1

Between this 4 CPUs, which one is the best for a lot of image editing/compression/optimization?

-Intel Core i7-2600K 3.40GHz
-Intel Xeon E3-1230v2 3.30GHz
-Intel Xeon E3-1270v2 3.50GHz
-Intel Xeon E3-1225v6 3.30GHz

I'm looking into renting a server that would be constantly performing image editing tasks in the background 24/7. I'm between those 4. Any ideas?

1

Out of these CPUs (and ignoring considerations other then performance), it's a toss-up between the e3-1225v6 and e3-1270v2.

In terms of raw power, the e3-1270 has about a 10% edge because of it's hyperthreading and faster clock speed - see https://www.cpubenchmark.net/compare/Intel-Xeon-E3-1225-v6-vs-Intel-Xeon-E3-1270-V2/3019vs1192 and https://ark.intel.com/compare/65727,97476 - they have similar instruction sets.

That said at this level the external factors come into play - eg memory size, cache, SSD/hdd performance. Depending on the types of manipulation, CPU offload might come in to play.

| improve this answer | |
0

It's a bit of a bet, but I'd go with E3-1225v6.

Two reasons:

  • Image processing is, by it's very nature, memory-bound. Meaning RAM speeds matter A LOT. 8MB of cache could, at best, fit a single 2MP 24 BPP image
  • Assuming your software supports it: AVX2
| improve this answer | |
  • Also the v6 are Kaby Lake S (2017) and v2 are Ivy Bridge (2012), there might be architecture optimizations in between – Jules R Nov 7 '18 at 7:45
  • @julesR there are optimizations on newer CPUs, but the. e3-1270 is MUCH faster then it's e3-1225 equivalent, which more then offsets its age. Surprisingly, despite larger litho, the e3-1270v2 has a lower tdp. – davidgo Nov 7 '18 at 13:16
  • @davidgo the 1270v2 has HT, the 1225v6 does not. For the TDP, they have changed the calculation method since Ivy Bridge, that's why KBL-S have larger TDP but probably not larger consumption. – Jules R Nov 7 '18 at 14:12
  • 1
    @JulesR on second thought... HT might just offset the slower memory. That's it's intended use after all. A lot also depends on the scheduler - Windows tends to bunch stuff at Core 0 or move stuff between cores. I experienced this on my system - Blender's GPU rendering is, unfortunately, single core on the CPU. – Jan Dorniak Nov 7 '18 at 15:06
0

You should definitely go to Intel Xeon E3-1225v6 3.30GHz, because it's the only that has AVX2, game changer for image processing, supported by compilers, OpenCV, FFMPEG, etc, and it probably won't compensate the lack of hyperthreading. It also has 14nm (vs. 22|32) and several generations of enhancements of the architecture, which increases efficiency.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.