Dual Xeon vs i7

M

MaxSk8

n00b
Joined
Dec 23, 2009
Messages
4
Hi,

I'm going to buy a 3D Workstation that will be running 3DS Max 2010, After Effects and Photoshop for professionnel use.

I would like to know what would be better between a Dual Xeon and a Core i7. Is there any benchmark out there that compare the two in 3D applications like 3DS and AE? I found some bench of Xeon and bench of Core I7 but none for Xeon VS i7.

Can you help me with that?

Thanks.
 
Which Xeons are you looking at? That makes all the difference. Naturally, a dual quad-core Nehalem-based Xeon system will outperform a Core i7 system if they are both running at similar clock speeds. Older Core 2-era Xeons might not perform as favourably, but would most likely still do better than a single i7. Also, if you go with an i7 system, will you be overclocking it at all?
 
I was looking at the Nehalem Xeon. The thing is that since they are way more expensive I didn't know if it was better to buy a i7 and change it sooner. I would like to compare rendering time between Dual Xeon Nehalem based vs i7.

I would not overclock the i7.
 
Response to MaxSk8:

Assumption
1. You do not want to overclock (as stated)
2. Timing :
2a. If you can wait, wait for 6-core Westmere
2b. If you must buy now, you will later upgrade to 6-core westmere.
2c. You do not mind hidden costs, (your own time, upgrading processor, logistic, handling hassle, misc)
3. ECC memory is not important to you. (Only Xeon officially supports ECC) Your software vendor/professional requirements do not think ECC is important.
4. Memory Amount - IF > 24GB reasonable cost I think you need to go Xeon

Then the following is my suggestion
1. If you wanted now, buy now with Core i7 920 $200 deal (other users can help you here). slow performance for several months.
2. Upgrade to 6-core westmere at 1000, sell the i7-920 to recoup cost.
3. You will save approximately 300-400 on processor cost and get within 88-95% throughput of a Dual Xeon E5540 (turbo enabled). excluding hidden cost and overhead. Perhaps less power bill later with single 32nm gear compare to dual-processor (not sure, need engineering measurement)
4. You will save more on other components because enterprise components for Dual-Processor environment are more expensive than consumer range generally.

Alternative : I believe lot of media professional will go for professional workstation judging from record profits of Apple computer so minor savings may not meant much to them.

Estimate, double-check with real engineer.
 
Additional notes for MaxSk8:

Sorry the previous comment about 3DS scaling concern is wrong.

I think as long as it is Multimedia/Audio/Visual type of commercial software environments you will get good scaling from all setups.

However, there are indeed concern how well Adobe software will scale pass 4-core, so I naturally favour higher clockspeed than more core in this scenario.

Finally, as already known Adobe is adopting GPU-based enhancement direction for its portfolio of products. I am not sure optimization on 8-core platform is their focus at this point. Perhaps you need to factor GPU into the calculation.
 
Last edited:
3dsMax will use multiple cores depending on what you're doing. If you're using some older features in 3DSMax (i.e.-Blobmesh) or vanilla Particle Flow, it'll be rather slow. However, I think Cebas Thinking Particles do much better with multithreading. In regards to rendering, Mental Ray, which comes with 3DSMax, will do very well with multicores, though some rendering and baking features with the default Scanline renderer can somewhat vary in how many cores they'll use.

Rumor has it the next 3DSMax will have GPU-based rendering via Iray/Mental Ray, which would be amazing. Nvidia's GPU accelerated renderer (Gelato) has basically been discontinued, so no use looking towards it for GPU-acceleration.

http://www.youtube.com/watch?v=TQsXualxLVs
http://www.nvidia.com/object/io_1254292325160.html

One concern, of course, is sequential events, like physics simulations, where one event has to run before the next, hence how to multithread this type of situation becomes difficult.

As for which choice, I would go with the i7 or wait. I'm currently using an q6600, and eagerly wait for the "next big" tons-O-cores CPU to come out before upgrading, especially since DDR3 is quite pricey compared to my 8gb DDR2.

PS: Some cool GPU-based real-time physics simulations,
http://www.youtube.com/watch?v=Ds8NjUslrwM&feature=related
(Rayfire for 3DSMax works great with Nvidia Physx, much better than Reactor! ;) )
 
Last edited:
If you're not overclocking, you're wasting the potential of the i7. At stock clock speeds, obviously two Nehalem Xeons beat one Core i7. This is a no-brainer...

You should reconsider overclocking, though. I am in the same situation (After Effects/Cinema 4D professional user) and decided to go Core i7 920 (for now) rather than the dual Xeon route. Reasons:

- Dual Nehalem Xeon motherboards don't seem to overclock at all - no info out there (that I've found)
- Upgrade path for dual Xeons seems unclear, whereas...
- Six-core i7 CPUs are coming out later this year
- Six-core i7 will be overclockable
- Extreme/enthusiast-branded hardware is generally cheaper than server-class and has more upgrade options/paths--and for my purposes, server class hardware is not necessary (when you deal with an application as innately unstable and buggy as After Effects, ECC RAM and such isn't going to save you from the inevitable crashes)
- More info and community support for overclocking enthusiast-class hardware in comparison to server-class
 
You must log in or register to reply here.
Back
Top