7 Intel Core i7 X58 Motherboards Tested and Compared

Motherboards/Intel S1366 by massman @ 2009-02-25

We put seven feature rich X58 motherboards for Intel´s latest Core i7 CPU to the test. Comparing performance, overclocking scaling in a multitude of applications and games. Which one comes out on top? Read on to find out!

  • next

Introduction

Introduction

It has been almost two months now that the big public is able to enjoy the wonders of the new Core i7 technology, codenamed Nehalem, but although we use the word wonder, there are no dramatic improvements over the previous high-end Intel quad-core processors. In fact, the novelties of the Core i7 can be broken down into four points:

  • Small clock-per-clock performance increase
  • Integrated memory controller in the processor
  • Introducing triple channel memory configuration, which improves memory bandwidth
  • Re-introduction of HyperThreading

    Let's have a look at the conclusion of our launch article regarding the Core i7:

    The Core i7 delivers what it promised: better performance. But there isn’t a huge boost in single threaded applications. We compared triple to dual and single, but that just gave close to 0 extra performances. Although it is working correctly, delivering up to 25 Gigabyte per second memory bandwidth, this platform doesn't need it for the moment.

    The current crop of mainboards based on the Intel X58 are feature rich and offer everything you'll need, including Crossfire and SLI support on a select group of third party motherboards.

    Price wise we were not expecting this new platform to be cheap, and we are not proven wrong; the cheapest Core i7 capable motherboards hover around €300, while the entry level i7 920 is also close to €300, combine this with a capable DDR3 memory kit and you’re looking at ~€1000 for CPU/Motherboard/RAM. Knowing that you can build a fast Core 2 Duo (or even Quad) system for less than half of that, the current pricing of the Core i7 doesn’t make it very cost effective.


    Referring back to the four points of change, we can conclude that:

  • The small increase in performance is indeed visible, be it in multi-threaded applications only due to the HyperTreading mostly. In single-threaded or very limited multi-threaded applications, such as most of the recent games, the increase in performance is not noteworthy.
  • The integrated memory controller does its job perfectly: no complaints in this area
  • The triple channel memory configuration does increase the memory bandwidth, although the difference with single and dual channel configurations is negligible. As Intel already confirmed: the added channel is only to feed the processor in 8-threaded applications.
  • In multi-threaded applications, the HT feature does help. It's a pitty though that the HT package comes with extra heat production.

    It has become very clear that the LGA1366 platform is not meant for the mainstream users, but only for high-end enthusiasts who are willing to spend a few extra bucks for the highest possible performance. The LGA1366 is an expensive platform (as mentioned in above paragraph); not only because there are only three types of processors that will fit onto you motherboard (920-940-965), which means no low-budget variants, not only because the only memory supported is expensive DDR3, but also because the motherboards are in fact very expensive. At least, expensive when comparing to the middle-end LGA775 motherboards, which show up around €100-€125, whereas we have not found an LGA1366 motherboard under €200 ... a significant difference isn't it.
    In this 7-way X58 motherboard round-up, we'll be comparing the following motherboards both on raw performance as on the overclocking capabilities, which often indicates the maturity of a motherboard:

  • Asrock X58 Supercomputer
  • Asus Rampage 2 Extreme
  • DFI Lanparty DK X58-T3eH6
  • Foxconn Bloodrage
  • Gigabyte X58-Extreme
  • Intel DX58SO
  • MSI X58 Eclipse SLI


  • Madshrimps (c)
    • next
    Comment from Kougar @ 2009/02/26
    I think this thread was meant to be in the Articles section?

    Good review, nice to see almost all of the major boards together in one thorough review. I like how the OC tests were split up and the specific areas focused upon.

    I know it would have lengthened the time with testing/overclocking but I would have much preferred to see 5-10 minutes of IntelBurn for stability testing... SuperPi 4M or even 32M only proves the system won't BSOD at desktop randomly. As overclocking is one of my top factors in deciding which board to chose to buy, this is important to me as a future X58 buyer.

    Testing all the boards with the same processor in a single review (after plenty of BIOS revisions have already been released) means this review is one of the best comparisons for showing which board overclocks the best... but SuperPi 4M means nothing in terms of stability so I can't really draw definitive conclusions from the OC tests.

    The only other thing I could ask was maybe throwing some UD3 or UD4 and either vanilla or deluxe P6T results in to show how they compare with the flagship boards in the OC tests. Just wishing, anyway...
    Comment from geoffrey @ 2009/02/26
    Huge!!!
    Comment from Massman @ 2009/02/26
    Quote:
    Originally Posted by Kougar View Post
    I know it would have lengthened the time with testing/overclocking but I would have much preferred to see 5-10 minutes of IntelBurn for stability testing... SuperPi 4M or even 32M only proves the system won't BSOD at desktop randomly. As overclocking is one of my top factors in deciding which board to chose to buy, this is important to me as a future X58 buyer.

    Testing all the boards with the same processor in a single review (after plenty of BIOS revisions have already been released) means this review is one of the best comparisons for showing which board overclocks the best... but SuperPi 4M means nothing in terms of stability so I can't really draw definitive conclusions from the OC tests.
    I can't disagree: 4M is not really a good estimate for 24/7 overclocks, but for me it was the better choice in terms of stability testing and available time. Testing one motherboard's overclocking capabilities took me 1 full day, and that's only if everything went alright. Take into account the troubleshooting and you're off for a long journey :-).

    The conclusions you draw are not supposed to be conlcusive in terms of absolute overclocking capabilities, but should be comparison ONLY. The overclocking process is being affected by more than just the motherboard (as you know): for instance, the memory overclocking results can be slightly better or worse depending on the quality of your memory chips. That's what the comments under the graphs are for, btw :-).

    Next time, I'm going to change some things, though. I now already know that the maximum CPU-Z BCLK frequency will be replaced by maximum boot BCLK frequency. Actual stability tests are not an issue, as long as I have the time to do propper testing; with 7 motherboards on the testbed, that was kinda impossible. In stand-alone reviews, it shouldn't be a problem, though
    Comment from Kougar @ 2009/02/27
    Quote:
    The conclusions you draw are not supposed to be conlcusive in terms of absolute overclocking capabilities, but should be comparison ONLY.
    I just think having done this much, it would be worth doing that extra bit more to make it a comprehensive OC article. For all intents and purposes, I think I can safely say most X58 users will be overclockers.

    Even 5 minutes of IntelBurn is not much longer than SuperPi 4M, and more simple to use than 4x1 instances of SuperPi. If time is that critical, perhaps you could set arbitrary CPU, VTT, QPI (etc) voltages/BCLK settings and just see what boards pass or fail at given settings? Just a thought, not sure if it was a good one.

    Quote:
    The overclocking process is being affected by more than just the motherboard (as you know): for instance, the memory overclocking results can be slightly better or worse depending on the quality of your memory chips.
    Yes, of course. But for your review you used the same kit of memory. Just as everything else except the motherboard was kept identical. Which is why your overclocking results have the potential to be the most useful to readers than any other single-board review. Same CPU, memory, tests, OS, and same date that takes into account revised BIOS's. This review is as close to apples-to-apples OC comparisons as one can get.

    Quote:
    In stand-alone reviews, it shouldn't be a problem, though
    That is partly my point. In a stand alone review often memory/CPUs and other hardware gets changed, more time elapses so BIOS's get updated and further refined, general OC knowledge for a new platform is improved, etc. All of those make it less of a direct comparison if doing ~ 7 individual reviews verses 1 large roundup. I know "ideal" is very often different from "practical", but still it would be "ideal" to have.
    Comment from jmke @ 2009/03/01


    those prefab voltage read out points are just awesome for the overclockers and testers out there, too cool
    Comment from Massman @ 2009/03/01
    Foxconn Bloodrage has them too.

    There's quite an interesting story to tell about who 'invented' those pre-fab voltage read-outs, by the way
    Comment from jmke @ 2009/03/01
    where are they located on this board? Can't spot them in the pics at first sight
    Comment from Massman @ 2009/03/01
    Next to the DIMM sockets. In the article, they're not visible (at least, not if you don't know where they are). I only noticed them when I prepared the board for this OC session :-)
    Comment from jmke @ 2009/03/01
    pics!
    Comment from Massman @ 2009/03/01
    Comment from jmke @ 2009/03/01
    thanks added to article. Not as fancy as the Asus board it seems.
    Comment from Massman @ 2009/03/01
    No, but it does the job

     

    reply