Overclock.net › Forums › Benchmarks › Benchmarking Software and Discussion › Best Benching Software for 3D Rendering?
New Posts  All Forums:Forum Nav:

Best Benching Software for 3D Rendering?

post #1 of 10
Thread Starter 
What is the best benchmarking tool that will unanimously determine how quickly computer 1 can render a file from something like RenderMan/VRay/MentalRay/Mantra/Etc versus computer 2?

The game benchmarks are all well and good...but the problem is that A) there is such thing as 'fast enough' when it comes to games if you can keep decent framerates with maxed settings and B) games aren't necessarily using your processor and machine in the same way as rendering software does.

My computers have one sole function in life: to render my 3D work. In feature film, there is no such thing as a frame that renders quicker than 1-2 hours (in fact, 1-2 hours is extremely quick). As such, there is no such thing as EVER having fast enough procs or computing power.

I'm guessing CineBench would be the best bet in terms of figuring out what computer setup is the single best renderer of files?

Are there any benchmarks that allow multiple computers to do a distributed render? Seems these days it would be much more cost effective to grab a couple SB consumer rigs and tie them together than do a dual Xeon build...but there's no real tests for that.
2P Workstation
(13 items)
 
  
CPUCPUMotherboardGraphics
Xeon E5 2699 v3 Xeon E5 2699 v3 ASUS Z10PE-D16 EVGA TITAN X 
RAMHard DriveHard DriveCooling
128GB Crucial DDR4 @ 2133MHz Intel 750 Series 1.2TB 4 x 4TB Constellation RAID5 2 x XSPC Raystorms 
CoolingCoolingMonitorMonitor
RS480 MCP655 Dell U3011 Dell U3011 
Case
Silverstone TJ07 
  hide details  
Reply
2P Workstation
(13 items)
 
  
CPUCPUMotherboardGraphics
Xeon E5 2699 v3 Xeon E5 2699 v3 ASUS Z10PE-D16 EVGA TITAN X 
RAMHard DriveHard DriveCooling
128GB Crucial DDR4 @ 2133MHz Intel 750 Series 1.2TB 4 x 4TB Constellation RAID5 2 x XSPC Raystorms 
CoolingCoolingMonitorMonitor
RS480 MCP655 Dell U3011 Dell U3011 
Case
Silverstone TJ07 
  hide details  
Reply
post #2 of 10
The only thing I know of is Cinebench for professional rendering and I'm not sure if you can spread the workload across multiple systems... And yes, I think it would be cheaper to get a couple 2600k CPUs, a cheap p67, and have a headless render farm.
Po' Pimpin'
(11 items)
 
  
CPUMotherboardRAMHard Drive
i5 2500k @ stock Biostar TZ68K+ [A3] 4GB  Sandforce 1222 64GB SSD 
Optical DriveCoolingOSMonitor
LG 22x DVD-+RW  Stock Windows 7 x64 Acer S211HL 1080p 
PowerCaseMouse
600w Diablotek Linkworld Electronic Inland 
  hide details  
Reply
Po' Pimpin'
(11 items)
 
  
CPUMotherboardRAMHard Drive
i5 2500k @ stock Biostar TZ68K+ [A3] 4GB  Sandforce 1222 64GB SSD 
Optical DriveCoolingOSMonitor
LG 22x DVD-+RW  Stock Windows 7 x64 Acer S211HL 1080p 
PowerCaseMouse
600w Diablotek Linkworld Electronic Inland 
  hide details  
Reply
post #3 of 10
Thread Starter 
Yeah that's what I've got set up right now. My main machine for working and 4 slaves with these specs:

ASUS P8P67
16GB DDR3 1600mhz
i7 2600k @ 4.9ghz
Corsair H70

Working in feature film as a lighter, I'm alloted 2 dual Xeon render blades from the farm to do distributed rendering during the day in addition to my dual Xeon workstation...honestly I think my home setup renders things faster, though of course I'm not allowed to take the files home to do a head to head comparison

It would be awesome if a benchmark software existed that allowed you to see how powerful your network or machine farm is. I can't imagine going back to owning a dual Xeon at home with how powerful and cheap it is to make a farm with consumer chips.
2P Workstation
(13 items)
 
  
CPUCPUMotherboardGraphics
Xeon E5 2699 v3 Xeon E5 2699 v3 ASUS Z10PE-D16 EVGA TITAN X 
RAMHard DriveHard DriveCooling
128GB Crucial DDR4 @ 2133MHz Intel 750 Series 1.2TB 4 x 4TB Constellation RAID5 2 x XSPC Raystorms 
CoolingCoolingMonitorMonitor
RS480 MCP655 Dell U3011 Dell U3011 
Case
Silverstone TJ07 
  hide details  
Reply
2P Workstation
(13 items)
 
  
CPUCPUMotherboardGraphics
Xeon E5 2699 v3 Xeon E5 2699 v3 ASUS Z10PE-D16 EVGA TITAN X 
RAMHard DriveHard DriveCooling
128GB Crucial DDR4 @ 2133MHz Intel 750 Series 1.2TB 4 x 4TB Constellation RAID5 2 x XSPC Raystorms 
CoolingCoolingMonitorMonitor
RS480 MCP655 Dell U3011 Dell U3011 
Case
Silverstone TJ07 
  hide details  
Reply
post #4 of 10
Lol! I never thought that a network rendering benchmark exists
But i was WRONG:
http://www.luxrender.net/wiki/Tests_...k_rendering.29
This is for linux systems only.
But your 3D application of choice should provide network rendering.
If so you can always take a test scene and benchmark it on a single system and compare it to the network renderer.
You just need ato create test file.
In Cinebench the test file can be found in cinebench\\plugins\\bench\\cpu .
There is the cpu.c4d file that is used for the cpu test render.
You can import this file in c4d and do a network renderer and compare it to the single system.
IMHO this makes more sense that any 3Dmark stuff or similar because you want to know how fast your 3D application is not something other.
I hope i explained it well enough. If not forgive me
/* Redemption*/
(14 items)
 
  
CPUMotherboardGraphicsRAM
I7 3930K Asus Sabertooth Asus GTX 680 8x4GB G.Skill@1337MHz 
Hard DriveOptical DriveCoolingOS
2xM4 64GB/ / F3 - 1TB / 2x2TB Baracudas some LG Modified EK 360 HFX 2x(Win7 x64) 
MonitorKeyboardPowerCase
SyncMaster P2770HD and SyncMaster 940NW Roccat Isku Corsair Gold AX750 NZXT 810 Switch 
MouseMouse Pad
Rocat Kone[+] Razer exactmat X 
  hide details  
Reply
/* Redemption*/
(14 items)
 
  
CPUMotherboardGraphicsRAM
I7 3930K Asus Sabertooth Asus GTX 680 8x4GB G.Skill@1337MHz 
Hard DriveOptical DriveCoolingOS
2xM4 64GB/ / F3 - 1TB / 2x2TB Baracudas some LG Modified EK 360 HFX 2x(Win7 x64) 
MonitorKeyboardPowerCase
SyncMaster P2770HD and SyncMaster 940NW Roccat Isku Corsair Gold AX750 NZXT 810 Switch 
MouseMouse Pad
Rocat Kone[+] Razer exactmat X 
  hide details  
Reply
post #5 of 10
Thread Starter 
Makes perfect sense. Only real problem is that you'd need to have C4D installed or other dependencies...also running the CineBench scene in C4D doesn't guarantee as standardized results as running it in their standalone application.

In terms of real, hard evidence of network rendering speed...all I can offer is this test I did on the same frame from the same file:

Test 1; Main PC only (i7 975x @ 4.2ghz)
Rendertime: 5:57

Test 2; Main PC + Slave (i7 2600k @ 5.1ghz)
Rendertime: 2:27

I haven't built my other two render slaves yet, so don't have measurements for that. At some point I will hit a bottleneck where adding slaves doesn't help at all due to network jamming plus the fact that my main PC is trying to render and send/recieve files from the slaves all at once.

Once that happens, I'd need a dedicated server to host all the files, send out render requests and texture loads and assemble the distributed product. Until I hit that though, I guarantee this is the single best and cheapest way to build the fastest possible system for rendering.

A dual Xeon @ $5-6k doesn't benchmark much higher than a single i7 2600k that's been OC'd and costs $1.1k. Benching software just doesn't really allow you to compare the two head on.
2P Workstation
(13 items)
 
  
CPUCPUMotherboardGraphics
Xeon E5 2699 v3 Xeon E5 2699 v3 ASUS Z10PE-D16 EVGA TITAN X 
RAMHard DriveHard DriveCooling
128GB Crucial DDR4 @ 2133MHz Intel 750 Series 1.2TB 4 x 4TB Constellation RAID5 2 x XSPC Raystorms 
CoolingCoolingMonitorMonitor
RS480 MCP655 Dell U3011 Dell U3011 
Case
Silverstone TJ07 
  hide details  
Reply
2P Workstation
(13 items)
 
  
CPUCPUMotherboardGraphics
Xeon E5 2699 v3 Xeon E5 2699 v3 ASUS Z10PE-D16 EVGA TITAN X 
RAMHard DriveHard DriveCooling
128GB Crucial DDR4 @ 2133MHz Intel 750 Series 1.2TB 4 x 4TB Constellation RAID5 2 x XSPC Raystorms 
CoolingCoolingMonitorMonitor
RS480 MCP655 Dell U3011 Dell U3011 
Case
Silverstone TJ07 
  hide details  
Reply
post #6 of 10
Quote:
Originally Posted by kweechy View Post
Test 1; Main PC only (i7 975x @ 4.2ghz)
Rendertime: 5:57

Test 2; Main PC + Slave (i7 2600k @ 5.1ghz)
Rendertime: 2:27
Again you are still comparing a quad core to a quad core. As you know the overclocked cpu will beat out the slower cpu. As the 2600k's have newer instructions and are faster clock for clock.

Now if you have something that uses more then 4 cores. Then you will see a huge difference between a 2600k and a say a 980x. As once the other 2 cores kick in the 2600k can't compare then. Then you look at time saved, over a few days. weeks or a year and it adds up.

Though, I agree a 2500/2600k is a great chip. but it just depends on if your software will take full advantage of the chips capability. Have fun and good luck.
post #7 of 10
Thread Starter 
No that result is with distributed rendering; Main PC + Slave on the same job. Not comparing Main PC to Slave.

The main problem with everything relating to benchmarks is they are designed for single systems. I do not run a dual Xeon anymore (because I realized that the price just isn't worth it) and run a single gaming rig with powerful slave machines; all using consumer level parts rather than workstation parts.

Benchmarking single systems exclusively will not allow you to understand the price vs performance you can obtain with multiple, cheaper systems.

A dual 6 core Xeon machine will render a scene faster than a single 2600k, but for the price, I can assemble 4 2600k machines. When networked together, they will put the Xeons to shame.

That's the real thing here.

Ideally you make a benchmark program that can run distributed. That way you can benchmark your network and not just single machines in it. Sure, a quad Opteron system gets 27pts for CPU in CineBench 11.5 and my 2600k only gets 10pts...but how would the results look if CineBench would allow me to use all 4 of my 2600k slaves + my 975x main PC at once?
Edited by kweechy - 4/27/11 at 2:14pm
2P Workstation
(13 items)
 
  
CPUCPUMotherboardGraphics
Xeon E5 2699 v3 Xeon E5 2699 v3 ASUS Z10PE-D16 EVGA TITAN X 
RAMHard DriveHard DriveCooling
128GB Crucial DDR4 @ 2133MHz Intel 750 Series 1.2TB 4 x 4TB Constellation RAID5 2 x XSPC Raystorms 
CoolingCoolingMonitorMonitor
RS480 MCP655 Dell U3011 Dell U3011 
Case
Silverstone TJ07 
  hide details  
Reply
2P Workstation
(13 items)
 
  
CPUCPUMotherboardGraphics
Xeon E5 2699 v3 Xeon E5 2699 v3 ASUS Z10PE-D16 EVGA TITAN X 
RAMHard DriveHard DriveCooling
128GB Crucial DDR4 @ 2133MHz Intel 750 Series 1.2TB 4 x 4TB Constellation RAID5 2 x XSPC Raystorms 
CoolingCoolingMonitorMonitor
RS480 MCP655 Dell U3011 Dell U3011 
Case
Silverstone TJ07 
  hide details  
Reply
post #8 of 10
Sorry, I missed the "+ slave" part. Do you have the means to do a 975x + 975x slaved -vs- 2600K + 2600k slaved. As i like to see a same type deal on both and results.

But for the value i grab some 2600k's and do it that way. Since you can grab 3 of those for one 980x price.
Edited by KILLER_K - 4/27/11 at 6:35pm
post #9 of 10
The c4d-file was just an example.
It sounded that you have some 3D applikation.
Cause if you had one you could do the test scene standalone renderer vs the net.
And nearly all 3D apps have network renderer nowadays
Edited by Mr.Eiht - 4/27/11 at 6:29pm
/* Redemption*/
(14 items)
 
  
CPUMotherboardGraphicsRAM
I7 3930K Asus Sabertooth Asus GTX 680 8x4GB G.Skill@1337MHz 
Hard DriveOptical DriveCoolingOS
2xM4 64GB/ / F3 - 1TB / 2x2TB Baracudas some LG Modified EK 360 HFX 2x(Win7 x64) 
MonitorKeyboardPowerCase
SyncMaster P2770HD and SyncMaster 940NW Roccat Isku Corsair Gold AX750 NZXT 810 Switch 
MouseMouse Pad
Rocat Kone[+] Razer exactmat X 
  hide details  
Reply
/* Redemption*/
(14 items)
 
  
CPUMotherboardGraphicsRAM
I7 3930K Asus Sabertooth Asus GTX 680 8x4GB G.Skill@1337MHz 
Hard DriveOptical DriveCoolingOS
2xM4 64GB/ / F3 - 1TB / 2x2TB Baracudas some LG Modified EK 360 HFX 2x(Win7 x64) 
MonitorKeyboardPowerCase
SyncMaster P2770HD and SyncMaster 940NW Roccat Isku Corsair Gold AX750 NZXT 810 Switch 
MouseMouse Pad
Rocat Kone[+] Razer exactmat X 
  hide details  
Reply
post #10 of 10
Thread Starter 
I guess I was just thinking it would be cool to somehow get my 'equivalent' Cinebench 11.5 score with Main PC and 4 Slaves pumping away...really see how good it is against an OC'd dual 6C Xeon machine or quad AMD Opteron.

@ KillerK, I can't really do those comparisons too well for a couple reasons...one is that the Slaves don't have any 3D apps actually installed, only rendering standalone software and the main reason is that I don't have a 2nd i7 975 computer

If it helps though, head to head, the i7 2600k Slave machine will render a file roughly 2x faster than the i7 975x machine...which actually speaks volumes for PassMark's software accuracy. My 975 scores around 7,920 and my 2600k scores around 13,800...close to 2x higher.

For lols I looked up the PassMark score for my first Xeon workstation that I ever owned...it got a score of 272. That wasn't even 10 years ago. Orders of magnitude much?
2P Workstation
(13 items)
 
  
CPUCPUMotherboardGraphics
Xeon E5 2699 v3 Xeon E5 2699 v3 ASUS Z10PE-D16 EVGA TITAN X 
RAMHard DriveHard DriveCooling
128GB Crucial DDR4 @ 2133MHz Intel 750 Series 1.2TB 4 x 4TB Constellation RAID5 2 x XSPC Raystorms 
CoolingCoolingMonitorMonitor
RS480 MCP655 Dell U3011 Dell U3011 
Case
Silverstone TJ07 
  hide details  
Reply
2P Workstation
(13 items)
 
  
CPUCPUMotherboardGraphics
Xeon E5 2699 v3 Xeon E5 2699 v3 ASUS Z10PE-D16 EVGA TITAN X 
RAMHard DriveHard DriveCooling
128GB Crucial DDR4 @ 2133MHz Intel 750 Series 1.2TB 4 x 4TB Constellation RAID5 2 x XSPC Raystorms 
CoolingCoolingMonitorMonitor
RS480 MCP655 Dell U3011 Dell U3011 
Case
Silverstone TJ07 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
Overclock.net › Forums › Benchmarks › Benchmarking Software and Discussion › Best Benching Software for 3D Rendering?