P67A-UD7 Most In-Depth Preview/Review!!

[bingo13]

Thanks I knwo why he pushed for it, and I believe/guess the reason they switched was because of VRD 12 and that the PWMs didn't meet spec, probably something very stupid like serial vid was not able to implement, its one a 3 wire system, but always incompatibilities arise.
I have never read an article that said Digital PWMs are better for Computer systems than Analogue PWMs, all the EE articles I have read said they are in their infant stages and mentioned ongoing work by volterra and chil. but I have never seen a comparison of them straight out. We are talking about power output, not straight mhz of OCing, max poweroutput is the 24 phase analogue by gigabyte, best precision over volt is from digital pwms, loadline is handled better by digital pwms, response time analogue vrms, Ripple controller digital pwm. I think everyone woudl agree with those. I did a ton of research before writing my review/articles, and i state facts that I have read from reputable sources, aka datasheets, papers from research institutes, i went to ga tech which is a research institute, i did a lot of research.

I stated those in my article, and i do not see how those bash digital pwms.

Actually you came in here at the behest of Gigabyte to promote their marketing message on these boards. Fact is, this exact same message is being passed on to other forums, resellers, and review sites by other personnel inside and outside of Gigabyte. I have refrained from addressing Gigabyte's "claims" that you have promoted but if you continue this then I think we are going to have a serious educational seminar on your points when I am not under NDA.

The fact you have continued to say that Digital PWM is in its infancy is incorrect and you have no idea about the CHiL setup (all new design) that ASUS is utilizing on their P67 boards. You claim you did research and read papers, well that is about the same as me claiming to be a heart surgeon because I read medical documents on becoming a heart surgeon.

You also hinted that UEFI might not be as stable as a BIOS design. This is just FUD as I can tell you for a fact that this is not true and just another marketing point from Gigbayte as they were unable to get UEFI on their boards due to engineering problems. They will move to UEFI shortly and also to a Digital PWM, so it will be interesting to see what messaging you provide from Gigabyte next quarter.

As a hint to the marketing BS in your post I leave you with these three nuggets to answer truthfully until launch when we can really discuss their 24-phase power system and just how it actually operates or you can actually read the Intersil documentation and then comment on the multiplexing aspects of the design and what that truly means in transient response among other things.

1. Why does Gigabyte default the Bclk to a non-standard 99.80MHz?

2. Sandy Bridge has a digital SVID design now, what is required by an analog setup to properly handle CPUVID?

3. What is the maximum TDC output of Sandy Bridge at the allowable cpu core voltage limit of 2.0V?

[sin0822]

I am sorry i thought they just tossed surgeons into surgery, wait no they educate them first, with that? Books and professors, no real world practice until you learn the basics first, that is why premed is 2 years school where they tell you to forget everything you learned in undergrad, then cram you with 60 credits+ a semester, after two years of premed you go into residency, where you practice, yes this is part you are talking about, but you practice based on what you read. When you are a doctor you don't post on forums either.

There is no reason for me to answer any of your questions, as I am not paid to do so. I also DO NOT represent Gigabyte, I post as it comes and as I see fit. I cannot and will not act on behalf of Gigabyte to fight you over this. If you have any other questions feel free to PM me, as you are just bashing me for no reason other than it goes against your product.

your 3 questions, woudl have me break NDA on on 2,3.
The first one is a pointless question.

I took the data I had and made use of it. i did my best to gather all the data I could on digital VRm design, read many articles, yes they were a bit outdated like 2-3 years old, and that is where I took that digital VRM design is infancy. The truth is though that analogue signal is still converted into digital signal and while that may take time, it still has many benefits.

While you represent Asus, I DO NOT represent Gigabyte.

Hey thanks for you opinion! criticism is always welcomed. Even though proof is welcomed as well.

Last i heard Uncle Sam signs my paycheck.

Gigabyte didn't tell me to come post here. Gigabyte's BIOS has a EFI loader already, I think they are working out the bugs so their BIOS doesn't turn out like PC Probe. yea I have owned ASUS boards since 2003. I have been around a long time i didn't just start to promote companies, i woudl be willing to give your boards a shot and give them my attention but don't make them sound disgusting as you are by the low blow of posting in my review.

They already have a unwritten rule manufacturers don't post in each others threads. They definitely shouldn't threaten reviewers with "education" what are you a professor? for tech support. please don't make me hate ASUS boards (because of the people who work for ASUS), i have always liked them, and I didn't say anything bad about digital PWMs you guys just get so offended easily, how do you live with yourself? There are many pros of Analogue VRM design such as efficiency, power output, and easy of implementation, i said good things about Chil PWM and I do like the company. I did base my analogue vs. digital on X58 and below designs, not the new P67 designs, which I am awaiting. Hopefully you guys improved a LOT! that woudl be great and then i can say a LOT of good abotu Digital PWMs. Don't make it hard to like ASUS, PLEASE!!!! I beg you!

This is what gigabyte has in their slides:


This is what I found, is this your education for me:
http://www.google.com/url?sa=t&sourc...9Y7pkFdI3tZQkQ

[dinos22]

hahah Gary you sound stressed mate

[the finisher]

Nice review, nice board!

Think the Asus guy need's to chill.

Have Asus in X48,P45,i5 laptop. Have a few Gigabyte and EVGA mobos as well

[sin0822]

Thanks man!
Yea I still own my P4P800 SE as well as my P5WD2-premium, and even my p5k deluxe, like ASUS you made me not wanna buy anything more ASUS. WHY?

[Leeghoofd]

C'mon guys, both concerned companies have great products... and PR slides are sometimes euh so so... it's good they take their own directions and believe in their products... but don't overdo it...

Could we do an Asus vs Gigabyte cagefight game soon ? I bet my money on Dino, he looks like the Undertaker after midnite and a zillion of brewskis. Think he will Rip Gary and Peter in two :p Scary stuff :p

[zalbard]

Keeping his post count in mind, he surely would.

[flopper]

every motherboard needs new bioses, asus does that also and you got to wonder why the need of a new bios, didnt you do it right the first time?

I love my Gigabyte board, it has a lot of super coppar inside it, I hear asus copied that concept from gigabyte, no wonder as the board runs better, my 1366 board is up and running 2 years and going and been stable as hell, and yes got new bioses for 2 years now, they didnt get it right the first time either.

I didnt get it right the first time I had sex either, never stopped trying tho.

Looking forward the sandy babe line from both companies, and no I dont stop looking for better boards either.

[Silver Bullet]

Great review ... thanks

[sin0822]

You guys are welcome. IDK who has 2 OZ copper first, but both companies have had it for a while! Stack Cool is Asus's type and they have had 20Z copper I think since P45 maybe a little later, but so did Gigabyte with Ultra Durable. Its a great thing to more copper into the board, but they didn't advertise it on their X58 boards which makes me thing it wasn't there, because my R3E and P6X58D-P weighed less than my UD5. Needless to say both companies make great boards, but this is one of my first experiences with Gigabyte and I just flat out love the board and design, hard for any company to beat.

One thing I should say is that DVID is also digital signal and analog PWMs use a DAC so i am guessing SVID is no different, but digital PWms still have 5-6 DACs for other parts of their PWM, hopefully efficiency has been improved as well as overall design. Both Digital and Gigabyte Analogue VRm designs are top notch, just liek I said in my article, idk why ASUS has their panties in a knot.

[chuchnit]

All I know is that I can't wait until NDA lifts so we can see what this new arch brings to the table.

[sin0822]

i am really interested so see how Sandy Bridges AVX instruction set pans out.

[cbutters]

Thank you for the awesome overview of this board! I have some questions for you.
For users who want to get decent CPU overclocks, but will not be using multiple GPU setups; is there any reason to purchase the UD7 over the UD5, UD4 or UD3P versions of the P67A ? I know these lower models have less phases, but will that necessarily translate in to lesser cpu overclocks or general stability? Which board would you recommend for someone who will only be using a single GPU?

[sin0822]

I would recommend the UD4P then, becuase you still have all the phases, but no descrete support for multi-GPU setup. BUT you also lose some features, such as a second USB chip and hub, as well as some extra SATA6G connectivity. Some other small features are missing as well. IDK what max TDP(wattage/power) and TDC(current/amperage) are for the 2500K and 2600K but with every die shrink comes better efficiency less leakage, and lower overall power consumption, probabaly less thani7 900series which was 145amps+35amp for uncore which and 130watt 80% TDP, so I would say get at least a 16 phase board. SHould bemore thanenough power for air/water.

[massman]

A large problem when engineering motherboards that are made for overclocking (maybe 5-10% woudl use DICE,LN2, Phase change and I am part of that 5-10% we aren't many) is that the majority of people are going to use air and water cooling and thus not really need or want to add more heat to things they do not think about like VRM.

Just want to come in and give some HWBOT numbers: 16.11% using LN2, 25.00% subzero (dice, phase-change, ...).

Of course, hwbot userbase is a lot smaller than the total market. Then again, the high-end boards are not made for the mass-market but for this performance enthusiast branch.

[sin0822]

I agree, but I do not agree why make that many motherboards just to give them away, they need to sell them, and they can't just target sub-zero OCers with more than one model. Sub-zero OCing was something MUCH more popular in the past, it is dieing art now, apart from LN2 you don't see that Many SS, no cascades anymore, no one uses peltier nothing, its keeping up with the time. Wait for SandyBridge it will probably die down more, feel free to disagree.

you know I actually enjoy the cooling aspect of OCing more than many other aspects. I loved my SS, chilly1 really knew how to make em, i let it go for too cheap 300.

[massman]

I think the amount of people using subzero cooling is just a function of the hardware's capability to operate in a certain temperature range. The better a CPU or GPU architecture can withstand extreme cold (and scales with it), the more people will be using extreme cooling to get higher overclocking results. With Sandy Bridge not scaling well under cold, I sure do expect the amount of liquid nitrogen users to be lower than on for instance Lynnfield or Clarkdale.

It's also partially about availability. In the old days (untill 3-5y ago), liquid nitrogen was definitly not as accessible as it is now. In the AXP and A64 era (when AMD had the performance lead) you also didn't need that much cooling as most chips stopped scaling around -50°C (AXP) or had an insane coldbug (A64).

The general trend I see in HWBOT, however, is that more and more people are starting to use extreme cooling. Here are some charts I posted in the VIP lounge of the forums a while back:

[chuchnit]

Massman, in the CPU cooling chart, air, both types of phase cooling, and water cooling are either flat or on a slightly downward trend. Yet dry ice has a ever so gentle upward trend and ln2 has a nice upward trend. Is the increase in the user base and submissions the cause of this or am I reading this wrong? Just from looking at the graph I don't see enough change in the other categories to account for the increase in dice/ln2 use. Graphs can be misleading though if you don't pull out the data and do the math.

[massman]

Those charts are quick'n'dirty work, so sorry for the quality. It's also based on amount of results, which doesn't always correlate with actual users (small group of extreme overclockers can produce huge amounts of extreme cooled scores).

Here's the data for unique users



And some useful math:



I'll pull some numbers for GPU as well. Imho, GPU cooling numbers are far more accurate for analysis of extreme cooling usage.

//edit: here they are

[massman]

Figured I should also check the ratio of type_x_cooling and total_hwbot_users. In the above charts one user is in multiple categories, which could lower the relative increase of extreme overclockers.

//edit:

Now charting using 'active users per month' as reference value for # users. More accurate data than above charts!





CPU cooling Figures

[Buckeye]

If I may say, the reason you so so many on HWBOT using extreme cooling is because if you want to be competitive thats what you must do.

I don't think many even post there Air cooling numbers because they would fall so far in the rankings ?

But outside of the bot and except a few hardcore gamers maybe pretty much Air and H2O rule the numbers I would suspect.

[massman]

If I may say, the reason you so so many on HWBOT using extreme cooling is because if you want to be competitive thats what you must do.

I don't think many even post there Air cooling numbers because they would fall so far in the rankings ?

But outside of the bot and except a few hardcore gamers maybe pretty much Air and H2O rule the numbers I would suspect.

You're right, but it's not really relevant to this discussion. This discussion is about the necessity for ultimate PWM designs with regards to product line. As Sin stated, vendors tend to over-engineer the entire product line meaning, for instance, elaborate PWM designs as well as extensive BIOS options. The vast majority of the users will not need that.

The data acquired from the HWBOT database shows a trend of an inceasing amount of overclockers using extreme cooling. Although this is, of course, a market trend in a niche market, it means that there's an increase in demand for products pushing the hardware to the absolute maximum performance, including the elaborate PWM and BIOS. One of the problems with using liquid nitrogen for overclocking is that it cools down more than just the CPU; it also cools down the PWM. Since most PWM lose efficiency below a certain temperature range, vendors need to take this into account when designing a new product and make sure the PWM will not cool down too much.

I think most of the overclockers are aware that they represent a niche market as I've heard more than one person asking a vendor to simplify things and bring one mainboard dedicated to overclockers on the market. The problem is that the resources put into the design of this one mainboard well exceed the amount of potential customers (although that is also debatable); in other words, vendors want to make sure different groups of the enthusiast market buy their high-end mainboard. To do that, they need to add excellent audio features (gamers), fancy color schemes (casemodders) and add an excessive amount of sata, usb and other useless ports. Too much useless features for overclockers, too much bios/pwm engineering for gamers.

As for the positive trend for usage of liquid nitrogen. As mentioned before, the popularity of a certain cooling type correlates postively with the charateristics of the prominent hardware. If the highest performing platform does not scale below -50°C, less people will use liquid nitrogen and more people will use cascade cooling. It's a pity HWBOT has only been around for 4y now and only 1y in its current format because the data <2008 is heavily skewed. If only Futuremark would have listened to the XS community in the past and added cooling fields to the ORB! The data of the AXP/A64 era would, without any doubt, have shown a large majority of the users sticking to cooling types varying between +10°C (chilled air) and -30°C (phase-change) because the most prominent technology back them (Athlon64) had a lot of problems with extreme cold. Liquid nitrogen was expensive, not easy to get and not really worth the time since you could reach the same frequencies with a single-stage or chilled water configuration.

So, yes, people use liquid nitrogen at HWBOT because it's necessary to be competitive. But, it's only a necessity because the technology is tolerating those operating temperatures. Think about it this way: it's only because AMD brought their Phenom II technology, which scaled incredibly with temperatures, that overclockers have used liquid helium.

Sandy Bridge brings us back to the A64-era in terms of cooling usage, although it will not be the platform giving you the highest absolute performance. I'm excited to see what that will bring for HWBOT and other overclocking ranking boards.

[chuchnit]

Great charts Massman!!

[massman]

Great charts Massman!!

You shouldn't thank me, but every single overclocker posting at HWBOT. Without their input, charts like this would be a lot more difficult to build. I really am kind of sad that the data of <2008 is so erratic and hardly useful.

[sin0822]

You're right, but it's not really relevant to this discussion. This discussion is about the necessity for ultimate PWM designs with regards to product line. As Sin stated, vendors tend to over-engineer the entire product line meaning, for instance, elaborate PWM designs as well as extensive BIOS options. The vast majority of the users will not need that.

The data acquired from the HWBOT database shows a trend of an inceasing amount of overclockers using extreme cooling. Although this is, of course, a market trend in a niche market, it means that there's an increase in demand for products pushing the hardware to the absolute maximum performance, including the elaborate PWM and BIOS. One of the problems with using liquid nitrogen for overclocking is that it cools down more than just the CPU; it also cools down the PWM. Since most PWM lose efficiency below a certain temperature range, vendors need to take this into account when designing a new product and make sure the PWM will not cool down too much.

I think most of the overclockers are aware that they represent a niche market as I've heard more than one person asking a vendor to simplify things and bring one mainboard dedicated to overclockers on the market. The problem is that the resources put into the design of this one mainboard well exceed the amount of potential customers (although that is also debatable); in other words, vendors want to make sure different groups of the enthusiast market buy their high-end mainboard. To do that, they need to add excellent audio features (gamers), fancy color schemes (casemodders) and add an excessive amount of sata, usb and other useless ports. Too much useless features for overclockers, too much bios/pwm engineering for gamers.

As for the positive trend for usage of liquid nitrogen. As mentioned before, the popularity of a certain cooling type correlates postively with the charateristics of the prominent hardware. If the highest performing platform does not scale below -50°C, less people will use liquid nitrogen and more people will use cascade cooling. It's a pity HWBOT has only been around for 4y now and only 1y in its current format because the data <2008 is heavily skewed. If only Futuremark would have listened to the XS community in the past and added cooling fields to the ORB! The data of the AXP/A64 era would, without any doubt, have shown a large majority of the users sticking to cooling types varying between +10°C (chilled air) and -30°C (phase-change) because the most prominent technology back them (Athlon64) had a lot of problems with extreme cold. Liquid nitrogen was expensive, not easy to get and not really worth the time since you could reach the same frequencies with a single-stage or chilled water configuration.

So, yes, people use liquid nitrogen at HWBOT because it's necessary to be competitive. But, it's only a necessity because the technology is tolerating those operating temperatures. Think about it this way: it's only because AMD brought their Phenom II technology, which scaled incredibly with temperatures, that overclockers have used liquid helium.

Sandy Bridge brings us back to the A64-era in terms of cooling usage, although it will not be the platform giving you the highest absolute performance. I'm excited to see what that will bring for HWBOT and other overclocking ranking boards.

I do agree with what you are saying, yes extreme cooling is required to be competitive in crowds that HWbot caters to. But HWBot isn't the majority of people who buy the ROG series or the highend BigBAng or UD7,UD9 boards, that number is the real question here. Is it worth it to just market a board to the extreme users? I know you would agree that a perfect benching board would have only native connectivity, no on-board sound, a powerful VRM along with a tolerable PWM as you stated, as well as excellent build, clearance around the socket, as well as extra PCI-E connectivity and spacing. As many would agree this would be the perfect benching board, something that is overkill in the VRM portion, as well as PCI-E. But would a board like that sell in any other market? We have to admit that our species while they come from all walks of life and every corner of the globe, is a rare one at best. Not all of us would buy from the same manufacturer, and I think we could all agree that we aren't normal overclockers we are the extreme (you guys more than me, I just OC to see how high I can go, then I use my computer at stock). manufacturers, need to ask themselves, will this sell in a store? at a local microcenter? will they carry this board? or am I going to have to put it on special order? i think the majority of overclockers, where we come from, use their OCed rigs 24/7. They use them for folding, as well as for many other tasks that can benefit from higher frequencies. Those average OCers don't have multiple motherboards for the same platform, nor do they own every platform. Money is a huge issue!! LN2 costs money as well!!
A single stage would be the maximum any normal overclocker would use(practicality), and I am sure many of you can admit that at least 80% of overclockers have never dabbled in extreme cooling. This would have been different back in the day of P4s, but now after the OC boom, overclocking has become comercial, before we were lucky to get a 4-8 phase board to take our P4s to A64s to 5-6ghz. back then those P4s ran just as hot(P4 561 @ 115watt TDP) as our current i7 900 series chips(130watt tdp), if not hotter. I remember chilly1 built me an SS that would do 220watts at -48C, and low and behold -48C was my load temp at 5.3ghz. Now we have chips that are more power effiecent, yet we have motherboards that put put over 1000watts of power! I think we have been catered too. Before we couldn't control the PWM, with digital systems you can now. again we have been catered too. We can only ask and hope for so much.

Now the choice is do you want a VRm that can push a lot of power at maximum, but also be very effcienct when at 1/8th its load? or do you want a VRM that is off the bat more accurate and controllable. The way Gigabyte has setup their VRM is so that it is extremely efficiecnt, voltage control is almost as good as a digital PWM could provide, and will supply more than enough power to the CPU. In many ways I see no need to control an analogue PWM the way you control a digital PWM, I don't see a need to mess with it. Never helped my OC other than putting frequency at maximum. I do see a need to mess with the switching frequency of a digital PWM to make sure its high when OCing and low when not(that is a luxury). There are a lot more facets to the differences between the two system, but in the end, can a UD9 with its 24 phase power take a i7 980x to the WR for a 980x? yes hicookie did, can a R3E take a 980x to the same WR and beat it by a few mhz, yes it did, at that point both VRM systems are able to achieve the same thing, the WR, whether one did it first or not doesn't matter, what matters it both can do it and have done it. The classy is up there too if not the best digital PWM/VRM design as you guys are saying, i do trust you.

While we know that analogue PWMs are trustworthy and do their job excellently is no argument, but digital PWMs? the truth is they have come a long way since(skip audio equipment) that first ABIT board, and then DFI and then everyone followed, but we did take a step back and say is this really better than an analogue vrm?

I think manufacturers came to the conclusion that either one can do the job correctly, yes there were some problems with early versions of Digital PWMs but this was one of the problems with having firmware drive the VRM system. Everything new has problems, and now that those problems are over I think that Digital PWMs will be come more and more common throughout computer systems. I feel as though they need to be proven first. Overclockers initially liked them because they allowed clearance around the socket, much easier to insulate, you could use rubber eraser instead of di-electric grease, or both of course. They were new and more accurate, handle the load line equation better and everything.

Now Analogue PWM design has come a long way, 6.25mv tuning compared to 5mv on digital PWM, no transient spikes(on (24 phase analogue(from GB slide)) that plagued LLC and scared the living crap out of normal OCers. The truth is Digital PWMs still control MOSFETs and MOSFETS(including low RDS (ON) MOSFETs and Driver MOSFETS) are always going to be a part of the VRM design, so are core chokes, and capacitors(or one cap), and I think the quality of those components are more important now than ever, i feel as though both analogue and digital PWMs can get the job done, now what we need to focus on is what are the good components and who builds em the best?

I think sandybridge is going to change a lot, I feel like maybe temperature is going to have much less effect on OCing, I could be wrong, I haven't had much luck.

All I can say is good job to the many manufacturers who provide us with boards and competition so the prices of these boards don't go through the roof!

Massman thank you for going through it with me, I do respect your views and the way you put everything, it is very insightful! Thanks! It is nice to have discussions over things like this other than straight up insults and trying to fight each other. For me this is a hobby and I love doing it, as I know it is for many of you, but it still is the internet and some people are just angry as hell.

Everyone have a wonderful Holiday Weekend!!!!

Merry Christmas and Happy New Years!!!