• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

I've Made A New, Heavy, Rendering Benchmark. Looking For Testers!

Status
Not open for further replies.
K

kweechy

n00b
Joined
Mar 17, 2012
Messages
21
It is essentially just a project folder for Autodesk Maya, but since they have 30 day free trials available, this benchmark is free to run. Here is the image that will be produced by your system (the bottle is from a commercial gig, just threw it in since I had it kicking around):



The idea is that this is a benchmark for your entire system and will stress all the different computational functions you see in a real world workload. Some single threaded ops, some hard drive access, RAM access and best of all (in my opinion) it takes approximately 1hr to render the frame on my 4.8GHz 2600K...this removes a MASSIVE amount of margin of error present in other benchmarks that are way too short to give a real picture of how well your system performs.

I'd love to get some people testing with various systems to really get a sense of what impact the RAM has on the final speeds, but also very interested to see various procs and architectures. Cinebench isn't affected by RAM much, but Cinebench also has the ability to more or less sit in L3 without much back and forth through the IMC.

Who's interested?

Dropbox links to the Maya project and scene file below:

Project Folder + Textures

Instructions:

1) Install Maya and launch the app.
2) File -> Set Project, navigate to this project folder you downloaded + extracted. You want to set "lib" as your project. If a message pops up asking if you want to create a default workspace, you picked the wrong folder. Hit cancel and retry.
3) Once the project is set, File -> Open and pick the file inside maya/scenes/lighting.
4) In the top row of icons in the UI, towards the far right hand side, there's a director's clapboard icon that, on mouseover, says "Render the current frame (mental ray)". If the render window is already open, skip to 6.
5) Click the button to the left of that one, it will launch the Render Window. You can resize this to your liking.
5) In the render window, click the clapboard button at the top left and watch the pretty picture get made. Maya will note the render duration upon completion at the bottom of that window.

Please let me know if you give this a shot. Very interested to see results. Here is my spreadsheet thusfar...I've normalized the scores so they resolve very similar to what a system's CineBench score is.

https://docs.google.com/spreadsheet...VkJMVDJZQWpSVkE&single=true&gid=0&output=html

I will dig up info for a Maya Linux trial as well since most of you are on Lin.

Edit...here's the Linux installer:

http://images.autodesk.com/adsk/file...nux_64bit2.tgz

In addition, here is a guide to help with the install process for folks using Debian / Ubuntu:

http://etoia.free.fr/?p=1611

It's not for the feint of heart sadly. I wish I had an easy way to let us benchmark all systems on all OS.
 
Last edited:
since i'm bored i'll try this just to see how bloody slow my system really is.

edit.. doing the rendering now.. holy crap i don't recommend doing this on a phenom II x4.. its sssslllllllloooooowwwwww its about 1/4 the way done and about 2 minutes in so far..
 
Last edited:
I'd be curious to frogstomp this with my quad 6172, but need linux version and idiot proof instructions to proceed.
 
well i gave up, it takes way to long.. i let it run for 40 minutes and it wasn't even half done. maybe if i'm off doing something else for a few hours i'll try it again but my ADD and the fact that my system is almost completely unusable during the rendering is to much for me. :D

though the one thing i did notice is that maya is detecting my processor as having 8 threads when i start the rendering even though it obviously doesn't. not sure if thats screwing it up or not.
 
If you want to try it out on some of the more awesome systems here you will have to figure out a way to render it on Linux, preferably from the terminal. That's one huge benchmark, it would probably take 5 hours on my daily driver laptop.
 
I'll run it on my SR2 and my wife's laptop, just for kicks. Post the results back later. Just out of curiosity, how much ram does this program use? I'm curious to see if I can render the image completely on a ram disk.
 
freeloader1969 said:
I'll run it on my SR2 and my wife's laptop, just for kicks. Post the results back later. Just out of curiosity, how much ram does this program use? I'm curious to see if I can render the image completely on a ram disk.
Click to expand...

the highest i saw on 32bit was ~1GB even though maya was only reportedly using 210mb didn't say what was using the extra ram in my process list though so i don't have an exact number on how much it used for me.. over all my total usage was 2.39GB while rendering idle ram usage is ~1.1GB(probably less since i closed chrome before doing the rendering). the program its self takes up 2.1GB.
 
This benchmark will take a long time on older hardware for sure, especially something like a Phenom X4 or a Core2Quad. I designed the test with the more bleeding edge, high powered stuff in mind...figuring that a benchmark that takes even a 4.8GHz 2600K an hour to complete should be future proof for a long time. The downside of course is that it would definitely take 5+ hours on a laptop or systems of that nature.

I also wanted long crunch times to avoid people running this with unstable voltage levels to produce unrealistic numbers. If your machine can't sustain its OC for an hour at 100% load, it doesn't really deserve to be setting benchmark numbers.

If you really wanted to, you can do this though...in the main Maya menu bars, underneath File, Edit, etc., there's a drop down box. Make sure it's set to "Rendering". Once that's done, you'll see a Render menu in the top along with File, Edit, etc. Click that and go to Test Resolution. You can then just set that to 50%, 25%, whatever you want to really. The results aren't as simple as multiplying your 50% render time by 4 to get your 100% render time, but it's close enough to give you an idea.

Here is a link to a Linux install for Maya:

http://images.autodesk.com/adsk/files/autodesk_maya_2012_english_linux_64bit2.tgz

In addition, here is a guide to help with the install process for folks using Debian / Ubuntu:

http://etoia.free.fr/?p=1611

It's not for the feint of heart sadly. It's a shame it's so difficult to benchmark Linux systems with the popular suites.

Folks with 4P systems could, in theory, grab a trial version of Windows Server or the Windows 8 Server Beta, then install Maya in Windows and be off and running.
 
Oh and this (unfortunately) uses small enough RAM that you could definitely throw the project folder onto a RAMdisk partition and render to and from there. Maya doesn't actually save out an image until you say so, and it will pull all the source images into RAM immediately as the rendering starts, but that should go a bit faster if it's reading from a RAMdisk.

I'm VERY curious to see what impact RAM speeds have on this benchmark though if anyone out there has access to a 4.8GHz 2600K with 2133+ memory.

Ideally, I'd be able to get us one of our studio's scenes from TRON or RoboCop or something that would eat up a solid 12GB+ mem...sadly those assets are locked down like Fort Knox.

I feel like not enough benchmarks test RAM and CPU simultaneously though. It's all extremely light RAM footprint, CPU intensive stuff. That tells you nothing about what your system does to real workloads.


sirmonkey1985 said:
well i gave up, it takes way to long.. i let it run for 40 minutes and it wasn't even half done. maybe if i'm off doing something else for a few hours i'll try it again but my ADD and the fact that my system is almost completely unusable during the rendering is to much for me. :D

though the one thing i did notice is that maya is detecting my processor as having 8 threads when i start the rendering even though it obviously doesn't. not sure if thats screwing it up or not.
Click to expand...

Totally understandable. Where are you seeing Maya detecting 8 threads btw? I'm guessing in the Output Window? If that's where you're seeing it, I think Mental Ray always says "mental ray: got 8 satellite CPUs." It does that on all my 3930Ks and 980X as well. Once it's rendering time though, all 12 threads light up.


pjkenned said:
I can probably get this on the dual Xeon E5-2690 system.
Click to expand...

I'll wager a guess here and say it would render the scene in roughly 24:22 from distilling all the information I've got about the 2690s. Absolute travesty those chips can't be overclocked...could easily render this scene in under 19 minutes.
 
Last edited:
Wish you would of had this out a few weeks ago when I was setting up new servers. I just put a couple Dual X5690 boxes into service.
 
A dual X5690 would probably score around 16.5 or so, which would mean a render time of ~35:20 or so.
 
Hmmm.... 5:11:57 on a i7-3930k @ 4.5GHz render happened at 2560x1440 though.
 
pjkenned said:
Hmmm.... 5:11:57 on a i7-3930k @ 4.5GHz render happened at 2560x1440 though.
Click to expand...

Yeah that's a rather huge resolution to render this image at. Probably looked nice though!

I have the scene defaulting to 1024 x 576 to keep render times down to around an hour on my 2600K, any chance you'd be willing to crunch it at that default resolution?


sirmonkey1985 said:
wow if it took you 5 hours i'm definitely not trying it again, lol.
Click to expand...

Nah it's only an hour on my quad core with a CB score of 9.4, a machine with a CB score of 4.8 should be capable of posting a 2hr score.
 
Last edited:
kweechy said:
Yeah that's a rather huge resolution to render this image at. Probably looked nice though!

I have the scene defaulting to 1024 x 576 to keep render times down to around an hour on my 2600K, any chance you'd be willing to crunch it at that default resolution?




Nah it's only an hour on my quad core with a CB score of 9.4, a machine with a CB score of 4.8 should be capable of posting a 2hr score.
Click to expand...

i did it for 40 minutes, the furthest it got was the green background on the top left corner and right before i stopped it 1 of the squares with some rocks finally finished. i'm pretty sure it'll take me a lot longer then an hour. but now that i think about it i did have the render window full screen.
 
59:56 at 1024x576 on a 3.5GHz i7-3930K... not sure why it is clocked that low.
 
pjkenned said:
59:56 at 1024x576 on a 3.5GHz i7-3930K... not sure why it is clocked that low.
Click to expand...

What memory speed was that with? The render time sounds pretty much right, almost identical to a 4.8GHz 2600K. 37% higher clock speed plus ~5% due to fewer inefficiencies from multi threading versus 50% more cores.

I'm guessing your RAM is 1600MHz since you scored almost exactly 20% slower than my 4.2GHz 3930K (20% higher clock).
 
Yes it was at 1600MHz.

When I put windows on the 2P E5-2690 I will give you an update. Super busy week here.
 
sirmonkey1985 said:
i did it for 40 minutes, the furthest it got was the green background on the top left corner and right before i stopped it 1 of the squares with some rocks finally finished. i'm pretty sure it'll take me a lot longer then an hour. but now that i think about it i did have the render window full screen.
Click to expand...

Definitely going to want to just try out the regular 1024 x 576 render.

One of my OC'd 3930K rigs took over 3 hours to churn out a 2048 x 1152 render...that's a pretty unfriendly test for Phenom X4 chips.

Interestingly and sadly enough...that still doesn't even begin to approach some of the actual frame times we see down at the office. However we're talking about scenes with upwards of 1 billion particles, area lights with soft shadows, bounce light, reflections and refractions...crazy stuff.

pjkenned said:
Yes it was at 1600MHz.

When I put windows on the 2P E5-2690 I will give you an update. Super busy week here.
Click to expand...

Ah cool, pretty neat that the RAM speed was guessable just by the render times! That indicates a pretty solid benching test to me.

No rush at all on the testing, I just find this data cool to have as there seems to be very few bench suites that run for such a long time on workloads that are literally real world loads.
 
kweechy said:
Definitely going to want to just try out the regular 1024 x 576 render.

One of my OC'd 3930K rigs took over 3 hours to churn out a 2048 x 1152 render...that's a pretty unfriendly test for Phenom X4 chips.

Interestingly and sadly enough...that still doesn't even begin to approach some of the actual frame times we see down at the office. However we're talking about scenes with upwards of 1 billion particles, area lights with soft shadows, bounce light, reflections and refractions...crazy stuff.
Click to expand...


yeah tried it again at the normal render window size, got further this time(30% or at least thats what maya said) but after an hour my ADD took over and i couldn't take sitting and staring at it rendering anymore since it makes my system completely unusable during the rendering. i think the problem is my DDR2 800 ram is just to damn slow.
 
Status
Not open for further replies.
Back
Top