It is time to introduce the fine print and disclaimers when using these boards with a Phenom 9850BE. As a recap, the critical aspect of running a 9850BE at stock speeds on current 780G products centers on the cooling of the PWM circuitry. Each and every manufacturer along with AMD agreed that cooling the MOSFETS properly was critical to the successful operation of the board at stock or overclocked speeds with the 9850BE - and to some degree, the 6400+ X2.  This is based on the board manufacturer utilizing a properly developed power delivery system that is designed to handle the 125W TDP processors and in the future, the upcoming 140W TDP Phenoms.

We still do not think the majority of users who purchase these boards will end up installing AMD’s top of the line CPU. Personally, we would not install a Phenom 9850BE or 9750 retail unit in these boards considering the vast majority of applications these boards will see can get by fine on an Athlon 4850e or Phenom 8450. However, for those who wish to continue down this path we have a few suggestions.

Everyone probably knows the old saying about the three rules of real estate, location – location – location. It holds true with our board samples only replace location with airflow.  We found additional airflow around the PWM/MOSFET area was critical when utilizing the stock retail fan/heatsink in our Silverstone SG03 case. We installed a secondary 120mm fan that provided enough airflow over the board to ensure stable operation under our load scenarios. While that was acceptable for the most part, it did increase noise levels a little more than we expected. In the end, we ended up replacing the stock case fan with one that offered slightly higher airflow with only a small penalty in acoustic levels.

 

Our other solution was to replace the stock heatsink unit with a radial design that improved airflow around the CPU area to the point where the boards did not experience thermal related shutdowns or catastrophic failure when we played “Pop goes the MOSFET”.   We found that increasing our case airflow provided temperature drops in the PWM/MOSFET area about 5~11C better than the radial fan design. However, our processor temperature dropped about 3C at idle and averaged 9C lower at load when compared to the stock heatsink. Additionally, acoustic levels dropped on average about 4dB(A) with the radial fan setup. 

Utilizing our standard test bed components with the 9850BE, we ran our Jetway, ASRock, and Gigabyte boards over 102 hours each with a combination of OCCT, Microsoft Flight Simulator X, and PCMark Vantage benchmark loops. We did this with the case fan modification first and then again with the Thermaltake RubyOrb installed.   We did not have any thermal related failures or shutdowns under load as we had experienced with just the retail heatsink installed in previous testing. We also overclocked the 9850BE slightly to 2.7GHz and ran an additional 48-hour combination of OCCT and PCMark Vantage loops without a problem utilizing the two front-mounted case fans and retail heatsink.

Our Biostar and Sapphire boards both passed the 102-hour test. We did experience thermal related shutdowns with OCCT around the 40-minute mark on both boards with just the retail heatsink and standard case configuration. This also occurred around the two-hour mark with PCMark Vantage. In fact, we had Biostar send us another board after a bad overclocking experience (our fault for trying) with the 9850BE on an open test bed setup. We did not test either board in an overclocked condition but feel safe in their capabilities to run the 9850BE at stock settings with proper cooling.

Our results just reiterate that adequate airflow is required when utilizing a 125W TDP processor on these particular boards that are qualified by the manufacturers to work with the Phenom 9850BE or retail 9750 (OEM version is rated at 95W TDP) processors.   At this time, AMD has not qualified the Jetway, ASRock, Gigabyte, Biostar, or Sapphire 780G boards to operate properly with the 125W TDP processors. Please keep this in mind when considering one of these combinations as only the manufacturer has stated compliance or in the case of Biostar and Sapphire, implied it.  That said, we have finally completed our testing and will have performance results from these boards and others in the near future.

Index
Comments Locked

21 Comments

View All Comments

  • ZootyGray - Monday, July 21, 2008 - link

    I have to say thank you for these articles on 780G manufacturers and cpu support lists. I have never seen such beautifully brutal testing. I have endless respect for real truth. And there is a strong consumer protection thing here too. Reviewing is not about being nice - it's about accurate info, and good journalism.

    I have been burned in the by bogus mobo manufacturer's claims and, even worse, ommisions. My suffering was also facilitated by rosy test reports. I eagerly bought hook, line and sinker and 2 days over the warranty period, I discovered I had been sold old tek. (The mobo was ABITch. The review was on toms hdwe guide.) The problem was the MS announced chg on acpi which brought 24+ irq's into play with the advent of winXP - my board was released after the announcement by Microsoft - it was a tek dump. I told them I knew. I told them I would put the word out. ABITch tek sport - no speakadee eenglish.

    So I was pretty much ready to run out and buy a 780G board. I planned on a small cpu for now - and a big Phenom later. Reviews were good everywhere I went - and then boom - pop goes the mosfet.

    I have been referring a lot of people to come and read up on the real story. And like others have said, it is really too bad that it appears as if AMD did this - rather than this being a mobo manufacturers design issue. And the whole issue of vague or absent CPU Support Lists.

    And you know it has not changed very much - some changes but a lot is still the same.

    But thanx for brilliant testing so far. More to come. This isn't over.
  • ZootyGray - Monday, July 21, 2008 - link

    Well, you said this .......

    (quote)
    That said, we have finally completed our testing and will have performance results from these boards and others in the near future.
    (quote)

    The near future has arrived. It's update time.

    I have been informed of changes by asus - apparently they have some new models in their 780G line of mobos that support 140 WATT CPU'S !!! They have some others that don't seem to have this 140w sticker rating also - and I found it hard to get EXACT CPU INFO to qualify for myself, the nervous consumer. This should be easy for me as a total idiot consumer! Why do I have to learn more tek than an engineer to make sure I am not being conned? Isn't that the bottom, line here? I would not buy most things in such an atmosphere of danger!!

    Anyway, 140 watts MIGHT be enough to handle an oclox Phenom 9950 Black Edition. Apparently also - 140 watts is the realm of the not yet released DENEB cpu. And yes, it begs the question - will the higher or oclok DENEB play pop? :) It seems they use a 4+1 phase.

    It seems the newer 140w mobos only appear on the International website - not on USA or Canada sites (yet?) (WHO KNOWS?). In fact there are no 780g boards on the Canada site at all (at this writing) and a search reveals nothing - no results! Maybe they were never there? - this makes me paranoid. There was a press release in June announcing these boards - about a month ago?

    Biostar has an interesting board with a "space pipe" heat pipe that connects various components including the mosfet area. I thought that was a really unique idea. PLUS BIOSTAR CLEARLY STATED 125 watts as ok on 1 model - IMPRESSIVE :)

    I, personally, am unable to assess clearly what Gigabyte is doing - although I have seen a version 1.1 with no explanation. They WERE my first choice - now, I don't know.

    My experience with MSI is that I have too many unanswered questions, and reading for answers brings only frustration in the face of more vagueness. And what seems like a language barrier.

    And I see a 3-year warranty offered by asus - but I really would prefer to avoid RMA costs and delays and days of computer down time. Has the problem been fixed? Is it safe to play now? Was there ever a problem? Feed me some honest information; because I am very hungry.

    The mobo co's seem content to indulge in self-promotion rather than really taking this issue by the horns and dealing with it completely.
    It would be oh so nice if I read that xyz mobo company admitted fault or problem and identified the exact problem and provided a clean solution that was crystal cleat to me, the average idiot consumer, and that enabled me to relax and trust these scared lil babies with my money; so I could simply buy the board and processor that I want.
    And everybody lived happily ever after and there was infinite peace in the universe too.

    Hey - what's the problem with that? How is it that up-front honesty is such a great difficulty in our beautiful world. If the rules of the game were clear, clean, fair, and announced up front - would that make it all just easy?

    Until then :) it would be wonderful to see a follow-up article to the initial brilliant reports.

    Thank you once again.
  • ZootyGray - Monday, July 21, 2008 - link

    Since I am on such a roll here - (rolling up the number of comments from 18 where it has been stuck for too long) (yes, I check back often)

    I further wish to compliment you on general accuracy of reporting. Unlike some (cough tom cough) sites where blatant bias seems to be a sick joke, you here at Anandtech seem pretty clean and factual.

    Recent example re Ruiz resigns was simply stated. Also the Q2 reports. That other site treated these issues with a lack of respect and furthermore disguised their stupidity by labelling an article "opinion".

    In my search for accurate information I have no use for damage communicated through blatant bias in the face of facts, no use for opinions because I am seeking truth and guidance - and will continue to attend, and refer people to, this site where people seem to have the simple courage to state truth.
    Sometimes all we have is opinion - that's ok - but when opinion is a cover for gloating over the difficulty of certain events - well, that's just trolling and flaming, and is indicative of more and worse.

    I am wandering off the good topic here.

    Peace.
  • dancinguy - Wednesday, May 21, 2008 - link

    Where is the full review of the 780g motherboards? I need more info before I buy!
  • Tarantella - Saturday, May 10, 2008 - link

    Currently, at Jetway and Gigabyte, their cpu support pages show otherwise for the 9850 phenom.

    http://www.jetway.com.tw/jw/support_motherboard.as...">http://www.jetway.com.tw/jw/support_motherboard.as...

    http://www.giga-byte.com/Support/Motherboard/CPUSu...">http://www.giga-byte.com/Support/Mother...d/CPUSup...





  • avi85 - Sunday, May 11, 2008 - link

    Hey!!! There is now a Rev. 1.1 of the gigabyte, http://www.giga-byte.com/Products/Motherboard/Prod...">http://www.giga-byte.com/Products/Mothe.../Product... anybody know what the differences are besides what I found while comparing:
    1. different shaped (larger?) NB heatsink
    2. NB fan header (although no fan on the new heatsink)
    3. no TPM header on the new one

    Does anybody know if they changed the power system???
  • strmbkr - Saturday, May 3, 2008 - link

    Some 780G models costs around the same or more than the full fledged "mid range" motherboards that could support the 125W AMD dual Core and the Quad Core (example Asus M3A vs M3A78-EMH HDMI); so, why it can't support the 125W? heck... even the M2A-VM (690G) motherboard can support the 125W dual core with no problems.

    Building a file server with these boards are cost effective, since cpu power, memory and HDD is more important, and you don't need a good graphics card, then why not pair it with a good cpu when building these basic file/data servers?

    The third point is AMD's announced intent for easy upgradability of the cpus. In the future, the later AM2+ and AM3 cpus will have higher voltages; wouldn't that make AMD's roadmap of "easy upgradability" moot?
  • talos - Friday, May 2, 2008 - link

    The Sapphire and Jetaway 780G boards look the same. Only the MOFSET coolers are missing from Jatway.
    Maybe they are crossflashable.
  • bgd73 - Thursday, May 1, 2008 - link

    A cpu with that much going on is doing more than frying mosfets. Is all old school rules deemed retard by a retard? put it on a bigger board, get a case big enough for a large heatsink and reverse cool precisely. The problems these monkeys are having is over 6 years old, with the first of the hottest cpus known to man, the p4 e line,and hardly a problem to the rest of builders out there with the given tdp already. Oh the pride of closed eared engineers is going to kill somebody.Gov't anybody? never thought I'd be scared enough to request that one...
  • bradley - Thursday, May 1, 2008 - link

    I've never heard of a mosfet going pop... they can crackle, and in instances catch fire. I'm talking about running poorly cooled and/or slightly out of spec, as opposed to extreme overvolting. Capacitors can pop, or even go bang. Although breakaway tops are meant to limit the explosive activity somewhat. Of the two AMD related stories: the Phenom TLB and 780G compatibility, both seemingly have been borne out of an overemphasized misunderstanding. This then spread like Internet wildfire and clearly hurt AMD as a corporation.
    A cynic might even call them hatchet jobs, but the intent is unclear. And I'm not even bothering to include the senseless and irresponsible writings of Ed Stroligo. :) I have seen on occasion potential customers forgoing this great chipset based on information propagated from these series of articles.

    I believe a collective urging for rigid verification and implicit compatibility standards would be the more genuine and germane angle. Manufacturers now have the tools to virtually design these boards within very exacting tolerances. This includes them using less efficient [and thus less costly] parts, which would convert more power to heat. It probably allows manufacturers to build within a very specific price point. I'd imagine you could have a 4-phase design run more efficiently than a poorly designed 8-phase. Less efficient parts, in turn, dampen the function of that very powerful CPU, as would a smaller form factor limit the threshold of current. Otherwise we will know what to expect when 140 Watt Phenoms hit the 780G scene, more of the same.

    Or maybe there needs to be a better understanding of electrical engineering and design. Sharing this inside knowledge right from manufacturer's own mouths would make for very illuminating articles. Perhaps start to list part numbers in these reviews so educated readers can do their own math. Let readers know what makes these boards tick, and the part specifications that allow: memory, cpu, videocard, storage shine. I want to see a greater accountability and responsibility from manufacturers and enthusiast writers alike. This is no longer just some hobby or preoccupation, but big business that affects many groups of people.

Log in

Don't have an account? Sign up now