Updates to our Testing Methodology

In light of various requests and as a naturally evolving process, some of our motherboard tests have changed.  This has the unfortunate consequence of making many of our previous results incomparable, but if they were never updated, we would still be on Catalyst 6.3 or something similar!  Also with experience allows us to pick tests that are able to run and give a nice score without having to sit with a stopwatch for twenty minutes each time.

For our CPU tests, we introduce Xilisoft Video Converter, a tool used to convert one type of video to another.  What I like about XVC is that it works using all threads available as much as it can, and if you have a GPU that is compatible, it will use it to help speed up conversion.  We also add in x264 HD Benchmark, a standard video encoding benchmark that a few other review websites use.

For the system benchmarks, we utilize two new features for USB testing.  In terms of hardware, we use a SATA 6 Gbps to USB 3.0 device connected up to an OCZ Vertex3 240 GB drive in order to really stress both USB 2.0 and USB 3.0.  For this test, we are also using software called DiskBench - a utility that records accurate times to copy from one drive through the USB to another.

Our GPU test suite also gets a bump.  Previously all benchmarks were run at 1920x1080, a reasonable gaming screen size.  Today we move all our benchmarks up to 2560x1440 using a 27" Yamakasi Catleap Q270 LED, a common Korean gaming monitor.  In terms of actual GPUs, we are retiring our now two generations old HD 5850s, and with many thanks to ASUS, now using a set of HD 7970s for up to four-way CrossFireX (should the motherboard allow it).  For NVIDIA, we still use a pair of ECS GTX 580s for SLI testing.  The games are also changing with AVP becoming a relatively trivial (120 FPS+) benchmark at 1920x1080, so we are replacing it with Civilization 5 at 2560x1440.

Test Setup

Processor Intel Core i7-3770K ES
4 Cores, 8 Threads, 3.5 GHz (3.9 GHz Turbo)
Motherboards ASRock Z77 Extreme4
ASUS P8Z77-V Pro
Gigabyte GA-Z77X-UD3H
MSI Z77A-GD65
Cooling Intel All-in-One Liquid Cooler
Power Supply OCZ 1250W Gold ZX Series
Memory GSkill RipjawsZ 4x4 GB DDR3-2400 10-12-12 Kit 
GSkill TridentX 2x4 GB DDR3-2666 11-13-13 Kit
Memory Settings XMP (1866 9-10-9)
Video Cards ASUS HD7970 3GB
ECS GTX 580 1536MB
Video Drivers Catalyst 12.3
NVIDIA Drivers 296.10 WHQL
Hard Drive Micron RealSSD C300 256GB
Optical Drive LG GH22NS50
Case Open Test Bed - CoolerMaster Lab V1.0
Operating System Windows 7 64-bit
SATA Testing Micron RealSSD C300 256GB
USB 2/3 Testing OCZ Vertex 3 240GB with SATA->USB Adaptor

Many thanks to...

Many thanks to ASUS as they have kindly allowed us to retire our now two generations old HD 5850s.   By providing us with a set of ASUS HD 7970s for up to four-way CrossFireX, we can really test any system that comes our way (coupled with that 2560x1440 monitor).

The ASUS HD7970-3GD5 was ASUS' first 7970 product on the market, using the reference design but being coupled with ASUS GPU Tweak software and support.  These cards offer solid performance, and I have easily overclocked them from 925/1375 to 1200/1800 with small bumps in voltage.  With any luck, the limiting factor for the next few generations of motherboards will be the motherboard rather than our GPU in the GPU testing, as well as some special X79 boards we are testing soon.

A little preview of some overclocking I have done with these cards for a future review

Many thanks should also go to G.Skill, for sending us one of their new TridentX memory kits designed specifically for the Z77 Platform.  The kit they have sent is the F3-2666C11D-8GTXD, or put simply, a dual channel 8 GB kit (two 4 GB modules) designed to run at DDR3-2666 MHz with 11-13-13-35 timings at 1.65 volts.

Having DDR3-2666 memory does seem a little far-fetched, given that Panther Point and the CPUs are only rated up to DDR3-1600.  However, with the integrated memory controller on Ivy Bridge, most motherboards should support up to DDR3-2666 to DDR3-2800 by default.  In the real world, this means that bulk memory transfers, or memory transfer limited situations, should get a boost.  This is in the realm of video processing, or integrated graphics gaming, or multi-monitor gaming (depending on how the game is processed).  However, given that motherboard manufacturers are releasing the fact that their motherboards should support these high frequencies, it does not hurt to test them!

In terms of G.Skill’s part in all this – their TridentX range will cover dual channel kits and quad channel kits, from 2x4 GB kits, 2x8 GB kits, 4x4 GB kits and to 4x8 GB kits.  The smaller 4 GB module kits will have speeds from 2400 to 2800 MHz, and are assured to be competitively priced, as well as overclockable.  In terms of our 2x4 GB 2666 11-13-13 kit, we were able to bump up the ratio to 2800 MHz without any change of voltage or subtimings.  With a little voltage push, we were able to go to 2950 MHz with some frequency adjustments.  As long as the CPU can take it, it seems that G.Skill have some overclockable kits ready for the taking.

 

MSI Z77A-GD65 - In The Box, Overclocking Power Consumption, Boot Times, Overclocks
Comments Locked

117 Comments

View All Comments

  • DarkRogue - Tuesday, May 8, 2012 - link

    First off, thank you for the review.

    I am a bit bummed that the UD5H and the Z77 Deluxe were not reviewed, since those were the two I was looking at. Especially since the UD5H can be had for under $200.

    Anyway, the voltage ripple/stability charts were quite interesting for me.
    But my main concern lies with the Gigabyte's chart. It looks good... but in my eyes, TOO good. It's too perfectly straight. On one hand, I thought, "Wow, this board has awesome VRM or something."

    On the other hand, it made me suspicious about why it was so stable. I have to ask, is it measuring the correct voltage?

    The reason is because I had a similar finding when I was looking at the vcore requried to OC an IVB CPU (or any CPU, really) on the new Z77 mobos.
    Per my thread here: http://forums.anandtech.com/showthread.php?t=22417...

    We found that Gigabyte mobos were incorrectly reporting its VTT voltage as the vCore, which resulted in "vcore" readings in CPU-Z and other programs reporting the same, or very similar, 1.0xx voltages regardless of what the CPU OC'd to.

    I hope to be able to get some clarification on this.

    Only other suggestion I have is to really test more of the features of each motherboard. (mSATA, firewire, audio, etc; how do they compare with other chipsets, how are their drivers, etc..)
    Thank you, and keep up the good work!
  • IanCutress - Tuesday, May 8, 2012 - link

    Hi DarkRogue,

    I have the Deluxe on my test bed right now, so keep your eyes peeled for when I finish the review.

    Regarding the voltage reading charts, it merely reads the OS reported voltage. This is loosely a smoothing of what ripple actually happens on board. After consideration, it only serves to show LLC on board, and how the board reacts to requested load by the processor. It's fairly easy for a manufacturer to override this to make sure only a straight line is reported. But, if it is a messy line, then there could be a problem (e.g. check my 990FX review a little while ago).

    I'd love the kit to test more features on the boards (mSATA etc), if you've got any kit spare! :) Though keep in mind that each test can't take 2 hrs, or we would end up with 1 board a month reviewed (as we do this part time)! I'm open to suggestions regarding tests if anyone has a good one with a simple output I can report and analyse.

    Ian
  • DarkRogue - Tuesday, May 8, 2012 - link

    Thank you for your response, Ian!

    I wish I had the funds to send in spare items, though. Unfortunately, I'm not quite rich enough, haha. We'll have to be bound by the generosity of the various vendors to this site.

    As far as tests go, I imagine some people would be interested in a quick RMAA test of the various audio chipsets. It's of no concern to me, since I insist onboard solutions are never going to be as good as a dedicated external DAC+amp, but it should be good for a lot of people.

    I'd also be interested in how the eSATA and Firewire performs, as I'm of the camp that says anything firewire related that isn't Texas Instruments is not worthwhile. The eSATA, mainly it's to see if there are any quirks with the drivers from each manufacturer allowing hotswap properly or not, and whether it causes DPC latency issues. My friend's ASUS board was plagued with problems related to eSATA not allowing him to eject drives, BSOD'ing on resuming from sleep, causing massive DPC latency when a drive was connected, etc. It's these little things that really make or break the experience of a board.

    I'd also like to see how well the fan controls are on each motherboard. ASUS' Fan Xpert 2 really drew me in, as it seems no one else can match the level of customization for fans. I dug a bit and found out that Gigabyte's boards not only cannot do this, but it even struggles to stay consistent between its various headers. (One header runs straight +12v no matter what, while the others make the fans spin at different speeds with the same settings.)

    ANYWAY, back to the issue at hand - Gigabyte's voltage readings.
    As I found in my thread linked above, Gigabyte appears to be reporting the wrong voltage, for some unknown reason. This to me seems to invalidate the test result for the Gigabyte board, because it's incomparable to the others.

    I know that the purpose of the test is to test for variation in the voltage to the CPU, not necessarily the exact ripples, but the VTT supplies voltage to a completely different segment than the vcore, unless I'm mistaken. I wouldn't think that the voltage supplied toward the IMC would vary as much when the CPU ramps up and down.

    Is there a way to force that program to probe a specific/different voltage reading, or have you already done this and the chart actually does represent Gigabyte's handling of vcore voltage? I wasn't able to figure that out from the article.

    Thank you again!
  • UltraWide - Tuesday, May 8, 2012 - link

    Thank you so much for covering the fan control features on each board! I truly appreciate this as it is often left out in other reviews.

    Keep up the great work!
  • AeroRob - Tuesday, May 8, 2012 - link

    I don't know about anyone else, but I'm really sick of seeing VGA and PS/2 ports wasting space on new motherboads.

    I know some gamers might think that PS/2 does the job better than USB, and I can appreciate that, but VGA? Who even uses VGA connections anymore? They should be avoided like the plague.

    And even if you do insist on using a VGA connection, what's the point of putting a DVI-D connector and a VGA together? Chances are you won't be using both, so just make it a DVI-I connector and throw in one of the cheap DVI>VGA adapters, and use the newly freed up space for a connector that isn't an ancient piece of garbage. Let's see HDMI or DP up there. Move things around so you can perhaps throw an eSATA connector on the back, or more USB ports--you can never have too many USB ports!
  • Paapaa125 - Wednesday, May 9, 2012 - link

    PS/2 has one single thing that makes it superior to USB: you can turn your computer on by clicking space bar on a PS/2 keyboard or clicking mouse button on a PS/2 mouse. USB does not have this feature which is a big problem if your computer case is not easily reachable.

    Agree about VGA ports. Nobody uses them anymore. Nobody.
  • AeroRob - Wednesday, May 9, 2012 - link

    As I understand it, gamers feel there's an issue with USB's polling rate, and prefer PS/2 for that reason.

    As for turning on your computer, I never heard about that. I rarely shut my computer completely off, and my wireless USB keyboard can wake it up from sleep just fine. Hell, my computer's so sensitive to any change, just flipping my monitor back on wakes it up (probably due to the built-in USB hub).
  • mcquade181 - Thursday, May 10, 2012 - link

    I know whole organizations that still use VGA, and there are tons of KVM switch boxes in development and testing centres everywhere that only support VGA. Yes I know you can get HDMI and DVI KVM's but most places won't have them yet.
    I still use VGA occasionally and would be annoyed if it wasn't there.
  • Paapaa125 - Friday, May 11, 2012 - link

    And how many of those organizations are switching to Z77 boards and still keeping their VGA?
  • Ramon Zarat - Tuesday, May 8, 2012 - link

    The latest AXTU version does not include XFastRAM anymore. XFastRAM is a stand alone utility now. Also, XFastRAM is much more than a 32bit 4GB RAM limitation extender. It's in fact a RAM disk on steroid, valid for both 32 and 64bit system. It can do the following:

    1- "Recycle" unused memory beyond 4GB on a 32bit OS into a RAM disk. A RAM disk of up to 32GB can be created on a 32bit OS.

    2- RAM disk of up to 8GB on a 64bit OS. Asrock is working on extending that limit on 64bit OS.

    3- Can choose any available driver letter to assign to your RAM disk

    4- Use part of the RAM disk as a Readyboost drive to accelerate your magnetic boot drive!

    5- Easy transfer of either or both the "user" and "system" temp file to RAM disk. No fooling around with Windows configuration.

    6- Easy transfer of IE and Firefox cache to RAM disk. XFastRAM take care of everything straight from its interface.

    7- Easy transfer of the page file to RAM disk. Again, directly from XFastRAM interface.

    8- Possibility to save the RAM disk to hard drive before shutting the PC down.

    It's fast (10 000MB/s with CrystalDiskMark on a 2500K @ 4.7 and 8GB of 1866 RAM) , it's free, it's amazingly flexible and can both accelerate your PC and prevent premature wear on your SSD by redirecting a lots of small writes to the temp folders and web cache! You can apparently gain 5X performance in some Photoshop operations when you configure it so use the RAM disk as the temp folder!

Log in

Don't have an account? Sign up now