Xeon E5-2680V2 worth the premium over an i7?

#1

Would it be wise to purchase a E5-2680V2 say over an equivalent i7?

From what I understand Xeons are designed for servers, but I’ve never understood how the ‘premium’ price translates over to workstations.

Do Xeons raytrace “faster” as well?

#2

The only significant reasons for you to pick a xeon cpu is that you can buy motherboard which will accommodate two or more of the cpus, and the chips themselves can have more processing cores. So instead of having a computer with a single Intel i7 cpu with 4 or 6 cpu cores, you could have a pair of Xeon cpus with between 4 and 12 cores each, for a total of 24 cpu cores in a single machine.

For rendering they will be the fastest. However consider that there are significant downsides, the obvious one being the very high price, you can pay thousands of dollars for a single chip, so xeons machines upwards of $10,000 are not uncommon.

The other thing is that for many of your intended daily tasks, a lot of the xeon chips will be much slower than the cheaper i7. A mistake I have seen many people make, is they go off and buy a pair of 2GHz 4 core xeons; giving them a total of 16GHz of rendering power (2 cpus x 2ghz x 4 cores). For much less money they could have purchased a single i7 chip with 6 cores at 3.6GHz for a total of21.6GHz of power. Plus all the software features which cannot use multiple cores will run faster.

#3

There is no equivalent Core i7 product as that processor has 10 cores at 2.8GHz. The highest end Core i7 doesn’t even come close in terms of performance for parallel workloads like ray tracing rendering.

Workstation and server class hardware like Xeon processors are designed for working with larger datasets and mission critical applications. In terms of computer graphics that means more processor cores for rendering and vastly larger memory capacity (simulations, compositing, lots of polygons). For comparison if you spend three times as much money on workstation and server class hardware it won’t be three times faster. There’s a considerable premium on the higher end hardware that isn’t always worth the investment.

It’s already been pointed out that sometimes a cheap gaming machine with fewer processors cores with a much higher clock speed can be faster than a hugely expensive machine with dozens of processor cores if the task is utilizing only on processor core.

If you can give specifics about the workflow and the budget we can make practical suggestions for hardware.

#4

Thanks for the replies.

I’m purely interested in maximum speed for rendering with V-Ray. At this stage I’m not concerned with single thread performance.

#5

At two grands street price you could most likely squeeze more out of a dual hexacore, or even more, if you don’t mind splitting across frames, from two boxes running i7s.

The E5-2680V2 will be perfectly fine for single threaded tasks as well by the way, overclocking aside it’s only a hair slower than a 4770k, and about 12% slower than the newest devils canyon. It’s not the worst bang for buck wise, actually, so if you absolutely need to fit the rendering bang all in one box it’s not bad. It’s not the absolute most cycles per dollar you can get though if compactness doesn’t feature. Two boxes with hexa i7s will definitely beat it to the finish line of a sequence of frames.

#6

I only partially agree with that. According to Cinebench15 a single 10core@2.8 score about 1350, an OC i7 can easily reach 1250(and I’ve seen some desktop CPU with more exotic cooling solutions on par with a single e5-2697v2). If you consider that a 6core i7 cost about a third of a 2680v2 I would say that this is close enough in performance and much better bang for the bucks.

#7

4930K CPUs will usually hit at least 4.3 with a bit of a base bump included, with stock-like air.
All cores smoldering you will still easily get about 22-23Ghz worth of switching (often more, depending on deltas in load) and a generally more aggresive rush on the choke points that are single or poorly threaded.
I found recent xeons are very, very throttledown happy and I’d be surprised if you saw more than about 29Ghz worth of switching in those deca cores, and a much poorer bum rush during choke points of any sort.

I wouldn’t be too surprised, actually not surprised at all, to see a 600$ 4930K hitting the same wallclock time per frame of the decacore e5s.

I can tell you for sure and tested my 4.5Ghz 4770K at home ends up routinely running circles around the high clocked hexacore xeons at work, and it’s only about 60-65% of the potential of a 4930k, let alone the X with the better on-die cooling.

#8

http://www.cpubenchmark.net/high_end_cpus.html

Intel Core i7-4930K @ 3.40GHz 13,222
Intel Xeon E5-2680 v2 @ 2.80GHz 16,799

Thanks for the replies, even though these benchmarks should be taken with a grain of salt - it does seem like the i7 offers a better sweet spot with price.

#9

Bear in mind the former will get a 10-15% bump as well with OCing, and that’s on stock air, a lot more if you pick lucky and go even just closed circuit liquid. The latter (AFAIK, but I haven’t checked in a long while) can’t be overclocked.
Whether that’s of any value to you or not is your call.

#10

Yes, the Core i7 offerings have better bang for the buck. The difference is you can put two Intel Xeon E5 series processors in a machine which widens the performance gap to more than double the Core i7 offerings. It comes down to the budget and how much performance you think you need (why I asked about the budget in a previous post).