"...During light load activities (idle and playing H.264 video), the DZ77GA-70K was noticeably energy inefficient compared to its H67 compatriots; there was a difference of up to 6W..."

Shouldn't the comparison be with Z68, not H67 since the progression from Sandy Bridge to Ivy Bridge is as I understand H67->H77 and Z68->Z77.

I actually almost bought this for my new build as it has an excellent layout but it has some design deficiencies that kept me from pulling the trigger (got the Asus Sabertooth Z77 instead - I will be removing the "thermal armor"):

(i) Small Issue. As mentioned in the review, nothing but an HDMI video output so besides not allowing multi-monitor with the IGP it is also a no go when using Lucid Virtu (if it eventually turns out to be something useful) on any monitor with a resolution above 1920x1200. Not likely to be a deal breaker for most people.

(ii) Medium Issue. As mentioned in the review, insanely high power consumption (http://hothardware.com/articleimages/Item1829/power.png). To make this even worse, in the same price bracket, Asus has the EPU feature which drops load consumption by a full 10W or so (http://techreport.com/r.x/z77-mobos-san ... r-load.gif) and you are starting from an even lower initial power draw point.

(iii) HUGE Issue. This is the deal killer. The board runs all of the following off of a SINGLE PCI-E x1 link - TWO PCI-E x1 slots, two PCI slots, the Marvell chipset which runs both two internal 6GB/s SATA ports and one eSATA port and Firewire. It states this explicitly in the manual (web pic here: http://images.anandtech.com/doci/5728/C ... iagram.png). Yes, Intel uses a switching chip but there isn't much choice given the amount of things running off of such limited bandwidth.

On the plus side, at least they include a USB 3.0 3.5" bay in the box unlike Asus.

Other reviewers of this Intel motherboard don't find this. For example, this one http://www.bjorn3d.com/read.php?cID=2209&pageID=11732. Here using Prime 95 the Intel Z77 board uses 2 watts less power at load than an Asus Z68. It also used 10w more at load than the MSI Z77A-GD65, and 2w more than that board at idle. But according to Hothardware, the gaps were 29w at load and 17w at idle with the very same MSI Z77A-GD65. So it looks to me that the Intel Z77 does uses more power than some of the other boards, but the margin is probably nothing like Hothardware suggests. And more importantly even the Intel Z77 is as good as or a touch better than most Z68 boards.


That's not correct. The two internal 6GB/s SATA ports run from a Marvel 6Gb SATA controller linked directly to the Z77 PCH. There is another Marvel 6Gb SATA controller linked to the PCI Express Switch, but it is only for the back panel eSata connector and nothing else. Look at the diagram in the manual again, count them, two Marvel 6Gb SATA controllers not one.

With regard to the power draw, thanks for the chart, it's interesting how divergent it is compared to the hothardware one. Even if we assume that the bjorn3d chart is the more accurate of the two, at the very least, the power draw is equal to the Asus Z77 boards. Using the EPU feature on the Asus boards leaves the Intel in a class all its own when it comes to power draw (not to mention that I can't really understand the power draw given the limited amount of additional options on the board).

With regard to the components running off of the switch, you are correct regarding the Marvel internal SATA ports running off their own PCI-E x1 link, I missed that. However, the point stands as there are way too many components running off of a single PCI-E x1 link not to suffer bandwidth starvation. Don't even look at anything else, just use the two PCI-E x1 slots and you've effectively halved the available bandwidth to each (notwithstanding that the switching chip will do its magic to try to load balance in such situation).

The most common separate PCI-E board now is likely to be a sound card, which will be x1. In practice, an x1 PCI-E board can be plugged into a spare 16x or 4x slot so the situation you envisage would only arise where two PCI-E x1 boards were being used, and both PCI-E 16x slots and the 4x slot were filled. So probably not a problem for most people, but you're right it could be an issue for some.

Thanks for the review. The feature mix for this board is kinda odd. One video port, two NICs, IR, the odd wifi/bluetooth internal module/no external antenna. Sort of an HTPC and high end desktop mashup where the best features of both were lost.

I am happy to see the new UEFI/fan controller and fast boot times. The fan controller could be a lot of fun to mess with. Sad to see the less than stellar power usage. Hopefully, some of the other Intel mobo's will have a better feature mix/power usage.

errata: It looks like OCZ whacked their RAM product line - it's all end of life. So, the link to their site's product page is broken.

Note that in the past using anything but video cards in PCI-E x16 slots was problematic (that may no longer be the case). Also, if you are using anything other than an Ivy Bridge CPU, if you install anything in that second PCI-E x16 slot, the first PCI-E x16 slot reduces to PCI-E x8 2.0 (I have no idea if that bandwidth limitation actually affects video card performance).

I was considering the Intel board due to its PCI slots. I wanted to use several add in cards/peripherals - Creative PCI-E x1 sound card, Terratec PCI sound card (was trying to keep hardware MIDI), ARECA PCI-E x1 RAID1 card and an eSATA backup drive for the RAID1 array. I would use the first PCI-E x1 slot for the Creative card and given that my VGA cooler would block the second PCI-E x1 slot (pretty much any VGA cooler will do that), the first PCI slot and the second PCI-E x16 slot, the Terratec would go in the second PCI slot and the ARECA card in the PCI-E x4 slot. In such a set up, the Creative, Terratec and eSATA drive would be sharing bandwidth. To be fair, I would not be using the two sound cards simultaneously (and the Sabertooth does not have PCI slots). But even then, you have the Creative and the eSATA sharing bandwidth. I know I could schedule the eSATA backups at times other than gaming times or maybe the bandwidth is sufficient for both or I could switch the eSATA to a USB 3.0 enclosure, but at $250 why should I have to schedule activities around each other, wonder about optimal performance or degrade external drive performance (USB 3.0 is still not as good as eSATA for small file transfers)?

While you are probably right in that most people with one to two PCI-E add-in cards will not likely run into any issues on the board and in some cases the arrangement might be preferable (for example, running a PCI-E x1 sound card and a PCI-E x4 RAID card where, for example, on the Sabertooth, using the PCI-E x4 slot at full bandwidth disables the other three PCI-E x1 slots and only leaves you with the second PCI-E x16 slot (see first paragraph)), it just seems to me to be a case of bad engineering by Intel. The solution on the Sabertooth whereby the PCI-E x4 slot can be used at full bandwidth or effectively split into four PCI-E x1 slots is on the other hand very sharp engineering (in fact, the arrangement even lets me use my Intel Pro/1000 PT NIC for a total of two Intel NICs like the Intel board (you might lose the option of NIC bonding in such arrangement though if that's important to you - I haven't researched this) - I need this as my PC is also a gateway to all other components in my place (VPN) - without any bandwidth sharing between any peripherals or eSATA).

Bar81, you are way over thinking this.

First off, almost no video card is affected by running at PCIe 2.0 x8. It's still plenty of bandwidth, so you can use the second x8 slot for anything you want without penalty.

Second, you are talking about sound cards. Even if BOTH cards were simultaneously doing something ridiculous like 8 channel 192 khz 32 bit, that is a whole 12 MB/sec. That shared 500 MB/sec PCI x1 interface isn't even going to notice. You would have to be running a fast SSD on the eSata port to even approach 500 MB/sec. Assuming you are using a mechanical hard drive for eSata you are definatly using less than half of that shared port, likely less than a quarter and that includes the sound cards.

For that shared interface to be a bottleneck you would need two PCIe 2.0 disk controllers with fast SSDs or RAID arrays both going at the same time. If you are doing that you should use the x4 or x8 slot.

Owning an ASUS board with a switchable x1/x4 slot I can say that Intel's way is much better. If the slot is in x4 mode then other things have to be disabled. With the bridge chip on the Intel board the x4 has full bandwidth AND everything else still works.

It's worth noting that this would only be an issue when using Lucid Virtu MVP in I mode (display plugged into the motherboard), not in D mode (display plugged into the graphics card).

It's my OCD . Thanks for the numbers; they make it much easier to conceptualize why Intel designed the board the way they did. I guess then I stand corrected; maybe I should stick to my day job and give up my career as a mobo engineer before it gets off the ground (or bursts into flames is more like it ).

At the end of the day, I had the luxury of choosing between what I now understand are two excellent motherboards and I think anyone would be happy with either the Intel or the Asus.

Thanks for the clarification. I just did some research and it appears that you can actually use quick sync in D mode (which is nice), so all you would really be missing is the power down of the 3D card from I mode (however, given reports about poor (?) Intel IGP image quality this may be one of those situations where you trade a positive for a negative).

Congratulations on the review. I just want to say thanks for the "Speed Fan Support" part. I know this aspect has beed covered in past mobo reviews and i hope the trend continues. It is sometimes dificult to find this info when chosing a new mobo. For a reviewer, it takes one minute to populate each header with a fan and test speedfan suport, but few other sites care.

To me this is the deal breaker - any motherboard i chose MUST have a minimum of 2 fan headers fully controlable with speed fan, doesn't matter if PWM or voltage. I just link toghether 2-3-4 identical fans on the same header if needed and all is good.

I don't know about image quality, but D mode will also give you a slightly lower latency (though it's supposed to be a pretty small difference in practice). It should also make it easier to disable Virtu MVP in case it causes some issues (though I don't know whether it's likely to happen or not): no need to tinker with your monitor cables to get back to a 'normal' configuration.

Anyway, modern graphics card at least tend to do a decent job at reducing their power consumption when idling, so I guess it probably won't really be worth using I mode on a desktop PC...

Unless you have 2 monitors and then your gpu card will idle at 17-90W, depending on the brand and model. I mode is a potential way around this.

Well, it looks like the numbers across the web seem to be trending toward point the fact that the power consumption in comparison to other boards may not be higher (and may actually be significantly lower than the other board with PCI I was considering, the Gigabyte Z77-UD5H):

http://www.techspot.com/review/521-inte ... age13.html

Wow, I had no idea that video cards could have such issues with power-saving states with two monitors - thanks for the link!

Hi guys, it seems that the high load power consumption was caused by the UEFI pumping up the multiplier to 39x rather than 34x which is standard for the Core i5-2500K. After re-testing I found the numbers were middle of the pack compared to the H67/P67 boards compared. I've updated the charts and comments accordingly.

Nice - thanks!

Good detective work. So the hothardware review was an outlier.

Some of the images are not working:

http://www.silentpcreview.com/article1256-page5.html

Fixed.

or maybe not?:

http://vr-zone.com/articles/asus-p8z77- ... 685-6.html

I give up.

Those vr-zone figures are for a CPU overclocked to 4.8Ghz. To do that they would have used a 48x multiplier and additional core voltage. So those figures are not representative of what the power consumption would be for the stock speed, 3.5Ghz.

Correct. Since they don't list voltages on the overclock, it is not possible to determine whether all the motherboards are at the same voltages such that you could extrapolate to stock performance in relative terms.