>> >>>>>> And the classroom scene from the "Octane Render" thread from a >>>>>> couple weeks ago. 2 minutes on a single GeForce GTX 470. >>>>>> http://s3.redshift3d.com/classroom.png >>>>>> >>>>>> I'd also like to thank all of those on this list who've answered my >>>>>> various questions about the XSI SDK. Your help has made our integration >>>>>> of >>>>>> Redshift with Softimage very strong. >>>>>> >>>>>> Cheers! >>>>>> >>>>>> Nicolas >>>>>> >>>>>> >>>>> >>>>> >>>> No virus found in this message. >>>> Checked by AVG - www.avg.com >>>> Version: 2012.0.2238 / Virus Database: 2641/5646 - Release Date: >>>> 03/03/13 >>>> Internal Virus Database is out of date. >>>> >>>> >>>> >>> >> >
