OTOY Forums

brasco wrote:

rappet wrote:I doubt that... doing some math on my benchmark recalculating the seconds it is appr the same,
and are 2 Z's bit faster (164 sec versu 170sec).
And this is in a case where they both have good space to breathe.

In practise stuffing them in my other case 780T with only 6slots on mobo.. then the two Z's have good air to breathe, and the triple X will be stuck together.. then the two Z's with breating space will perform better then triple X getting too hot.. that is my guess... I wish I could practice this by having triple X

So, again.. it all depends. If I only had this 780T case twith mobo 6slot..double Z is definitely favour.
In another build with enough space for triple X, the triple X would be preferable.

cheers,

Yeah, I agree, on air you'd definitely want space around the back of the cards, they draw air from the front to back, so they stack nicely I get ~5% temp increase on the middle card.

OctaneBench shows 2 TitanZs @ an average of 362% the 3 X's 382% so ~5% between them overall, would be nice to test that in a real scenario.
Can you throw me the benchmark scene you're using, I'll can run it on the 3xTitanX's, that way we could get a better real world idea of the difference.

cheers
brasc

I used 2.22 with the 'standard' Octane Benchmark.... octane_22_benchmark.ocs
http://render.otoy.com/forum/viewtopic. ... +benchmark
The printscreen is 8000 samples rendered with double Titan Z not overclocked and with slot space between cards.

greetz,

edited: the other test we could do is run the OctaneBench http://render.otoy.com/octanebench/
I'll drop numbers in a few minutes.

edited 2:
I just got 324 doing the OctaneBench, and that is the lowest number on dounle Z test, but the higher numbers are watercooling.
Ofcourse the numbers to compare X and Z are already in the OctaneBench and that is the better test ... you are right triple X beats double Z... but then, these numbers don't tell anything about spacing cards, watercooling or not, and so on.

Anyway, the Titan Z is very hard to get, and prices went up because there are only a few left and soon will be unavailable.
So having a choice between x and Z will very shortly be only theory

see you

btw: triple X? man, that's fantastique

Notiusweb wrote:Okay, I know availability and pricing are going to be factors in real world, but as a theoretic exercise, which would you rather have:
(1) 3x GTX Titan X - 12 GB Vram w/ 9,216 Cuda Cores; or
(2) 2x GTX Titan Z - 6 GB VRAm w/ 11,520 Cuda Cores

This is assuming you had $3,000 US and Titan X is $1,000 and Titan Z is $1,500.
Also assume your platform can handle it operably.

A) 6GB additional VRam (with X), or 2,304 more cores (with Z) ?
B) What if it was for projects on a deadline, vs your own artistic endeavors?
C) What if power and heating were an issue, vs not an issue?

Oh well...Thought it might be fun to see what people think

If I had only $3k (USD) I'd rather have 2x new Titan Xs now that the price of a new Titan Z that is actually in stock sells for > $2k (USD) and the price of a Titan X that is actually in stock sells for more a little more than $1k (USD). But if I could by any GTX cards for that $3k budget, then I'd rather have 8x EVGA refurbished GTX 780 6Gs for $350 ea. from NewEgg. However, since my budget was $2k, I bought only 5 more GTX 780 6Gs.

I do like the way Guys use word "only" =DDD..

rappet wrote:edited: the other test we could do is run the OctaneBench http://render.otoy.com/octanebench/
I'll drop numbers in a few minutes.

edited 2:
I just got 324 doing the OctaneBench, and that is the lowest number on dounle Z test, but the higher numbers are watercooling.
Ofcourse the numbers to compare X and Z are already in the OctaneBench and that is the better test ... you are right triple X beats double Z... but then, these numbers don't tell anything about spacing cards, watercooling or not, and so on.

Yeah, you're right easier to use OctaneBench.

Here's my 24/7 setup:
421.51% on the latest run 1403-1466Mhz boost. Temps are all below 82C with a more aggressive fan profile than the default.

The cards are just stock cooled cards stacked with zero gaps, that's not a problem since they draw air from the front to the back not vertically. The top card is the hottest so clocks the lowest boost @ 1403Mhz
I've run these clocks for +48Hrs straight over a weekend and not had an issue - they do, however, heat the house up!



cheers
brasc

Great numbers Brasco!
Looking at the bench results triple Titan X without boosting looks like giving 366 instead of 421, so that is great boost you have.
Comparing double Z with 324 to triple X with 366 seems more fair, and the triple X is still number one in this comparison.

Yeah that 366 seems low, the cards can't be working optimally since all of them have Nvidia's GPU Boost 2.0, they automatically boost as you probably know.

Mine aren't stock BIOSes (easy to change) so they can boost higher and draw more power, if I was going to only get 366 with 3xTitan Xs, I wouldn't have bought them!
If people don't need the VRAM, I'd say wait and see what the GTX 980 Ti offers, or even the AMD cards once 3.0 drops.

cheers
brasc

Brasco: your 3 Titan X are just stock air, 3 slots in a row? how's the noise on that? what mobo do you have? I'm curious because I'd like to do a 3x Titan X setup if I can snag them when they come back in stock. I'd rather not do water if I don't have to just to simplify things, so your results are promising.

Yes, stock air 3 slots in a row, the fan profile is aggressive so yes it gets loud, but I have them further away from me and I wear headphones, I don't think people who are used to silence could cope, but the environment around me isn't silent and the fans are a very constant noise that ends up as background. Even with a less aggressive fan profile you'd still see a good boost and keep the noise more palatable, I'd say it's more important to have good air flow in the case and water would be ideal.

cheers
brasc

brasco wrote:Yes, stock air 3 slots in a row, the fan profile is aggressive so yes it gets loud, but I have them further away from me and I wear headphones, I don't think people who are used to silence could cope, but the environment around me isn't silent and the fans are a very constant noise that ends up as background. Even with a less aggressive fan profile you'd still see a good boost and keep the noise more palatable, I'd say it's more important to have good air flow in the case and water would be ideal.

cheers
brasc

ok, yeah, i'll probably just do water then. My computer is in my bedroom (the joys of NYC living ) so I definitely can't have it being super loud. Right now I have 2x 780 with those evga acx coolers on them so i'm accustomed to a pretty quiet setup.

any benefit over doing hydro copper vs installing the water block myself, other than ease of setup?

Hah, yeah, it's probably not a bedroom setup

@smicha would be the man to talk to about blocks, but I imagine the Hydro will come at a serious premium over adding your own EK block. Under water you could run it quieter and still get an excellent boost. It's on my long term plan for sure.

cheers
brasc