P67 Motherboard Roundup: Nine $150-200 Boards

[stasdm]

@elbert

Intel spec. of all Sandy bridge processors states 32GB (using 4 x single-rank 8GB modules) at 1333MHz.

So, 16GB is the overclocked (at max. board supported speed) volume.

 

[christop]

Nice review... Go Asus!!!!!

 

[stasdm]

@dgingeri
Both boards will support all four of your cards with 3Ware working w/o inteference with southbridge SATA (Nehalem southbridges had only 1GB/s bandwidth vs. 2GB of Sandy Bridge ones). That also allows better three-way SLI/CrossFire speed. Once again, with the first x8 slot all high/higher-end nVidia cards will loose about 12% of the speed - that's $20-50 money loss per board installed, compared to the mobos with the first full x16 slots.

@iLLz
Do not mix up processor/northbridge PCIe speed and southbridge PCIe speed on Nehalems - the processor/northbridge PCIe was thrue v.2.0 vs. half-speed southbridge lanes.

 

[stasdm]

@rusbee

Total LGA1155 processor PCIe bandwidth is 10GB/s, with the processor x16 PCIe at 8GB/s and 2GB/s on all southbridge devices.

Any independend nVidia (non GF110) Graphics card is no more than 0.7GB/s
AMD theorethical max. and nVidia GF110 (and derivatives) - no more than 1GB/s
second/third card in SLI/CrossFire - no more than 0.5/0.25 of that (respectively). So, even three-way GF110 will not take more than 2GB/s.
So,the card with true v.2.0 switch on the processor's x16 link may nicely support 3-way graphics plus one full-speed x8 and one full-speed x4 links.

On the southbridge (with additional SATA card) you may nicely support up to 7-drives software RAID 0 on 250MB/s SSDs (for Nehalem southbridge the effective number was only 3).

 

[dgingeri]

[citation][nom]stasdm[/nom]On the southbridge (with additional SATA card) you may nicely support up to 7-drives software RAID 0 on 250MB/s SSDs (for Nehalem southbridge the effective number was only 3).[/citation]

Definitely don't want to deal with software raid. hardware raid only for me from now on. I have 4 WD Caviar Black 750GB drives that simply would not work properly (one drive at random would drop from the raid every week) with the Intel SW raid. I put these same drives on the 3ware controller, and they have worked just fine for the last 2 months. Plus, I get far better performance out of them on the 3ware controller. Using the Intel raid, I got ~150-175MB/s, but with the 3ware controller I get a consistent 350MB/s for the first 2000GB, with it tapering off to 120MB/s at the end.

I'm never using chipset raid ever again.

 

[Crashman]

[citation][nom]stasdm[/nom]@Crashman on USB 3.0 on-board pin connectorAre they backward compatible with 2.0 pin connectors?[/citation]Here's a photo for you:

USB 3.0 and USB 2.0 are separate entities. People talk about backwards compatibility but the signals are still carried on separate pins. The adapter shown eliminates the USB 3.0 pins and passes through only the USB 2.0 pins. Hopefully these will also be available as a separate component...[citation][nom]dgingeri[/nom]Problem with the AsRock and Asus board review portions: the last slot is not meant for triple SLi. It's there for PCieX4 and X8 accessory cards. What I want to know is how these boards work with raid cards in that slot. Do I have to forgo using the other PCIe slots to use that last slot in X4 mode? More importantly, do I have to disable the addon SATA 6G and USB 3 controllers to use that slot in X4 mode? I currently have 2X GTX470 video cards, a PCI sound card, and a PCIeX4 SATA raid controller (8 port 3Ware 9650). Can I use either of these boards? This is an important question. you guys just answer it with smart remarks about using a third video card, which obviously wasn't intended.[/citation]Oh no, it WAS intended! According to the marketing materials, 3-way CrossFireX is a key feature.[citation][nom]iLLz[/nom]"stasdm, i was under the impression that lane bandwidth has doubled on P67 vs P55."Agreed. The PCIe lanes are now a full 500MB/s (1GB/s Bidirectional), whereas, with the older chipsets x58 and P55, they were only 250MB/s (500MB/s Bidirectional). With this doubling of bandwidth an x8/x8 configuration should yield the same bandwidth as a previous x16/x16 (x58 or P55), right? Even though the previous gen chipsets say they are PCIe2.0 they were limited to half the bandwidth and now with P67 they are fully PCIe2.0.[/citation]The X58 has 36 5Gb/s lanes (32+4) from the northbridge and eight 2.5Gb/s lanes from the southbridge, which helps to explain the debate over shared-pathway slots vs crippled onboard controllers.

 

[Crashman]

[citation][nom]rusbee[/nom]Thanks for the review.A major point missing the board info is the power phases. While most manufacturers have switched to digital, some boards have stayed with analog (Gigabyte GA-P67A-UD7 for one, I am not sure about the rest of their line-up). 2nd is the number of power phases and how it affects the life-time of the boards. While Asrock Extreme4 has 8+2, Extreme6 has 16+2. Asus P8P67 Pro and Evo have 12+2; Delux version has 16+2. How much does it matter (does one need less power phases with digital ones)?[/citation]You really need an extreme overclocking comparison to find out, because the quality of the components differs. Eight 50A phases are more durable than 12 30A phases, for example, and its difficult to rely on the data supplied by manufacturers..."trust but verify". The problem with extreme overclocking is that around 1/4 of the time that a voltage regulator pops, its last gasp is to send a voltage spike to the CPU. This CPU needed to be preserved at least until all nine boards were tested[citation][nom]rusbee[/nom]Another point which I am curious about is the quality of components used. MSI and Asrock use Polymer caps, where Asus seems to have cheaped out on a little here (their lower end H67 bards still use Polymer caps, but not the P67s). Is this going to significantly increase the probability of the boarding failing a few years down the road when the guarantee is over? Perhaps this is the reason VRMs on Asrock and MSI are running so much cooler compared to Asus despite using less phases?[/citation]Although capacitors age faster, overcurrent usually takes out the MOSFETs. It's been a long time since we've seen a failed capacitor on a name brand product.[citation][nom]rusbee[/nom]Currently, I am looking at two boards in particular: Asrock Extreme6 vs. Asus P8P67 Delux. Quality-wise, Asrock seems to use better components overall while being cheaper at the same time, but it does not use the Intel network controller which Asus does for one of the network controllers. I wonder how this affects ping times for online-gaming.[/citation]Google reviews on those specific network controllers? Intel's is a good one and generally considered a little better than a PCIe based codec, but I don't believe the difference is noticeable to *most* gamers[citation][nom]rusbee[/nom]As pointed out in other comments, I also want to know how and when the PCIe slots get saturated with two high end graphic cards and a 3d device as PCIe x4. I have two 6950s and a Revodrive 120GB which is going to use it's software-raid. Can any of the boards (even Asrock Extreme4, Extreme6 or the Asus P8P67 Delux which all have the PLX chip) handle and balance so much band-width?[/citation]These are actually separate questions, because the graphics cards are controlled by the processor's built-in hub. Adding a second card drops you from x16 to 2x x8 mode, with a minor performance drop already covered in several of our PCIe scaling articles. Separate from that is the P67's PCIe hub, which is limited by its 2Gb/s DMI interface.

 

[greenmachineiijh]

Sorry, Tom's. But is this a joke? Like all other Motherboard comparisons before, when the testing included default everything, the differences between boards with the same components is minimal and will have no real impact for anything!! The ONLY differences EVER seen from board comparisons such as this is in the board options (layout, connectivity, etc) BIOS options, and it's over-clock range. All of these other comparisons at default speeds and settings and saying that one board won over another by a fraction of a percentage is completely useless. the differences are so small and actually fall into margins of error. I would rather see all that time and effort go into the usability of the board, options, and more on the over-clock. Otherwise these tests are just verifying the components used in the testing, and not really what the BOARDS can do.

 

[Crashman]

[citation][nom]greenmachineiijh[/nom]Sorry, Tom's. But is this a joke? Like all other Motherboard comparisons before, when the testing included default everything, the differences between boards with the same components is minimal and will have no real impact for anything!! The ONLY differences EVER seen from board comparisons such as this is in the board options (layout, connectivity, etc) BIOS options, and it's over-clock range. All of these other comparisons at default speeds and settings and saying that one board won over another by a fraction of a percentage is completely useless. the differences are so small and actually fall into margins of error. I would rather see all that time and effort go into the usability of the board, options, and more on the over-clock. Otherwise these tests are just verifying the components used in the testing, and not really what the BOARDS can do.[/citation]

This simply isn't an accurate assessment of the review's content. There's an overclocking section at nonstock voltage levels, which is pretty much what you're saying the article should have been about. This is a major focus of the conclusion, so your approval is appreciated.

As for the benchmarks, these are only included for two reasons: 1, to identify problems and 2, because other readers expect them to be there. You mentioned reason 1 in your own conclusion, and indeed the benchmarks showed how one of the board's performed worse than the rest.

Thanks!

 

[cmartin011]

I love the UEFI interface very spiffy. ASRock is has a very nice USB 3 connector. like the power reset too... common gigabyte catch up!

 

[mapesdhs]

stasdm writes:
> Also, in x8/x8 PCIe configuration nearly all NVIDIA cards (exept for low-end ones) will
> loose at least 12% productivity

Absolute rubbish. Try running real tests and gathering genuine data before posting
such hokum. I've run loads of tests:

http://www.sgidepot.co.uk/misc/pctests.html
http://www.sgidepot.co.uk/misc/stalkercopbench.txt
http://www.sgidepot.co.uk/misc/x3tcbench.txt
http://www.sgidepot.co.uk/misc/uniginebench.txt
http://www.sgidepot.co.uk/misc/uniginebench2.txt
http://www.sgidepot.co.uk/misc/uniginebench3.txt
http://www.sgidepot.co.uk/misc/viewperf.txt
http://www.sgidepot.co.uk/misc/cinebench.html
http://www.sgidepot.co.uk/misc/ptboats.txt
http://www.sgidepot.co.uk/misc/callofjuarez.txt

Notice how my results compare to my friend's X58 rig, beats it every time, whereas
your comment implies my system should be 7% slower.

The amount of FUD surrounding PCIe scaling is ridiculous. Only thing worse is you
posted that garbage twice.

Ian.

 

[elbert]

[citation][nom]stasdm[/nom]@elbertIntel spec. of all Sandy bridge processors states 32GB (using 4 x single-rank 8GB modules) at 1333MHz.So, 16GB is the overclocked (at max. board supported speed) volume.[/citation]
Many motherboards limit their support to the 16GB no matter OC'ed or not. The one I suggest supports up to 32GB's. At some point down the road we may need more than 16GB's and no longer require overclocking. Other CPU's down the road may change limits. The end result is the ASUS P8P67 EVO supports 32GB's of memory where many here do not for $199.99.
http://www.newegg.com/Product/Product.aspx?Item=N82E16813131695

 

[Sabiancym]

Thankyou so much for this. I've been torn on which mobo to go with for my new Sandy Bridge Build.

 

[stasdm]

@mapesdhs
Diar sir,
Your tests show nothing (well, exept, might be, that your friend did not use SLI bridge or installed it improperly).

Run the following tests:
nVidia GPU alone (second x8 free)
nVidia GPU plus ANY card in the x0/x8 slot
and you will see what I am talking about.

Also run SLI test with and w/o SLI bridge installed - feel the difference.

 

[stasdm]

@elbert


4 x DIMM, Max. 32 GB, DDR3 2200(O.C.)*/2133(O.C.)/1866(O.C.)/1600/1333/1066 Non-ECC,Un-buffered Memory
Dual Channel memory architecture
Supports Intel® Extreme Memory Profile (XMP)
* According to Intel® SPEC, the Max. 32GB memory capacity can be supported with DIMMs of 8GB (or above). ASUS will update QVL once the DIMMs are available on the market.
* Hyper DIMM support is subject to the physical characteristics of individual CPUs. Some hyper DIMMs only support one DIMM per channel. Please refer to Memory QVL for details.
* Due to CPU behavior, DDR3 2200/2000/1800 MHz memory module will run at DDR3 2133/1866/1600 MHz frequency as default.


Kingston prodices single-rank 1333MHz 8GB modules for nearly a year, some other vendors too.

 

[stasdm]

see http://usa.asus.com/product.aspx?P_ID=erWGi4vbeZa8oxg8&templete=2

 

[elbert]

[citation][nom]stasdm[/nom]@elbert 4 x DIMM, Max. 32 GB, DDR3 2200(O.C.)*/2133(O.C.)/1866(O.C.)/1600/1333/1066 Non-ECC,Un-buffered MemoryDual Channel memory architectureSupports Intel® Extreme Memory Profile (XMP)* According to Intel® SPEC, the Max. 32GB memory capacity can be supported with DIMMs of 8GB (or above). ASUS will update QVL once the DIMMs are available on the market.* Hyper DIMM support is subject to the physical characteristics of individual CPUs. Some hyper DIMMs only support one DIMM per channel. Please refer to Memory QVL for details.* Due to CPU behavior, DDR3 2200/2000/1800 MHz memory module will run at DDR3 2133/1866/1600 MHz frequency as default. Kingston prodices single-rank 1333MHz 8GB modules for nearly a year, some other vendors too.[/citation]
History dictates a possible Gen2 700 series. An i7 2700k may come a year or so down the road and this mobo will support the 4 x 8GB DIMM's vendors should have out. Wouldn't that be a great upgrade scenario? I think this possibility is worth a few extra dollars for the ASUS P8P67 EVO. Over the mobo's listed here the EVO is clearly in the price range and 32GB is the most value added options I can think of for a $199.99 motherboard.
http://www.newegg.com/Product/Product.aspx?Item=N82E16813131695

 

[arkadi]

blah i really having "vista" like fillings...about all this....know what i mean?

 

[akula2]

Damn, looks like Sandy Bridge posing quite a big challenge (to resist temptation!). BTW, what happens to the people like me who have invested more than $2,000 to build high end P55 based PCs? It was hardly an year for me, this feeling sucks

 

[stasdm]

@akula2
Any PC is outdated at the moment you plan to by it

 

[Crashman]

[citation][nom]akula2[/nom]Damn, looks like Sandy Bridge posing quite a big challenge (to resist temptation!). BTW, what happens to the people like me who have invested more than $2,000 to build high end P55 based PCs? It was hardly an year for me, this feeling sucks[/citation]This is a fairly big jump from P55-based processors...but hey, it's possible to replace just the CPU and motherboard!

 

[akula2]

I won't replace anything Crashman because they are working quite heavily in my Business and a few at my home but I would build a new one based of P67 chip for sure

But, what MOBO should I go for? Hmmm, for P55, last year I've bought 23 MOBOs (P55A-UD4P, one P55A-UD6 and one P55A-UD7). I'm confused between Gigabyte/Asus. Honestly, Gigabyte's color kinda dull. Anyway, am not very much into color but still...just wondering about which one should I go w.r.t BIOS/EFI, lack of few features such as no Power, Reset buttons in Gigabyte board etc. Any suggestions?

 

[akula2]

One more thing, most of these reviews are pounding with blah blah about 'K' series over clocking. In my experience, any PC's health won't last longer with a consistent and continuous high OCing (say, stock of 2600K to over 4GHz. No reviews says what is the safe OC levels for these 'K' processors?

Last but not least, we know these processors are 2nd gen but will there be huge difference between i7-875K vs i7-2600K? I think yes but roughly by what %? Say, 15-20%? I wish to know the performance between:

i7-875K vs i7-2600 vs i7-2600K (sort of).

Please share some info, thanks

 

[Crashman]

[citation][nom]akula2[/nom]I won't replace anything Crashman because they are working quite heavily in my Business and a few at my home but I would build a new one based of P67 chip for sure But, what MOBO should I go for? Hmmm, for P55, last year I've bought 23 MOBOs (P55A-UD4P, one P55A-UD6 and one P55A-UD7). I'm confused between Gigabyte/Asus. Honestly, Gigabyte's color kinda dull. Anyway, am not very much into color but still...just wondering about which one should I go w.r.t BIOS/EFI, lack of few features such as no Power, Reset buttons in Gigabyte board etc. Any suggestions?[/citation]Gigabyte uses flat black, everyone else uses gloss black. ECS has the most opaque color, while the others allow you to see through the dye. A representative of another company actually told me that ECS' color is too expensive to work with...

Anyway, Gigabyte or Asus, I'd sort it out over features. I'm sure some people will prefer either shiny or dull finish, but I'm not going to allow that preference to effect my decisions

 

[stasdm]

@akula2
At the same frequencies, floating point benchmarks will be higher on Sandy Bridge. Do not expext any real difference in real life apps - they do not "know" anything about new possibilities.

If you use RAIDed s/b SSDs- storage part may (and most possible - on 3-4 drives arrays - will) work a bit faster

As you use the processors with GPU part in idle - no difference in graphics.

Non-extream OC does not harm the processor (keep it cool - that's all) and mobo.