• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

CPU for scientific image processing: X6 1100T or i5-2500K or i7-2600k or Fx-8130P

CPU for engineering-class image processing

  • AMD Phenom II X6 1100T

  • Intel i5-2500k

  • Intel i7-2600k

  • AMD Bulldozer Fx-8xxx


Results are only viewable after voting.

harshal

Member
Dear All,

I am planning to buy a system for engineering-class image processing. Budget is equivalent of USD 800 :'(. After following forums, I feel being stuck. 🙁

Which CPU is good for doing high throughput image processing ?
I would couple the CPU with either GTX 480 or ATI 6950 2GB DDR5.

My choices are:

AMD Phenom II X6 1100T (primary choice)
Intel i5-2500k (primary confusion)
Intel i7-2600k
AMD Bulldozer Fx-8130P (waited gamble?)

Thank you.
 
Last edited:
2600k > 2500k > X6 1100

How the AMD Bulldozer Fx-8130P will compare is anyones guess at this point.

If the software you use supports GPU acceleration though, this whole thing becomes meaningless, and you should look at that instead.

In the end I voted i5-2500k, because it's better than the X6, and because I don't think you can fit the i7-2600k in your budget.
 
Last edited:
If the software supports GPU acceleration, get a 990FX board($200), a cheap quad core Phenom II($100), and as many video cards as you can afford. The GTX 480 is not very good value for money. You can get 4x GTX 460 for around $500. That's a total of $800 for everything.
 
The processor may not matter in this instance ... look at GPU options. That said, Intel is the better platform right now.
 
In the end I voted i5-2500k, because it's better than the X6, and because I don't think you can fit the i7-2600k in your budget.

I did the same. Obviously, if you can make an i7-2600k work with your budget then that would be the best choice.

I'm surprised there are so many votes for an unreleased processor...
 
I shall be using combination of OpenCV, R, gsl, GNU Octave and custom C++ code.
No decided on scope of R and Octave.

Morever, if I buy Intel, I would have option of C++ libraries such as MKL, TBB and IPP. For AMD, I don't know equivalent of these.

Yes, even I am surprised over see many votes for Bulldozer and for 2600k also. I thought 2500k is better VFM than 2600k. Am I missing something here?
 
Who voted for BD? That isnt even available and we have no idea how fast it will even be once it is...
 
I vote wait for the FX-8xxx 😀

If we are to believe that AMD will once again be competitive at all, FX-8xxx should faster than the 2600K in MT situations.

Depending on how lazy you are feeling, what I would do is pick some representative workload, go to a brick and mortar store, and run some benchmarks yourself. That's what I did and the X6 turned out to be the better choice (vs the i5 750, this was last summer).

2500K is a solid choice if you need it tomorrow, though.
 

990FX has 40 PCI lanes.

Being able to do 4x PCIe x8 is why you should get a 990FX board, and a Phenom II or a Bulldozer, then put 4x GPUs in there in CF or SLI.
 
Who voted for BD? That isnt even available and we have no idea how fast it will even be once it is...

If he isnt pressed and is able to wait till it shows up, why not? I'd assume he is aware of the general time frame for BD release.

If he needs to buy fairly soon and be up and running sims then I'd say the 2500k might be the best option.
 
Is it a school thing or a job thing this image processing?

If it's a school thing I'll use the money for beer and just pound the various servers available for use on a typical campus network.
 
This is personal investment to acquire professional skills.

Wait, why do you need a new processor at all? If you for some reason have a single-core processor now, or need a better instruction set (AVX, SSE4.2), or need hyperthreading experience, get an i3 2100. Otherwise, don't waste your money on a new build.

If you're looking for GPU experience, get a GTX460 (or 450, or 430 if you're desperately poor), and put it in your existing build.

Edit: P.S. I'm not against building a super-fast computer; I just don't want you to do it for the wrong reasons.
 
Last edited:
Wait, why do you need a new processor at all? If you for some reason have a single-core processor now, or need a better instruction set (AVX, SSE4.2), or need hyperthreading experience, get an i3 2100. Otherwise, don't waste your money on a new build.

I appreciate and concur with you. In fact this computer should replace my old laptop Dell Inspiron 8600 of 2004 running on Intel Centrino 1.6 GHz. It is still working fine for browsing and entertainment.

If you're looking for GPU experience, get a GTX460 (or 450, or 430 if you're desperately poor), and put it in your existing build..
At least one GTX 460 (or ATI 6950).

Edit: P.S. I'm not against building a super-fast computer; I just don't want you to do it for the wrong reasons.
Appreciated. However I would like to buy a decent computer so that I need not to buy another computer in next 3-5 years.
 
OK, in that case, consider the following:

For highly parallel loads, an i3 will be at least 10 times as fast as your existing processor. For single-threaded loads, I'd guess about 5 times as fast.

For highly parallel loads, an overclocked i5 could be up to 3 times as fast as the i3. For single-threaded loads, it's only about 50% faster than the i3 at best. For about $100 more than the i3.

For highly parallel loads (8 threads), the i7 might be 20-40% faster than the i5, at best. For about $100 more than the i5.

Then there's one more thing to consider: Ivy Bridge is coming out next year and can fit in the same socket as all these new Intel processors. So you could get a cheap i3 now and easily upgrade in a year or two.

P.S. Current nVIDIA GPUs are much easier to program for than AMD, IMHO.
 
Appreciated. However I would like to buy a decent computer so that I need not to buy another computer in next 3-5 years.

Time-zero economics aside, looking at this from the human psychology angle (and obviously speaking from a limited sample size - namely one - myself) very few people lament having bought the more powerful processor come two years down the road.

I doubt you will find too many forum members here who bought the 2600K will come back a year from now and lament "gee I sure wish I hadn't bothered buying the 2600k, I should have just gone with the slower/cheaper cpus (2500k or 1100T) and saved myself the coin".

People do lament, down the road, having not gone with the higher performing CPU at the time of their purchase though.

Plenty of "gee, I was already spending $200 on the 1100T, why oh why did I not just drop another $100 and go all the way for the 2600K?".

I might be prejudiced about this perspective, but I've just never come across the case where a member felt like they bought themselves too much CPU, whereas I feel like I've interacted with plenty of people who come to feel like they were "a penny wise but a pound foolish" by scrimping and buying the cheaper CPU at the time.

If you are looking for a CPU that will be used 3 yrs from now I suspect you will torment yourself then (3yrs from now) for having saved a few bucks by buying something less than the 2600K. Just my opinion, heavily biased with projections of how I would feel if I were in your shoes of course.
 
People do lament, down the road, having not gone with the higher performing CPU at the time of their purchase though.

One important reason to put BD in the poll. Few dollars+ and few days+ may give me a marathon class machine.

For image processing applications, as I understand, integer operations would dominate over floating point operations. I suspect BD's architecture may actually outperform sandy bridge architecture due to separate integer units and shared FP units. Moreover, with Intel's Tick-Tock model, Ivy Bridge is only die shrunk implementation of SB. So if BD arch can beat SB arch, it's going to be very interesting.

Or I may be too naive to make this point.
 
Back
Top