Tech View: Motherboards Galore

by Gary Key on 5/3/2007 3:00 AM EST
Comments Locked

33 Comments

Back to Article

  • DigitalFreak - Friday, May 4, 2007 - link

    Where was the board with the Via chipset? I needed a good laugh.
  • danielackerman - Friday, May 4, 2007 - link

    I dont understan why anand hasnt done any good reviews on amd based mobos. I dont own intel, ive never owned intel, i will never own intel. there are many like me. there are many who love this website.

    PLEASE BE A BIT MORE FAIR AND BALANCED. less monopoly of intel board reviews and more skinny on amd based boards please.

    thnax
  • strikeback03 - Friday, May 4, 2007 - link

    umm, not much has happened on that front since the launch of AM2. They had a review of the new AMD integrated graphics, I'd expect more with the uATX review whenever it comes. Otherwise I'd expect more motherboard tests after the Barcelona launch.
  • kilkennycat - Thursday, May 3, 2007 - link

    As for many others not wishing to upgrade to a new motherboard every year - in spite of the money that activity makes for the MB manufacturers and the chip-set suppliers, grrrr... --- compatibility with upcoming processor-families becomes a most important decision-making parameter. In the case of the Intel motherboards, the obvious candidate is the desktop version of Penryn. Considering the motherboard voltage-regulator fiasco with the move from P4 to Conroe, it seems that there could be a repeat of that fiasco, or a chip-set incompatibility fiasco (as with the 715/725 and dual-cores) with the move to Penryn.

    For example, I expect to build a new PC in the Fall this year. I would like to buy a motherboard with a few months of production and BIOS-updates "under its belt". I will probably initally invest in a fast dual-core Intel Conroe system ( if AMD does not pull a rabbit out of the hat in a month or two ) and potentially later upgrade to a Penyrn quad-core when the "price is right". Replacement of my just-purchased motherboard to address that latter contingency is just "not on the cards"..... However, for me, a BIOS update to an existing MB is perfectly acceptable in the case of a Penryn upgrade.

    I'm sure that Anandtech has sufficient clout to spring free a beta-phase quad-core Penryn or two and some MB Alpha-BIOS updates to verify Penryn-compatibility in your formal reviews of the Intel-compatible versions of these new-generation motherboards.

    Of course, if you could also spring loose some desktop K10 AMD CPUs to verify AMD motherboard compatibility for their upcoming desktop-CPU family, I am sure that you will make some more of your readers eternally grateful.
  • kilkennycat - Thursday, May 3, 2007 - link

    ... sorry line #4, 915/925, not 715/725....
  • mattt79 - Thursday, May 3, 2007 - link

    quote:

    We will also be introducing our Vista and Linux benchmarks for our mainstream motherboard reviews.


    Finally! Could you also mention possible compatability problems... such as the JMicron EIDE controller issues that almost all of the 965 boards have?

    Thanks,
    Matt
  • strikeback03 - Thursday, May 3, 2007 - link

    the JMicron controller is supposed to work with Linux kernel 2.6.18 and up. I can't give any firsthand experience though, as Ubuntu 7.04 does not like my card reader or wireless card and I have not gotten time to actually install and try the JMicron.

    Any tests on whether those heatpipe chipset coolers have issues when using a good CPU cooler, such as an Ultra-120 or Tuniq, that move some airflow away from right at board level?
  • yacoub - Thursday, May 3, 2007 - link

    The cooling on the MSi P35 board looks like a friggin' rollercoaster! :D
  • sprockkets - Thursday, May 3, 2007 - link

    That Albatron board has been replaced by a newer version, but good luck finding it.
    Also, someone made a point elsewhere that again, Intel just made the current chipset obsolete in 6 months, again. Not a big deal for the end user though.

    That Sapphire board? Uses Teapo caps and like 2 mosfets per a channel, that's absurd for a near $100 board. Take a look at the abit nview and see how it has Rubycon caps all over the board, bearing much higher quality.
  • yyrkoon - Friday, May 4, 2007 - link

    Yeah, I was about to write a post saying that the Saphire board looks to be a knockoff of the MF-M2 nView, which we have two here on premisis (I own one, and I love it)

    Good looking out on the caps/mosfets, I did not notice that myself :/

    On a side note, someone needs to inform mini-ITX makers, that socket 754 is pretty dated, and time to move to more availible/inexpencive CPUs . . .
  • baronzemo78 - Thursday, May 3, 2007 - link

    Any news when the new Bearlake X or any boards that have support for dual x16 pci express electrical for crossfire?
  • kmrivers - Friday, May 4, 2007 - link

    Actually I found some stuff out today. The 975X replacement will be the X38 boards.

    The intel version, which is a direct replacement to the Bad Axe 2 is the Intel DX38BT 'Bone Trail
    Source: http://www.thetechrepository.com/showthread.php?t=...">http://www.thetechrepository.com/showthread.php?t=...

    Here are some details on it:
    DX38BT Product Highlights:

    * Based on the next-generation X38 Express chipset (MCH)
    * Intel Duo/Quad/Extreme (dual and quad) Core support at up to 1333MHz FSB, dual channel DDR2/DDR3 memory support with Fast Memory Access technology
    * Intel ICH9 southbridge, 12 USB ports, supporting up to 6 additional PCI Express 1.1(a) lanes [physics card upgrade support], GBit LAN
    * 2x PCI Express 2.0 16x ports (backwards compatable with PCI Express 1.1 specification), Crossfire is supported
    * 6x SATA 3.0GB/s ports, Intel Maxtrix Storage Technology (RAID)
    * Intel High Definition (HD) Technology
    * Full overclocking/overvoltage options via BIOS and NEW Windows GUI utility for overclocking, including auto-tune capabilities


    I have no idea where TPR got this information. Here are some details on the Bearlake chipsets:
    http://www.dailytech.com/article.aspx?newsid=4588">http://www.dailytech.com/article.aspx?newsid=4588
  • Chadder007 - Saturday, May 5, 2007 - link

    PCI Express ....2.0? Good Lord, are they going to pull another AGP on us again???
  • vailr - Thursday, May 3, 2007 - link

    I'd like to see a short summation of P35 boards having:
    1. No legacy parallel port
    2. 4x DDR2 memory slots
    3. Minimum of: 3 PCI card slots
    4. Intel Matrix Raid
    The 2 P35 boards known to have these 4 features are:
    Asus PK5, and variants
    Abit P35, and variants
    Any others?
    And, will DFI delay another 6 months (like they did with their just-released 965 Infinity board) before offering their version of Bearlake board?
  • Mogadon - Thursday, May 3, 2007 - link

    While we're on the subject of heatsinks, it seems unfortunate that Gigabyte decided to use exactly the same heatsinks on their DS3R as the DS3, considering everyone that's tried to overclock with a DS3 knows how hot they get.
  • mostlyprudent - Thursday, May 3, 2007 - link

    I was glad to see a new DFI borad, but very disappointed to see it, along with the MSI 680i board, had a fan on the chipset. Come on guys...HEATPIPES! Perhpas the fans allow for better O'Cing, but having just replaced another dead one - I'm done with boards that have chipset fans!

    BTW, nice article. I appreciate seeing what's coming up. With so many paper launches, it's very niec to know that AT already has multiple P35 boards for testing.
  • lsman - Thursday, May 3, 2007 - link

    I will prefer a chipset fan.
    because now, I have to find some 40mm fan to screw on those "$100" mobo to OC them. The NB is very hot.

    And there are many cases that ppl take their heat pipe off just to find out that only 1/3 of the chipset is cover with thermal paste.

    http://www.overclock.net/faqs/100443-how-reseat-as...">http://www.overclock.net/faqs/100443-how-reseat-as...
  • TA152H - Thursday, May 3, 2007 - link

    Also keep in mind this chipset is made on 65nm lithography, not 90 or 130 like the P965 and 975 respectively. It shouldn't give off as much heat.
  • mostlyprudent - Thursday, May 3, 2007 - link

    Interestingly, most (if not all) of the complaints I hear about passive cooling solutions on MB chipsets are along the lines "that heatsink was really hot to the touch". The temperature of the heatsink is not necessarily indicative of the temperature of the underlying chipset. Stick a little fan on top of the heatsink and:
    1) you won't be able to feel how hot the heatsink gets anymore because you can't touch it so easily;
    2) the surface of the heatsink will likely be cooler with the fan running, but that does not necessarily mean the chipset is being cooled any better; and
    3) as we've seen in the recent CPU cooler reviews here, there is not always a direct correlation between temperature and overclockability (at least not a direct linear correlation throughout the range of temps and mhtz).

    Besides, we've seen from ASUS and others that their passively cooled boards have no problem keeping pace with boards that sport chipset fans. I suspect the primary motivation for using these little fans is that the chipset fans are a much cheeper cooling solution (from a design standpoint).
  • Stele - Thursday, May 3, 2007 - link

    Passive cooling solutions have the significant advantages of being perfectly silent and, having no moving parts, are practically immune to mechanical failures - and the thermal failures resulting from that. Heatpipes, in particular, are also far more effective than plain copper heatsinks in removing heat from the component(s) - they generally have an effective thermal conductivity up to thousands of times that of solid copper of equivalent cross-section. Furthermore, heatpipes allow heat to be quickly and efficiently transported from the source of the heat (usually chipset components tucked amongst many other components/expansion cards) to another location where dissipation/cooling may be easier (e.g. in the path of airflow from CPU fan and/or in a more open space where better airflow can be obtained).

    If the component(s) become exceedingly hot, one possible and typical cause is that the manufacturer did not deploy a solution sufficiently capable of handling the heat dissipated under certain circumstances (some manufacturers may design their solutions with a 25-30°C ambient temperature in mind, vs. the 30-40°C more commonly found in packed casings in say, warm climates). The other reason is that some manufacturers have a target spec - say, keep NB operating at x°C or less under ambient temperatures of say, 30°C - and then go on to spec a fairly high temperature for x - say, 60°C - so that the NB operates under the maximum allowable temperature (even if barely) yet ensures that the cooling solution needed would be inexpensive and simple.

    Both these are mistakes/negligence in the engineering - under-specifying the desired cooling performance of the cooling solutions to be used - and are equally possible when designing in HSFs too, so it's a bit unfair to say that all overheating proablems are entirely because of the use of passive solutions per se. Same goes for heatsink seating issues - applies to both HSF and passive solutions and depends more on operator training/care than actual heatsink used, though a little more is required for exotic heatpipe system installations than simpler passive solutions.

    All told, however, a passive cooling system still has the edge over traditional chipset fans, which is why more and more motherboards are deploying them instead of tiny screaming meemies. :)
  • michal1980 - Thursday, May 3, 2007 - link

    what was the point of posting this. to spit in our faces or something?

    'We know something you don't, nanananananana'

    thats all I hear from this article.

    should have just waited until you could have said something that was more usefull then a press release from a company.

    plus over at hardforums, they are saying the p35 were moved up 2 weeks from may 21st to like the 7th.
  • xsilver - Thursday, May 3, 2007 - link

    and why are you posting here?

    "I'm the biggest whinger here,,, nannanannana" ???

    1) AT tries to pride themselves on being the best tech site out their, hence they usually need extra time to post articles or get very little sleep (anand/wes/derek/gary can probably speak for themselves)

    2) In order to stay "bleeding edge" some "preview" articles may need to be written so that AT doesnt appear to be last on the bandwagon.

    satisfied?


    on a more pleasant note can anyone confirm whether this launch date of the 7th/21st is going to be a hardlaunch and some indications of what prices are going to be?
    what about ddr3 availability?
  • Wesley Fink - Thursday, May 3, 2007 - link

    According to Intel, the 7th is an OEM "stocking" date and the NDA lifts on the 21st. In fact Intel has not yet sampled their own versions of the Bearlake boards and the 21st NDA applies to people like Asus/Gigabyte/ECS/MSI/Abit/DFI/etc.
  • yacoub - Thursday, May 3, 2007 - link

    Dude, just use "whiner" like normal people. "whinger" looks so smarmy and a bit too much like Winger. =P
  • michal1980 - Thursday, May 3, 2007 - link

    bleeding edge does not = press release.

    or wait, is every site just a shile for the companies making the product.

    At least label it as such. 'non paid ad for companies xyz'.

    And they threw that line in about performance them selves.

    Thats where the nannannana comes from.

    "Surprising .. performance.. results... can not saying anything ethier way, or hint"

    NANNANNAANA We know stuff.
  • Boushh - Thursday, May 3, 2007 - link

    Besides the normal test you do, could you also check if coolers like the Thermalright Ultra 120 fit on the motherboards ? I think that many people that fit after market cooling solutions are interested in this information. Specially the MSI P35 Platinum has some odd looking contraptions in the CPU area...

    Besides that I'm still disapointed that so many manufactures have such a poor layout of the expansion slots. Specially the PCI slots. I personally need a minimum of two. And I also no the my gfx card will be a double slot one. I also want at least one slot between the gfx card and the other cards. That does not leave me with much options.

    I'm disappointed that many manufactures are going for dual gfx solutions whil I recon that only a few people actualy make use of it. The dual PCIe slots eat up room. And many boards that do have a decent layout miss extra features (many current P965 boards with a decent layout miss the ICH8R). Some how people that want all the options but no dual PCIe slots messing up the layout are left in the dark.

    Maybe something you could mention to the manufactures ?
  • Stele - Thursday, May 3, 2007 - link

    quote:

    "...Gigabyte's Ultra Durable 2 technology. This technology is a combination of ferrite core choke coils, low RDS MOSFETs, and solid aluminum polymer capacitors..."


    Actually it's more of marketing hype - making mountains out of molehills as is unfortunately often the case with some manufacturers, who try to capitalise on the fact that your average PC buyer is not electronically-inclined.

    Choke coils on motherboards, for example, have generally be wound on ferrite cores for the longest time! Same goes for 'low RDS MOSFETs'... by the way, it should be 'low RDS(ON) MOSFETs', which means MOSFETs with low Resistance, between their Drain and Source, when they are in the 'ON' state (i.e. when current is flowing through them). This allows minimum obstruction to current flow, which translates into higher efficiency and lower power dissipation (the higher the resistance, the more power is lost as heat). Again, MOSFETs with low RDS(ON) have been pretty much standard on decent motherboards for some time now... the difference is that top manufacturers use MOSFETs with very low RDS(ON), which are of course better but cost more.

    As such, these claims are about as big a deal as saying that the new motherboards feature onboard 10/100Mbps NICs and onboard audio....

    To be fair, Gigabyte probably meant that they're using higher-quality ferrite-core coils, and especially-low RDS(ON) MOSFETs. However, from experience, it'd most likely turn out that the components they use aren't super-outstanding as their marketing hype makes it out to be.
  • TA152H - Thursday, May 3, 2007 - link

    Gary,

    There is a slight contradiction in your article. On one hand you mention how people should be interested in the motherboards you're testing based on the old chipsets, and then you mention how surprising the P35 (they are beginning to sound like fighter planes!) was for you. Hmmmm, if you were buying and you read something like that, how inclined would you be to buy the current stuff? Agreed, most people don't need the latest and greatest, but with a motherboard it's often worth it because of the technologies it supports. It's not at all clear the P35 will be expensive, at least I've heard nothing about that (maybe you know the prices better though), although it is very clear the performance is outrageous, particularly paired with DDR3. I wouldn't even consider a P965 anymore, and never would consider that rubbish NVIDIA sells at all. I think most people, particularly people reading this site, are going to feel the same way. DDR2 was a crappy technology, but from all indications DDR3 is a pretty good one, and it's becoming common knowledge the P35 is going to be the next "BX". It's worth waiting a few weeks. But then again, outside of companies like Intel and Supermicro (and even with them, to some extent) you take a risk when something new is introduced. So, that's something positive for the older stuff; you can buy stuff from second rate companies like Asus and Gigabyte and they'll probably be pretty stable by now. Buying a P35 from them would entail a lot more risk, so I guess there should still be interest in the P965, particularly since Intel and Supermicro can be quite expensive, and generally do not have the best performance.
  • ss284 - Thursday, May 3, 2007 - link

    There have been plenty of performance reviews of the bearlake platform. Most of your post is pretty much useless.

    http://www.tweaktown.com/articles/1062/1/page_1_p3...">http://www.tweaktown.com/articles/1062/...earlake_...

    Xtremesystems forum also has some ddr-3 testing.

    The point is, performance wise bearlake will make little to no difference. Maybe 2-3% at most. ddr-3 performance suffers from high latencies and poor initial performance, much like ddr-2 when it came out. The chipset itself will be more expensive than the p965, which isn't exactly cheap right now.
  • TA152H - Thursday, May 3, 2007 - link

    Are you crazy? So far no chipset has been able to show broad superiority over the P965, not anything from NVIDIA, or even the 975. They win at some, and lose at others. Now you see a preproduction motherboard, that is far from running optimally, and it's already outperforming it in every way! No, that's not impressive. Even with final silicon, you see improvements in performance just from changes in the BIOS, so to beat a mature design so early on is very, very impressive. When the intial boards come in, the difference will be pretty startling, unless you really have no perspective. Also keep in mind that improving performance on something with a 4MB cache is a lot harder than improving it on something with a 512K cache. Hopefully they'll test the Core 2 with the smaller cache here to show the memory controller differences a little better.

    I'm not going to go into all the technical merits of DDR3 because there are enough articles out there that do. It's what DDR2 should have been. DDR2 sucked did nothing but disappoint, and I'm still scratching my head as to why they released this half-baked technology instead of going straight to DDR3 like AMD wanted to. Then again, we saw RDRAM for the Pentium III, so weird things do happen.
  • defter - Thursday, May 3, 2007 - link

    quote:

    DDR2 was a crappy technology, but from all indications DDR3 is a pretty good one


    Why do you think that DDR2 is "crappy" and DDR3 is much better? Both introduce higher bandwidth with the same or slighly higher latencies (compared to previous generation) and lower power consumption.

    quote:

    and it's becoming common knowledge the P35 is going to be the next "BX"


    Biggest advantage of BX was its longlevity, with adapter you can use socket-370 CPUs in slot1 BX motherboard. Current P35 boards will be obsoleted in the next year (or early 2009) by Nehalem based CPUs that will require a new socket.
  • TA152H - Thursday, May 3, 2007 - link

    The strength of the BX was the incredible performance difference it had, even over succeeding generations. Intel came with the 810, 820, 840 and then 815. Only the 840 could touch it, and then in only some benchmarks, and this while using super expensive Rambus in dual channel. Even after it was discontinued, people clamored for this chipset, however obsolete it was, because of the crazy performance of it. VIA couldn't even approach the scores of it either.

    Any chipset from that time could be fitted with a Slocket and used with later processors. There was nothing unique about the BX from that perspective. And it was replaced just like Intel does with all their chipsets, and even made so it couldn't use the newest processors (remember FCPGA2 and the Tualatin and Coppermine-T). People found ways of using it because the performance was so good. Probably some of it had to do with politics though, Intel may have slowed down the 815 so it wouldn't hurt RDRAM sales. Why else would it be slower than the BX?
  • markopolo - Thursday, May 3, 2007 - link

    „… On May 21st, Intel will launch the P35 chipset …”

    Intel has brought forward the launch of its Bearlake chipset and, from May 27th to May 9th.

Log in

Don't have an account? Sign up now