Page 16 of 62
Re: OCTANE BENCHMARK: Submit performance scores for your rig
Posted: Thu Jan 31, 2013 9:16 am
by whersmy
@ Glimpse: Yes, a iGPU would be perfect. I think I`ll wait till haswell, then the 1155 with iGPU`s will be much cheaper to buy.
In the meantime I`ll stick with 3x 580s...
@above: High temps are common on 580s, I use afterburner to increase the fanspeed. Gets loud....but temps drop from 95C to 75C (with a 4 extra casefans)
Cheers
Re: OCTANE BENCHMARK: Submit performance scores for your rig
Posted: Sat Feb 02, 2013 1:16 am
by FrankPooleFloating
I think we may have possibly taken a step backward with v1.05....
[2] EVGA GTX460's
v1.02 = 4.00 M/s
v1.05 =
3.58 M/s
Can anyone else confirm somewhere around a 10% decrease in performance?
And are we supposed to be using that same benchmark scene (which I think is from 2010)?
(yes - I do understand that changing the scene would defeat (at least one) purpose... I just
want to be sure I am using the right one nonetheless)

Re: OCTANE BENCHMARK: Submit performance scores for your rig
Posted: Sat Feb 02, 2013 7:56 am
by cfrank78
Here too!
About 10% decrease from 1.02 to 1.05.
tested with standart benchmark scene!
So if you compare 2.58 and now 1.05 it is about 35% decrease of speed to 2.58.....
35% slower but more features. Seems like Octane goes in the wrong direction in my opinion!
Chris!
Re: OCTANE BENCHMARK: Submit performance scores for your rig
Posted: Sat Feb 02, 2013 12:19 pm
by Refracty
When features are (allready) introduced that minimize the noise (e.g. importance sampling) then it compensates the decrease in sampling speed.
Re: OCTANE BENCHMARK: Submit performance scores for your rig
Posted: Sat Feb 02, 2013 1:39 pm
by cfrank78
Refracty.....no!
Importance sampling is, in my opinion better sold than it really is. If i make a scene without importance sampling i have about 16 ms/s. with importance sampling turned on i get in the exactly same scene about 9 ms/s. So....Otoy sells importance sampling, for example the noise free Head model with lower count of frames to get it noise free than without, what is completeley correct - BUT these samples, even if they are much fewer, take much longer to render than without importance sampling turned on. I have made some tests and under the line - importance sampling IS faster but only about 9%. Not as nearly as THAT fast or as great, as all are talking about!
So you get 9% more and ONLY if you use a HDRI. Well. In comparison to other versions, you are still about 26% slower and that ONLY if you have a scene where you can use importance sampling. Otherwise it is 35% slower.
Under the line.......Octane/Otoy implements new features, makes Octane slower but still the BASIC things like working fast with keyboard shortcuts, applying materials fast with mouse klicks, UV texturing, ctrl z, and so many more are not implemented. New features and working slow....Is that the solution for a render engine that is sold as THE fastest?
Dont get me wrong. We all like the new features, but octane is not anymore THE fastest now. I dont wanna tell any names here, but i beta tested 2 other engines that will come online in the next few days/weeks/months and they will make octane look slow, cause they are even faster than octane was earlier days....and......have all features that octane has and many, many, many, more like for example great and simple to use UV mapping tools IN the render program, applying materials extremely fast with ALT right mouseclick, move-import-scale-show objects on/off from other scenes or 3d applications, obj export, live plugins, dirt, 10 different channels to output(zdepth ....), fog, instancing, motion blur, predefined and offline available great looking materials, handling natively exported camera paths and lights and materials from other 3d applications, (nearly) realtime sss with realtime DOF, perfectly in rhino integrated with a 1 klick plugin, and and and. Many, many more.
But to be honest they will also cost A LOT more. So....for its price, Octane is still a winner.
Chris!
Re: OCTANE BENCHMARK: Submit performance scores for your rig
Posted: Sat Feb 02, 2013 3:37 pm
by r-username
I am looking to pick up 2 new cards to build out my system, while I it may be an easy choice I always feel like I'm missing something and would greatly appreciate your opinions.
Current:
(1) 1920x1080 monitor, Sabertooth x58, 1200 PSU, evga Gtx 590
The first card is just for the monitor, but I'm unsure wether to get a 500 or 600 series card and the best value currently out there. The second is another rendering card, again the 5 or 6 series question, I can pick up both for almost the same price, $999. for 590 or 690, and 450ish for for 580 or 680.
Thanks in advance
Re: OCTANE BENCHMARK: Submit performance scores for your rig
Posted: Sun Feb 03, 2013 12:09 am
by buzby
whersmy wrote:How did you stuff a z77 asrock 9 with 4 duals slots 580 gpus?

Its a beautiful thing

Our first attempt at an industrial type rendering machine.
Tried a couple different Mobos, 1250 W power supplies, 1600 W PSU backup, and varieties of GTX 580 graphics cards but finally settled on the classified GTX 580 set up with it's overclocked 855 core and 3GB memory for the scenes. Ended up trading out the case to a CoolerMaster Storm Trooper as well to get everything flowing right.
I gotta say this machine brings satisfaction
Re: OCTANE BENCHMARK: Submit performance scores for your rig
Posted: Sun Feb 03, 2013 12:19 am
by buzby
rappet wrote:buzby wrote:Finally got the 4 classified in the same unit and not only did it increase my benchmark to 15.39 but they run cool from 59-65C
Hi buzby,
I also got 4 classifieds like you. I got 16.72 Ms/sec (4x580, OR 2.57, driver 285.62) on the benchmark,
But my temps are much higher 75-85 with aircooling using appr 10 fans. I am curious how your cooling is done?
Greetz,
Arggh, I was happy now I am jealous...
I am sure you chose the cards for the same reason I did as they blow the hot air out the back of the case. I am only using air cooling but I think a big factor is the CoolerMaster Storm Trooper case. It really handles the air nicely.
You did not mention if you were overclocking but I am sure you must not be if you are surprised you are running moderately high temps. I only dabbled in overclocking these and was running 69-72 tops. What are your core and shader settings?
Re: OCTANE BENCHMARK: Submit performance scores for your rig
Posted: Sun Feb 03, 2013 8:05 am
by enthewhite
Hey cfrank, would you mind sharing the names of those beta programs you tested? I am curious to know about them.
Re: OCTANE BENCHMARK: Submit performance scores for your rig
Posted: Sun Feb 03, 2013 11:55 am
by cfrank78
Hey enthewhite!
I am not allowed to and to be honest, i don“t want to! You will see it soon. Just watch the web

. I am not here to make advertisement for other render engines. Thats not fair.
When i came into the jewelry business and i made my apprentice ship, my master allways teached me one special thing! He allways said and told me: No matter how complex or how difficult your handmade jewelry is, your goal allways has to be time and quality. You allways have to be able to produce the best jewelry in the shortest time but WITHOUT loosing quality. When you are not able to do this, you can not be a master in your business. It is as simple as that! And he was right!
I just wanted to say, that from my point of view, octane goes at the moment in the wrong direction. But thats only my opinion!
Cheers - Chris!