How is this computer so much faster?...

[FenixFan08]

Alright, I have 3 computers. Computer A, B, and C, I will call them.

I am rendering an animation using Luxology Modo. They are all rendering the same frame. But Computer A is doing it in a hour less than the others. Can anyone tell me why? I have tried many things and can not figure this out. They are all being rendered on there main hds too. Nothing is being rendered over a network.

Computer A Specs - 51 min to render frame

-i7 3930K @ 3.2Ghz but turbos to 3.8 Ghz (6 Core)
-16 gb of ram
-SSD

Computer B Specs - 2 hr 18 min to render frame (4 Core)

-Xeon E3 1230 @ 3.2 Ghz
-16 gb of Ram
-10000 VR hdd

Computer C Specs - 1 hr 45 min to render frame

-Phenom II x6 1090t @ 3.2 Ghz Overclocked to 4.0 Ghz (6 Core)
- 16 gb of ram
-10000 rpm VR

I know these are huge frame times, but that isn't my question. I am wondering why is Computer A rendering 1 hour faster per frame then lets say Computer C that is OCed to 4.0 ghz.

My thoughts possibly, is the SSD making that huge of a difference?

Is the hyperthreading on the i7 really that huge of a difference compared to the OC of the 1090t?

I will take any thoughts on why this is into consideration and any help is very appreciated.

Thanks,

Fenix

 

[OS-Wiz]

Some thoughts, but a question first. Can Luxology Modo exploit all six cores of the 3930K? In other words, what's the maximum number of threads it can spawn in parallel?

The 3930K has the newest architecture and probably has a few extra opcodes that lend themselves well to rendering.

Writing to the SSD will be at least 10x faster than writing to an HDD, even your VelociRaptor.

 

[Hameister]

Some thoughts, but a question first. Can Luxology Modo exploit all six cores of the 3930K? In other words, what's the maximum number of threads it can spawn in parallel?

The 3930K has the newest architecture and probably has a few extra opcodes that lend themselves well to rendering.

Writing to the SSD will be at least 10x faster than writing to an HDD, even your VelociRaptor.

Aren't you happy you listened to me and bought the 3930k? Huh, huh, aren't ya?

"modo can be used on up to 10 Mac or PC workstations for network rendering; it is multi-threaded to use all available cores on each of your systems. Setup is extremely easy and is largely automatic; you can schedule rendering clients to render scenes across named groups of systems in your workgroup for maximum throughput."

Luxology.com :: modo Rendering

 

[FenixFan08]

Yeah I know it does use them all, so are you saying that 3930k is THAT much faster? I have read the SSDs dont make that much of a difference when it comes to rendering, but how much do you think that is better than a standard 10000 VR?

 

[OS-Wiz]

Yeah I know it does use them all, so are you saying that 3930k is THAT much faster? I have read the SSDs dont make that much of a difference when it comes to rendering, but how much do you think that is better than a standard 10000 VR?

There's a lot of "it depends" involved here. But yes, the 3930's architecture, being the newest design of the CPUs under study, will be faster in general. It has a more efficient method of handling the opcodes, thus more instructions per second will be completed. And it has on processor cache memory, L1, 2, and 3 cache that is larger and better managed. Also, if the rendering process is memory intensive (I don't know) the 3930 has four memory channels which provides for a higher bandwidth (meaning it can move more data in and out memory) than the other systems under study. The list goes on ... As far as I/O goes, I can only answer from a commercial, large system, standpoint as that's the only time I've had the professional tools it takes to measure the life-cycle of a process.
In general, the I/O portion of a process is anywhere from 10 - 30%, given how intense the I/O for that process is, e.g. a write intense process, like an large insert or load of data into a database will be near the 30% mark. However, I've never measured a rendering process, so again, I can only give a very general answer. And, as I've said before, writes to an SSD will be at least 10x faster than writes to an HDD given the disk or disks under study were dedicated to that process.

 

[OS-Wiz]

Aren't you happy you listened to me and bought the 3930k? Huh, huh, aren't ya?

"modo can be used on up to 10 Mac or PC workstations for network rendering; it is multi-threaded to use all available cores on each of your systems. Setup is extremely easy and is largely automatic; you can schedule rendering clients to render scenes across named groups of systems in your workgroup for maximum throughput."

Luxology.com :: modo Rendering

Thanks for that info Ham. And yes, very happy!

 

[Hameister]

I tell ya Wizzer, these 3930k processors are so strong, I'm really having second thoughts about the IB-E upgrade next year.

We haven't yet seen the software the will so much as approach the full potential of the 3930k.

 

[Celegorm]

What are you rendering and what video cars are in those systems? If you are doing video or image rendering or even cad work videocards can make a big difference.

 

[FenixFan08]

I dont think it has anything to do with the video cards when it comes to modo. Because in the Computer B with the xeon cpu there is a Nvidia Quadro 5000. We got the computer built for GPU rendering and now we dont use that beast of a GPU.