Time to upgrade! Monster rig quad gtx 590

Generic forum to discuss Octane Render, post ideas and suggest improvements.
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
KaroBastardKiter
Licensed Customer
Posts: 132
Joined: Mon Sep 19, 2011 8:53 am
Location: London, UK

Hi guys

I'm looking to upgrade my current 'gear' into a monstrous quad gtx590 rig for some nice animations in max-octane.

Since I am not a big computer expert and I'm sure there's people who have much better understanding then me on this, do you have some advice on the componets i should use (power, cooling, cpu, ram ...)?

Thank you!
Alienware M17x R3 | GTX 460M | Quad 2.0 GHz | 8 GB
Win 7 x64 | Blender 2.59 | Octane
User avatar
gabrielefx
Licensed Customer
Posts: 1701
Joined: Wed Sep 28, 2011 2:00 pm

KaroBastardKiter wrote:Hi guys

I'm looking to upgrade my current 'gear' into a monstrous quad gtx590 rig for some nice animations in max-octane.

Since I am not a big computer expert and I'm sure there's people who have much better understanding then me on this, do you have some advice on the componets i should use (power, cooling, cpu, ram ...)?

Thank you!
buy 4 GTX580 with 3GB....you will saturate 1.5Gb of vram with medium complex scenes.

then wait for a dual Kepler 3+3GB, it will fit in your pci-e 16X 2.0 slot
quad Titan Kepler 6GB + quad Titan X Pascal 12GB + quad GTX1080 8GB + dual GTX1080Ti 11GB
KaroBastardKiter
Licensed Customer
Posts: 132
Joined: Mon Sep 19, 2011 8:53 am
Location: London, UK

but gtx 580 is 512 cores only against 1024 of gtx 590.... i thought that it's #cuda cores that matters

you mean it's better of because of the price?

how many cuda cores and total ram is the kepler going to have?
Alienware M17x R3 | GTX 460M | Quad 2.0 GHz | 8 GB
Win 7 x64 | Blender 2.59 | Octane
User avatar
Refracty
Licensed Customer
Posts: 1599
Joined: Wed Dec 01, 2010 6:42 pm
Location: 3D-Visualisierung Köln
Contact:

I am using 3 590s plus a 580 in case if the scene is more heavy.
That rigs works pretty good for my needs. If you use more then 3 590s you will need a second PSU - 1500W will not be sufficiant.
KaroBastardKiter
Licensed Customer
Posts: 132
Joined: Mon Sep 19, 2011 8:53 am
Location: London, UK

i like your bazooka rig Refracty
are you using the 580 for display only? or you can use them all together for rendering?
i remember you must have the same cards

disclaimer: sorry if my questions are dumb! :lol:

dankeschoen
Alienware M17x R3 | GTX 460M | Quad 2.0 GHz | 8 GB
Win 7 x64 | Blender 2.59 | Octane
User avatar
Refracty
Licensed Customer
Posts: 1599
Joined: Wed Dec 01, 2010 6:42 pm
Location: 3D-Visualisierung Köln
Contact:

With the new kernel you don't need the same cards any more. You can mix different cards and the 580s as well as 590s have both NF110 chips so they usally get along well.
I use all cards for rendering but if I need to work on other stuff simultaniously I deactivate the 580 so that I can use this card for 3D applications ect.
When I render a larger scene or resolution the 590s will drop but the 580 will continue to run because of the larger VRAM.
For me 1,5 Gigs of VRAM are usually enough. Because I use the 580 as primary card I can use the full VRAM of the 590s.

For your descission keep in mind what your scene complexity and output resolution is. A large output resolution will need more VRAM.

I had a recent 3D project where I needed a very fast setup. I had to hand over up to 12 renderings per night. Usually I wouldn't invest much money in such a monster rig. Before I used 4 * 460s (2GB). That was 3 times slower but still fine for most jobs.
KaroBastardKiter
Licensed Customer
Posts: 132
Joined: Mon Sep 19, 2011 8:53 am
Location: London, UK

let me get it right then. suppose my rig is 2x580s 3GB and 2x590s 1.5GB, a total of 4 cards.

1) My scene is 2.5 GB, primary card is 580:
- only the 2x 580s will work? I'll have 2x512=1024 CUDA cores available?
2) My scene is 2.5 GB, primary card is 590:
- only the 2x 580s will work? I'll have 2x512=1024 CUDA cores available?
3) My scene is 1 GB, primary card is 580:
- all 4 cards will work? I'll have 2x512+2x1024= 3072 CUDA cores available?
4) My scene is 1 GB, primary card is 590
- all 4 cards will work? I'll have 2x512+2x1024= 3072 CUDA cores available?

Am I getting it right? If yes, I don't like this compromise :lol:

I've asked for a quote and given the option to buy a GTX 590 3GB SLI - is this really 3GB??

I'm asking for quotes around and I also have the following quesitons

What processor would you suggest me to go for?

Is Silverstone Temjin TJ07 with 6 fans a good case for a 3x or 4x cards rig?

16GB RAM will do the job?

THANKS!!!
Alienware M17x R3 | GTX 460M | Quad 2.0 GHz | 8 GB
Win 7 x64 | Blender 2.59 | Octane
User avatar
t_3
Posts: 2871
Joined: Tue Jul 05, 2011 5:37 pm

KaroBastardKiter wrote:I've asked for a quote and given the option to buy a GTX 590 3GB SLI - is this really 3GB??
no. some sellers write 3gb, because it's the total amount of vram on the card (sounds better), but it is still 2x 1.5gb
KaroBastardKiter wrote:What processor would you suggest me to go for?
core i7 2600k - best performance for the money, and overclockable as hell (if you dare ;))

btw: the internal gpu of the 2600k is a nice addon too - it handles ~2mio tris scenes in open gl good enough to work with it as primary card (not for heavy duty work though), so this could help to free up vram resources for octane.
KaroBastardKiter wrote:Is Silverstone Temjin TJ07 with 6 fans a good case for a 3x or 4x cards rig?
i own the tj07, but i'm not sure if it would handle 4 cards properly (thermally). imo the tj11 is better suited if you really want to go for 4 cards.
read on here: http://www.refractivesoftware.com/forum ... j07#p77321 - and take a look at what i wrote about the fan configuration of the cards...
KaroBastardKiter wrote:16GB RAM will do the job?
i work with scenes around 2mio tris, but large texture sets (enough to fill up the 3gb of my 580s), and 16gb always were enough to have octane runing, and still having the scene open in my host, as well as doing some photoshp works in parallel.

regarding your combination examples - i think you got it :)
The obvious is that which is never seen until someone expresses it simply

1x i7 2600K @5.0 (Asrock Z77), 16GB, 2x Asus GTX Titan 6GB @1200/3100/6200
2x i7 2600K @4.5 (P8Z68 -V P), 12GB, 1x EVGA GTX 580 3GB @0900/2200/4400
KaroBastardKiter
Licensed Customer
Posts: 132
Joined: Mon Sep 19, 2011 8:53 am
Location: London, UK

thank you t_3, i'm starting to get a handle on the technical side!

so i think my best configuration would be 2x590s + 1x580 or 2x580s + 2x590, given that I aim to produce archiviz animations so what i need is SPEED!!!

with the instancing coming up i might even be better off with 3x590s and 1x580...?

I have a quesiton about this:

if i have an object that takes 10MB VRAM and i instance it say 100 times, what is the VRAM utilization at the end?

Thanks a bunch t_3 i owe you a 'helles'! :lol: :lol: :lol:
Alienware M17x R3 | GTX 460M | Quad 2.0 GHz | 8 GB
Win 7 x64 | Blender 2.59 | Octane
User avatar
matej
Licensed Customer
Posts: 2083
Joined: Fri Jun 25, 2010 7:54 pm
Location: Slovenia

KaroBastardKiter wrote:f i have an object that takes 10MB VRAM and i instance it say 100 times, what is the VRAM utilization at the end?
10 MB + few kilobytes for transformation matrices & 'stuff'
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
Post Reply

Return to “General Discussion”