Almost as soon as the 875/865 chipsets were introduced last year, we were dealing with news stories about PAT being implemented on the 865 chipset. This Intel "875-only" feature was quickly and cleverly being introduced on the mainstream 865 chipsets by the Taiwanese motherboard makers. This was not what Intel had intended, but it was a testament to the engineering capabilities of the major board manufacturers.

The just-introduced Intel 925X and 915 chipsets seem destined to continue the tradition of Intel trying to limit the chipset features. This round, the hot topic is Intel's attempt to limit or lock-out overclocking on the new chipsets, and the clever engineering that companies, like Asus and Abit, are doing to get around this new issue.

Our early testing confirms what you may have already heard -- Intel has limited the overclocking on their new chipset to about 10%. In looking at boards from smaller or less savvy board makers, you simply can't overclock beyond 10% over specification. We first became suspicious of a lock because some Prescott chips are already overclocking as much as 50% and higher on Intel 875P systems. Given that fact, why should these same cores suddenly be such poor overclockers on 925X/915? The answer, of course, is that the chipset is different and it handles overclocking differently.

When an attempt is made to boot at more than about 10% on a 925X or 915, the system simply reboots or shuts-down. We have not been able to get any kind of official explanation from Intel, but board makers tell us that Intel has added an overclocking limiter that resets a PLL and reboots or shuts down the system if overclock attempts are made at speeds over about 110% of specification.

So, are we stuck with 10% overclock limits if we choose the new Intel chipsets? Fortunately, some of the major manufacturers have found ways around this issue.

UPDATE: Intel's Reply. Intel declined to officially answer our question about whether they have implemented a 10% Overclock Lock on the 915 and 925X chipsets. Intel provided the following official statement:
"Intel has done extensive evaluation of our 915G/P and 925X chipset products and have designed them to run robustly at 800FSB. Any use or operation of these chipsets beyond their specifications, including overclocking, has not been evaluated or validated by Intel. If a board provider decides to overclock their platform, they do so at their own risk, since this action will void the warranty for the Intel products. Additionally, from time to time, Intel may choose to implement functionality that helps to ensure that the Intel product experience is not degraded by those who may try to run our products out of spec."
Asus Breaks the Lock . . .


View All Comments

  • TrogdorJW - Wednesday, June 23, 2004 - link

    Going along with my last post, look at overclocking with the newer Athlon 64 chipsets. On socket 754 with second generation chipsets, what was the maximum overclock achieved? Anandtech managed to get 248 MHz, oddly enough.

    [H] had this to say when using the Abit KV8 Pro: "To put it frankly, this is the first time I have ever been excited about overclocking the AMD K8 core. This is the first time we have really seen tangible results that I think are going to be within the reach of the mainstream enthusiast that wants to experiement a bit with performance gains without having to reinvent the wheel in order to get there." Note that they had to run the HyperTransport bus at 3*245 for stability.

    What I'm trying to get at is that with the numerous changes to the platform, I wouldn't be at all surprised to see that initial overclocking is not able to match the mature 865PE/875P overclocking results. Perhaps we'll see some BIOS revisions that allow adjustment of the NB-to-SB bus speed, or maybe PCIe in general is causing problems with overclocking right now? As usual, this is revision 1.0 hardware we're looking at, and there are plenty of bugs left to work out.
  • TrogdorJW - Wednesday, June 23, 2004 - link

    I think the article gives a pretty good representation of the current status, but it might be a bit too anti-Intel right now. Remember, we're dealing with a BRAND NEW chipset that has all sorts of changes. New Northbridge and Southbridge, PCIe support, DDR2 support, and a new interconnect between the NB and SB, to name a few points. So, all you AMD fanboys, answer me this: how well did the first generation Athlon 64 chipsets overclock? And they didn't even add new RAM or PCIe support; all they did was moved the system bus to HyperTransport!

    I'm not saying Intel is perfect and AMD blows; just giving perspective. On the Asus board, AnandTech was able to reach a 248 MHz bus speed on 16x, 15x, and 14x multipliers. As the article pointed out, either the multiplier lock still exists but has been raised to 248 instead of 220, or else the bus simply doesn't run properly right now at speeds higher than 248. Until we get more information, it's a little premature to make a final judgement. We're looking at ONE Asus motherboard right now - they didn't even have several boards to try the CPU on! There's still a very good possibility that further tweaks to the BIOS will allow more than a 248 MHz bus, but Asus is working with version 1.0 hardware right now.
  • Pumpkinierre - Wednesday, June 23, 2004 - link

    I was wondering why those 865/S775 board like the ABIT AS8 were appearing. It could be because of Prescott's power consumption that could blow mobos but it just may be that the new mobos are real screamers (already tuned to higher bus speeds for DDR667). I mean, it seems strange that ASUS can only unlock the FSB to 248MHz. You either circumvent the lock or not. So ASUS might be playing politics as well, not raising the ire of Intel too much while being still ahead of the other mobo manufacturers as far as enthusiast appeal goes.
  • Creig - Wednesday, June 23, 2004 - link

    Intel - "Additionally, from time to time, Intel may choose to implement functionality that helps to ensure that the Intel product experience is not degraded by those who may try to run our products out of spec."

    "from time to time"? When HAVEN'T they attempted to suppress enthusiast overclocking?

    Their so-called "Intel product experience" is simply them telling you what system you want and how you want to run it. Irregardless of what you ACTUALLY want.

    What a bunch of weenies.
  • vedin - Wednesday, June 23, 2004 - link

    Wow..Anandtech got a responce from Intel in under 24 guys are fast. ::thumbs up:: Reply
  • Fr0zeN2 - Wednesday, June 23, 2004 - link

    #17 - I've seen thermal comparisons elsewhere on a [H] link, and yes, 64's unclocked gives off about 50 and prescotts give about 80-90. My Northwood gives off 130w though, and that's not even with a wild oc. Reply
  • sprockkets - Wednesday, June 23, 2004 - link

    Overclocking a 2.4ghz prescott to 3.6 is no accomplishment when that was a low frequency on the Northwood. Prescotts slower than 3.2ghz are in my opinion underclocked, since why would a new generation of processors have to be clocked lower than its predecessors, WAY LOWER? Reply
  • rjm55 - Wednesday, June 23, 2004 - link

    #18 -
    What a load of horse manure!! You would think Intel could just say "Yes, we added a 10% OC lock" instead of that convoluted explanation that leads to the same answer.
  • Wesley Fink - Wednesday, June 23, 2004 - link

    The article has been updated on Page 1 with Intel's Official Reply to our question of whether they have implemented a 10% overclock lock on the 915 and 925X chipsets.

    Intel declined to directly answer the question, but provided a statement which is quoted on Page 1.
  • dvinnen - Wednesday, June 23, 2004 - link

    Normally I don't, but they are the only ones I see with such a comparison. Plus, they are usally a Intel favoring site, so anything that shows AMD so far ahead, I figure most be alright. Hard to skew a temperature test unless you attach the dio in some obscure place. So, AMD could be more ahead. Reply

Log in

Don't have an account? Sign up now