iray cloud server

kyoto kid · February 2018

Szark said:

Richard is correct no additional plug-ins required. I know because I use Jack's server for long winded renders on my PC. Hour and a half on my rig, 2-3 mins on his server.

..so how does it know how to interpret duf scene files? Even Lux and Octane, which are standalone engines, need a plugin for Daz.

My interest is I am looking at using an older Dell workstation as an "in house" render server, not sending files to a remote server on the Net.

kyoto kid · February 2018

Ivy said:

kyoto kid said:

Ivy said:

kyoto kid said:

...however what if you wanted to build your own instead of renting time on someone else's server? (of course that would also mean getting an Iray Standalone licience[s]).

you can do like I did . I run windows 7 pro. I got a external 6 slot PCI express Alienware Graphic Amplifier Box, I put 2- 980ti 12 gigs GPU in , it runs on windows7 pro desktop using the MS visual-basics download and install a MS server RUN ASP.NET run the commands for C+++ with the included instructions in the kit to stack your graphic cards GPU cores and Vram. It took a little while to set up and some help from someone I knew that was good at setting up servers for crypto-mining

But it turned 12gigs SLi of vram into 18gigs of useabel of vrams. I can't say I notice if it sped up the renders any more . But i can build much larger scenes now with out BSOD because I maxed out my GPU . One thing which we have done recently was get a cheaper dedicated gtx980SC 4 gig card just for the dual monitors display. so we did not have to do a loop back for the display. Render farm Lite - no render farm GPU license required,

...VRAM doesn't stack for rendering purposes. If you have two 11 GB cards all you get is 11 GB. You do get the benefit of double the cores but that is it.

Really then you better tell my system that..lol Yes gpu will stacks vram in a graphic amplifier and running it in a windows server. do the research thats what they are made for .. lap tops, desktops and and lesser devices that need GPU & vram Boost

Edit: I should have mention the only reason I'm not getting the whole 24 gigs vram is because I have no independent ram chips set in the gpu box yet, this is a fairly new set up so I using some of the Gpu vram as shared memory to run the graphic amplifier. I do have 8 gig ram chipset coming from crucial to replace the shared memory. But yeah this was the only solution for stacking vram

...NVLink (Volta Quadro 100 and Tesla 100) was developed to allow for "memory pooling" and that is only for computational purposes.

The attachments appear to be the tab for the Daz remote server.

If memory pooling for rendering with GTX cards (particularly older generation cards as the 980 Ti [Maxwell]) was possible then why haven't people already been doing it?

Kendall Sears · February 2018

kyoto kid said:

Szark said:

Richard is correct no additional plug-ins required. I know because I use Jack's server for long winded renders on my PC. Hour and a half on my rig, 2-3 mins on his server.

..so how does it know how to interpret duf scene files? Even Lux and Octane, which are standalone engines, need a plugin for Daz.

My interest is I am looking at using an older Dell workstation as an "in house" render server, not sending files to a remote server on the Net.

You might want to investigate your cost to run that box. As an example, just ONE of my render servers will cost approx $100/month in electricity if left running 24/7 and I'm in a very low cost/kWh region. This is a combination of both rendering and idle time. When I fire up the whole rack things get ugly.

Kendall

Richard Haseltine · February 2018

kyoto kid said:

Szark said:

Richard is correct no additional plug-ins required. I know because I use Jack's server for long winded renders on my PC. Hour and a half on my rig, 2-3 mins on his server.

..so how does it know how to interpret duf scene files? Even Lux and Octane, which are standalone engines, need a plugin for Daz.

It doesn't get the .duf file, any more than the local Iray does - it gets data aleady prepared (by DS) for Iray to render.

Greymom · February 2018

Ivy said:

Greymom said:

prixat said:

https://www.nvidia.com/content/dam/en-zz/Solutions/design-visualization/documents/iray-plugin-faq-sheet-20171120-r5-lr-FINAL.pdf

This comes from the changes last November, nVidia no longer sell the Plugins or the Server. They've off-loaded the sales and support to third parties.

Migenius got the Iray for Rhino plugin and the Iray Server. No wonder they were first to market with this service!

What? They are passing it off already? Sounds like they have lost interest in the Rendering business. Maybe the third party will discount the licenses.

Yes I can . I keep saying its Alienware. don't ask me why buts its not actually..lol

We are using the VisionTek it has 6 slots for 3 dual card. but will hold 4 standard cards. i have 2 - 980ti's in it now https://www.bestbuy.com/site/visiontek-thunderbolt-3-egfx-external-graphics-accelerator-enclosure/6163995.p?skuId=6163995

it looks like this when set up, sitting under your desk, make sure to get the liquied side cooler add on for it it not much more

Many thanks, must look into this!

kyoto kid · February 2018

Kendall Sears said:

kyoto kid said:

Szark said:

Richard is correct no additional plug-ins required. I know because I use Jack's server for long winded renders on my PC. Hour and a half on my rig, 2-3 mins on his server.

..so how does it know how to interpret duf scene files? Even Lux and Octane, which are standalone engines, need a plugin for Daz.

My interest is I am looking at using an older Dell workstation as an "in house" render server, not sending files to a remote server on the Net.

You might want to investigate your cost to run that box. As an example, just ONE of my render servers will cost approx $100/month in electricity if left running 24/7 and I'm in a very low cost/kWh region. This is a combination of both rendering and idle time. When I fire up the whole rack things get ugly.

Kendall

...well I wouldn't be running it 24/7 only when actually rendering.

I'm also talking about one box with dual 8 - 10 core Xeons, 64 - 128 GB of memory and no dedicated GPU cards (as they are too stupid expensive today) networked with my current systems, not a rack of multiple blades that would need it's own dedicated power trunk due to the high amount of current it would draw (I live in a rent controlled apartment, that kind of modification is a no-no).

100$ per month comes to about 3.50$ per day which is still lot less than 70$ - 95$ per day or 900$ per week renting time on someone else's servers. Also live in a fairly low KW/h region as well and if I didn't have to run the heat like I did during a cold snap in December, my last bill would have been under 40$.

Ivy · February 2018

kyoto kid said:

Ivy said:

kyoto kid said:

Ivy said:

kyoto kid said:

...however what if you wanted to build your own instead of renting time on someone else's server? (of course that would also mean getting an Iray Standalone licience[s]).

you can do like I did . I run windows 7 pro. I got a external 6 slot PCI express Alienware Graphic Amplifier Box, I put 2- 980ti 12 gigs GPU in , it runs on windows7 pro desktop using the MS visual-basics download and install a MS server RUN ASP.NET run the commands for C+++ with the included instructions in the kit to stack your graphic cards GPU cores and Vram. It took a little while to set up and some help from someone I knew that was good at setting up servers for crypto-mining

But it turned 12gigs SLi of vram into 18gigs of useabel of vrams. I can't say I notice if it sped up the renders any more . But i can build much larger scenes now with out BSOD because I maxed out my GPU . One thing which we have done recently was get a cheaper dedicated gtx980SC 4 gig card just for the dual monitors display. so we did not have to do a loop back for the display. Render farm Lite - no render farm GPU license required,

...VRAM doesn't stack for rendering purposes. If you have two 11 GB cards all you get is 11 GB. You do get the benefit of double the cores but that is it.

Really then you better tell my system that..lol Yes gpu will stacks vram in a graphic amplifier and running it in a windows server. do the research thats what they are made for .. lap tops, desktops and and lesser devices that need GPU & vram Boost

Edit: I should have mention the only reason I'm not getting the whole 24 gigs vram is because I have no independent ram chips set in the gpu box yet, this is a fairly new set up so I using some of the Gpu vram as shared memory to run the graphic amplifier. I do have 8 gig ram chipset coming from crucial to replace the shared memory. But yeah this was the only solution for stacking vram

...NVLink (Volta Quadro 100 and Tesla 100) was developed to allow for "memory pooling" and that is only for computational purposes.

The attachments appear to be the tab for the Daz remote server.

If memory pooling for rendering with GTX cards (particularly older generation cards as the 980 Ti [Maxwell]) was possible then why haven't people already been doing it?

okay I'm not a tech person as I have stated to you a billion time..lol so I'm not even going to try to explain it

so I wrote kal my friend from geek squad this is his replay back

-------------------------------------------------------------------------------------

ME: Hi Kal

I had a convo with one of my friends today about my gpu you put in . is my vram stalking or is it still in SLI

Kal: No neither., it does not stack proper as you would think. but it does stack in the server.

AFR requires EACH card to have a copy of ALL graphical data as it has to render the ENTIRE screen.
SFR does NOT require each card, but at the same time, each card is working on completely different portion of the screen, so they only have the data they need to render their portion. HOWEVER, if there's any object that cuts and/or duplicated across both portions, then that data is also duplicated in BOTH card's VRAM. They do NOT access one from the other.
The reason is because the time and processing power the memory controllers are in each GPU used in communicating between the GPUs just to transfer data is time lost.
So, while you can "technically" call it "stacking" in the sense you're not duplicating data, each GPU is still using it's own memory pool by itself through the windows server we set up. so, it's not shared vram stacking "technically" its server stacking the vram together, which is what we are doing for bitecoin mining for using block rates. and how render farms set up their gpu's . I set your gpu's up the same way. so even though each card is only 12 gigs on their own, running through the windows server combines gpu usage as a memory pool. Tell your friend he is right and pay the bet. lol

Hurdy3D · February 2018

I just bought a Laptop with nVidia gtx970 GPU. This renders quick enough for me and I think it‘s from a long term perspective much cheaper than using a render farm.

drzap · February 2018

I'm diggin' this at the moment. Really cool. 1 Quadro P6000 workstation for 90 cents an hour. Lots of options out there at the moment. https://www.paperspace.com/

kyoto kid · February 2018

..so how do their subscriptions and charges work? There is the hourly and the 5$ monthly storage fee. The P5000 would probably be sufficient enough for most rendering and 65¢ an hour comes to only 15,60$ per day far better than 90$/day.

Sounds also like you have to install your software on the rented systems. Isn't that against most EULAs and licencing (like if you only have a single system non floating licence)?

AllenArt · February 2018

Greymom said:

Ironically, an ad from Wal-Mart just popped up in my browser:

6 Packs of NVIDIA GeForce GTX 1080 Founders Edition, 8GB GDDR5X PCI Express 3.0 Graphics Card

By: NVIDIA

$6,299.99 List $7,748.99 Save $1,449.00 Only one left!

Darn, and I wanted two packs.....

Hmmmm....maybe if I sold a kidney.....

drzap · February 2018

kyoto kid said:

..so how do their subscriptions and charges work? There is the hourly and the 5$ monthly storage fee. The P5000 would probably be sufficient enough for most rendering and 65¢ an hour comes to only 15,60$ per day far better than 90$/day.

Sounds also like you have to install your software on the rented systems. Isn't that against most EULAs and licencing (like if you only have a single system non floating licence)?

They have a page that explains the charges in detail. You have to work out your own licencing problems. For Daz Studio, this isn't a problem. My free Maya license allows me to install on two machines. but unfortunately they don't have a multi-gpu option so I'll just use them to host my video server.

kyoto kid · February 2018

...so do you have to install Daz on their system, and what about the content library and runtimes so it can find all the meshes/textures? The price is pretty decent considering the other rendering services I've seen, and even the resources of a single P5000 would be a pretty huge step up from rendering on the CPU like I'm stuck with now.

drzap · February 2018

kyoto kid said:

...so do you have to install Daz on their system, and what about the content library and runtimes so it can find all the meshes/textures?

It's your machine, so you would operate it exactly as you would your own. I wouldn't bother uploading my entire runtime, just the files I need for rendering the job, but the process is the same as if it were on your local desktop. or you could have the machine access your hard drive like on a WAN.

kyoto kid · February 2018

...hmm so I wouldn't get a bunch of nicely rendered grey boxes, or multiple path errors?

adacey · February 2018

Ivy said:

Yes I can . I keep saying its Alienware. don't ask me why buts its not actually..lol

We are using the VisionTek it has 6 slots for 3 dual card. but will hold 4 standard cards. i have 2 - 980ti's in it now https://www.bestbuy.com/site/visiontek-thunderbolt-3-egfx-external-graphics-accelerator-enclosure/6163995.p?skuId=6163995

That link still appears to not be what you've pictured in your setup. You say there's room for 6 slots, 3 dual slot cards, but the link appears to only hold 1 dual slot card. I'd really like to know what the enclosure is that you're using.

Ivy · February 2018

When clicked the link it appearss like the site may have changed the products link that was originally posted to that one on sale., I suggest you go to Best Buy that is where I bought mine., like I said Kal Radcliff from the Geek squad set me up. you'll find that Best Buy actually have a bunch of different sizes so if you need bigger ones they do have them. . The display they used was with a box that had 12 - 1080ti's -demos-cards in it (drool), so they do have bigger ones for your needs

Mine is s a VisionTek - its Thunderbolt3 , GFX series , its just a plain jane External Graphics Accelerator Enclosure, this one has 6 slots for 3 dual card, which I have 2 cards in it,

laststand6522732 · February 2018

This link talks about these boxes, which they call eGPUs.

Kendall is right about electricity costs being noticeable. I have a 1500VA UPS that keeps track of electricity costs for what's plugged into it. There is also an inexpensive ($20) device called a Kill A Watt Electricity Usage Monitor that can measure power usage.

Ivy · February 2018

I had Geek squad set up mine. I'm pretty computer dumb , I'm on win7 using -windows server 2014

If you were to use a Data mining program for the GPU server to do bit-mining it cost like $5000 for a block chain to get started . But if you are combining GPU's in a graphics Accelerator Enclosure, all you need is a way to connect it & to control your GPU's I'm connected through USB 3 using NIVIDA EVGA precision x oc server controller. which comes with its own GPU server & was included with my EVGA titan cards when i bought them.. This options lets me changes my gpu's on the fly like the over cocking and power settings etc. the graphics Accelerator Enclosure is just a place to keep your equipments from being knocked around.

But a lot of people just hang ther GPU'S from wires rigs, so you really don't even need a graphics Accelerator Enclosure, just some kind of wire rack like this so you can cable tie your gpu cards to it

All you have to do just do a Google search for homemade wire rig for gpu cards and you'll see thats you can combine a ton of GPU's like that. its how you connnet, power & control them that whats matters not the enclosure

laststand6522732 · February 2018

Ivy said:

I had Geek squad set up mine. I'm pretty computer dumb , I'm on win7 using -windows server 2014

If you were to use a Data mining program for the GPU server to do bit-mining it cost like $5000 for a block chain to get started . But if you are combining GPU's in a graphics Accelerator Enclosure, all you need is a way to connect it & to control your GPU's I'm connected through USB 3 using NIVIDA EVGA precision x oc server controller. which comes with its own GPU server & was included with my EVGA titan cards when i bought them.. This options lets me changes my gpu's on the fly like the over cocking and power settings etc. the graphics Accelerator Enclosure is just a place to keep your equipments from being knocked around.

But a lot of people just hang ther GPU'S from wires rigs, so you really don't even need a graphics Accelerator Enclosure, just some kind of wire rack like this so you can cable tie your gpu cards to it

All you have to do just do a Google search for homemade wire rig for gpu cards and you'll see thats you can combine a ton of GPU's like that. its how you connnet, power & control them that whats matters not the enclosure

At the moment I don't have any GPU storage space problems myself, but anything is possible. I've seen the naked rigs. There's something about them that says money is meaningless, like lighting a cigar with a $100 bill. I do think that we are living in a time where processors are making an evolutionary leap. RAM was once $500 a meg. When I bought two meg, all the box would hold, the dumbfounded salesman asked "What are you going to do with it?" I said, "Everything." I'm still working on it.

Old man with new GPU, below.

callmepaul · August 2018

I'm having a problem that when I try to render a large scene with a remote iRay Server, DAZ crushes while I'm uploading the scene. It just closes and I have to start over. Any idea why this may be happening and how to avoid it?

Thanks!

JazzyBear · August 2018

callmepaul said:

I'm having a problem that when I try to render a large scene with a remote iRay Server, DAZ crushes while I'm uploading the scene. It just closes and I have to start over. Any idea why this may be happening and how to avoid it?

Thanks!

My UPLOAD speed was too slow when I tried one. Their server kept dropping the connection.

callmepaul · August 2018

JazzyBear said:

callmepaul said:

I'm having a problem that when I try to render a large scene with a remote iRay Server, DAZ crushes while I'm uploading the scene. It just closes and I have to start over. Any idea why this may be happening and how to avoid it?

Thanks!

My UPLOAD speed was too slow when I tried one. Their server kept dropping the connection.

Yes, my upload speed was not fast either, but that should not make DAZ to crush. I tried contacting support but they told me "it's not their problem" and that I should deal with NVIDIA. I tried explaining that the iRay server was not the problem since what was unexpectedly closed was their software but they still did not answer.

peteanderson1212 · May 2019

Szark said:

Jack Tomalin has a Iray server working I will drop him a message for you.

Hiii, I'd like to know about these servers too and if you could drop a message to him for me too

Daz Jack Tomalin · May 2019

peteanderson1212 said:

Szark said:

Jack Tomalin has a Iray server working I will drop him a message for you.

Hiii, I'd like to know about these servers too and if you could drop a message to him for me too

Dropped you a PM

Notifications

iray cloud server

Comments

Adding to Cart…