But you can set up RR to work per GPU though, right Tim? Could you elaborate about how the 4 card systems are setup?
On Tue, Jan 19, 2016 at 2:48 PM, Matt Morris <[email protected]> wrote: > This is the thread I was talking about, haven't heard back on how is done > yet though. > > https://www.redshift3d.com/forums/viewthread/5658/ > > > > On 19 January 2016 at 19:39, Leoung O'Young <[email protected]> wrote: > >> We are using rr v7.0.16 and it is not implemented in this version. >> It is not a problem for us at the moment since we only have 2 GPU per >> machine. >> >> >> >> On 19/01/2016 2:20 PM, Matt Morris wrote: >> >> I'm sure there was a post on the redshift forum from someone who had >> managed to get RR7 working with jobs per gpu, but it sounded like it needed >> a bit of time to get it right, and conversations with Holger. >> >> On 19 January 2016 at 19:13, Dan Yargici <[email protected]> wrote: >> >>> Might want to double check with Holger first, I may be wrong about RR. >>> >>> DAN >>> On 19 Jan 2016 6:59 pm, "Tony Bexley" <[email protected]> wrote: >>> >>>> Oh okay, I think I got it now. We're still on RR 6 but maybe we need to >>>> upgrade to 7 so we can have multiple render nodes on machines. >>>> >>>> Guess I better talk to the bossman and see about getting a license >>>> upgrade. >>>> >>>> Thanks Tim! >>>> >>>> On Tue, Jan 19, 2016 at 1:21 PM, Tim Crowson < >>>> <[email protected]>[email protected]> wrote: >>>> >>>>> We settled on 2GPUs for most machines. A couple have 3. Workstations >>>>> have 1. Like Matt and Mirko said, because of non-linear performance, >>>>> you'll >>>>> need to determine at what point you want to parallelize. >>>>> >>>>> And yes, you can have a machine with 4 cards, and 4 instances of RS >>>>> each rendering on a separate card. However, be advised that all 4 >>>>> instances >>>>> of RS/XSI will pull from shared resources like the CPU and system RAM. >>>>> >>>>> If you can swing Titan Xs, their 12GB of vram make a huge difference >>>>> in performance, even if the specs of the card itself might not seem as >>>>> appealing as the 908Ti. >>>>> >>>>> -Tim >>>>> >>>>> >>>>> On 1/19/2016 2:52 AM, Matt Morris wrote: >>>>> >>>>> When I was at A Large Evil Corporation we replaced around 40 cpu nodes >>>>> with 8 dual gpu nodes. With 2 cards the scaling is pretty good (around >>>>> 1.85x). With 3 or 4 cards it would definitely be worth doing one frame per >>>>> gpu. >>>>> >>>>> On 19 January 2016 at 00:39, Artur Woźniak < <[email protected]> >>>>> [email protected]> wrote: >>>>> >>>>>> SO the question. >>>>>> How do you setup your farm. Which is better? Multiple cards per frame >>>>>> or a card per frame? >>>>>> >>>>>> Artur >>>>>> >>>>>> 2016-01-19 0:50 GMT+01:00 Tim Crowson < >>>>>> <[email protected]>[email protected]>: >>>>>> >>>>>>> We're use a 'garden' really... about a dozen dedicated machines, >>>>>>> most with 2xTitanX cards, and we have some workstations that can kick >>>>>>> in if >>>>>>> needed. You can get through a lot with far less than before. And if you >>>>>>> need to upgrade, just pop in a different card. We went from dual 770s to >>>>>>> dual TitanXs with zero hassle and the performance improvement was as >>>>>>> expected: awesome. >>>>>>> >>>>>>> -Tim >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>>> On 1/17/2016 11:38 AM, Graham Bell wrote: >>>>>>> >>>>>>> I'd really like us to look at Redshift but too embedded with Vray. >>>>>>> Keeping an eye on the possible Max integration as that could turn heads. >>>>>>> How big a farm are people using for Redshift, because we do a lot of >>>>>>> rendering. >>>>>>> On Fri, 15 Jan 2016 at 16:57, balazs kiss < <[email protected]> >>>>>>> [email protected]> wrote: >>>>>>> >>>>>>>> Hi Morten, >>>>>>>> >>>>>>>> we're also on redshift here, and I've made another small studio to >>>>>>>> switch too ( by showing a few frames and the corresponding render >>>>>>>> times ).. >>>>>>>> great stuff :) >>>>>>>> >>>>>>>> On Fri, Jan 15, 2016 at 2:54 AM, Kris Rivel < <[email protected]> >>>>>>>> [email protected]> wrote: >>>>>>>> >>>>>>>>> Loooooove RS! Switched over after briefly playing with Vray and >>>>>>>>> never looked back. It has been amazing so far. >>>>>>>>> >>>>>>>>> Kris >>>>>>>>> >>>>>>>>> On Fri, Jan 8, 2016 at 8:11 AM, Ognjen Vukovic < >>>>>>>>> <[email protected]>[email protected]> wrote: >>>>>>>>> >>>>>>>>>> I thought they mentioned having rendermaps in soft actually >>>>>>>>>> working partially, they were just looking for a method on how to >>>>>>>>>> implement >>>>>>>>>> it across the board. >>>>>>>>>> I could be completely wrong though. Either way i would like the >>>>>>>>>> volumes more :), sorry baking guys. >>>>>>>>>> >>>>>>>>>> On Fri, Jan 8, 2016 at 12:13 PM, Matt Morris < >>>>>>>>>> <[email protected]>[email protected]> wrote: >>>>>>>>>> >>>>>>>>>>> I'd say the timescale for rendermapping/baking will be months >>>>>>>>>>> rather than weeks - seems like complex volume rendering will come >>>>>>>>>>> before >>>>>>>>>>> baking. >>>>>>>>>>> >>>>>>>>>>> On 8 January 2016 at 09:28, Ognjen Vukovic < <[email protected]> >>>>>>>>>>> [email protected]> wrote: >>>>>>>>>>> >>>>>>>>>>>> It wasn't supported up till now, I think they will be >>>>>>>>>>>> introducing it in the coming weeks. >>>>>>>>>>>> >>>>>>>>>>>> On Fri, Jan 8, 2016 at 9:18 AM, James De Colling < >>>>>>>>>>>> <[email protected]>[email protected]> wrote: >>>>>>>>>>>> >>>>>>>>>>>>> sorry for the potentially silly question, but how is redshift >>>>>>>>>>>>> with Rendermap? we use it extensively with MR, and would need >>>>>>>>>>>>> redshift to >>>>>>>>>>>>> have the same capability >>>>>>>>>>>>> >>>>>>>>>>>>> James, >>>>>>>>>>>>> >>>>>>>>>>>>> On Fri, Jan 8, 2016 at 8:01 PM, Morten Bartholdy < >>>>>>>>>>>>> <[email protected]>[email protected]> wrote: >>>>>>>>>>>>> >>>>>>>>>>>>>> Wow, it is quite impressive what an impact Redshift has made >>>>>>>>>>>>>> already. >>>>>>>>>>>>>> >>>>>>>>>>>>>> >>>>>>>>>>>>>> >>>>>>>>>>>>>> I have done a fair amount of testing with Redshift too and >>>>>>>>>>>>>> find on average scenes to be 5-10 times faster than Arnold with >>>>>>>>>>>>>> comparable >>>>>>>>>>>>>> quality, plus a number of things are simpler to set up the way >>>>>>>>>>>>>> you want it, >>>>>>>>>>>>>> in part because of the rapid feedback in the renderregion. I >>>>>>>>>>>>>> have yet to >>>>>>>>>>>>>> test it on really complex scenes, so that will be the next thing >>>>>>>>>>>>>> to check. >>>>>>>>>>>>>> I agree on a lack of shader support in certain parts, especially >>>>>>>>>>>>>> compared >>>>>>>>>>>>>> to Arnold/Sitoa, so it is reassuring that Burtnyk and Co are so >>>>>>>>>>>>>> responsive >>>>>>>>>>>>>> :) >>>>>>>>>>>>>> >>>>>>>>>>>>>> >>>>>>>>>>>>>> >>>>>>>>>>>>>> We will likely also incorporate it to some extent when it >>>>>>>>>>>>>> supports vdb and volumetric rendering and see where it takes us. >>>>>>>>>>>>>> >>>>>>>>>>>>>> >>>>>>>>>>>>>> >>>>>>>>>>>>>> >>>>>>>>>>>>>> >>>>>>>>>>>>>> Morten >>>>>>>>>>>>>> >>>>>>>>>>>>>> >>>>>>>>>>>>>> >>>>>>>>>>>>>> >>>>>>>>>>>>>> >>>>>>>>>>>>>> >>>>>>>>>>>>>> Den 5. januar 2016 kl. 19:50 skrev "Emilio Hernández" < >>>>>>>>>>>>>> <[email protected]>[email protected]>: >>>>>>>>>>>>>> >>>>>>>>>>>>>> Softimage and Redshift. Best marriage ever! >>>>>>>>>>>>>> >>>>>>>>>>>>>> >>>>>>>>>>>>>> >>>>>>>>>>>>>> I have 4 licenses. >>>>>>>>>>>>>> >>>>>>>>>>>>>> >>>>>>>>>>>>>> >>>>>>>>>>>>>> >>>>>>>>>>>>> >>>>>>>>>>>>> >>>>>>>>>>>> >>>>>>>>>>> >>>>>>>>>>> >>>>>>>>>>> -- >>>>>>>>>>> <http://www.matinai.com>www.matinai.com >>>>>>>>>>> >>>>>>>>>> >>>>>>>>>> >>>>>>>>> >>>>>>>> >>>>>>> -- >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>>> *Tim Crowson **Lead CG Artist* >>>>>>> >>>>>>> >>>>>>> *Magnetic Dreams, Inc. *2525 Lebanon Pike, Bldg C, Suite 101, >>>>>>> Nashville, TN 37214 >>>>>>> *Ph* 615.885.6801 | *Fax* 615.889.4768 | >>>>>>> <http://www.magneticdreams.com>www.magneticdreams.com >>>>>>> <[email protected]>[email protected] >>>>>>> >>>>>>> *Confidentiality Notice: This email, including attachments, is >>>>>>> confidential and should not be used by anyone who is not the original >>>>>>> intended recipient(s). If you have received this e-mail in error please >>>>>>> inform the sender and delete it from your mailbox or any other storage >>>>>>> mechanism. Magnetic Dreams, Inc cannot accept liability for any >>>>>>> statements >>>>>>> made which are clearly the sender's own and not expressly made on >>>>>>> behalf of >>>>>>> Magnetic Dreams, Inc or one of its agents.* >>>>>>> >>>>>>> >>>>>>> >>>>>> >>>>>> >>>>> >>>>> >>>>> -- >>>>> www.matinai.com >>>>> >>>>> >>>>> -- >>>>> >>>>> >>>>> >>>>> >>>>> *Tim Crowson **Lead CG Artist* >>>>> >>>>> >>>>> *Magnetic Dreams, Inc. *2525 Lebanon Pike, Bldg C, Suite 101, >>>>> Nashville, TN 37214 >>>>> *Ph* 615.885.6801 | *Fax* 615.889.4768 | www.magneticdreams.com >>>>> [email protected] >>>>> >>>>> *Confidentiality Notice: This email, including attachments, is >>>>> confidential and should not be used by anyone who is not the original >>>>> intended recipient(s). If you have received this e-mail in error please >>>>> inform the sender and delete it from your mailbox or any other storage >>>>> mechanism. Magnetic Dreams, Inc cannot accept liability for any statements >>>>> made which are clearly the sender's own and not expressly made on behalf >>>>> of >>>>> Magnetic Dreams, Inc or one of its agents.* >>>>> >>>>> >>>>> >>>> >>>> >>>> >>>> -- >>>> -Tony Bexley >>>> Yes, Thee Tony Bexley >>>> >>> >> >> >> -- >> www.matinai.com >> >> >> > > > -- > www.matinai.com > -- -Tony Bexley Yes, Thee Tony Bexley
