Intel 925X: Exploring the Overclock Lock
by Wesley Fink on June 23, 2004 12:05 PM EST- Posted in
- CPUs
. . . or Do They?
However, further tests with the 3.6 Engineering Sample shed more light on the question.
| Asus P5AD2 (Intel 925X) - Highest Stable Overclock - 3.6 ES | |||||
| CPU | Multiplier | Voltage | Highest FSB | CPU Speed | % Overclock |
| 3.6 ES | 14 | Default | 992 (248) | 3472 | 24% |
| 3.6 ES | 15 | Default | 992 (248) | 3720 | 24% |
| 3.6 ES | 16 | Default | 992 (248) | 3968 | 24% |
| 3.6 ES | 17 | 1.425V | 952 (238) | 4046 | 19% |
| 3.6 ES | 18 | 1.425V | 896 (224) | 4032 | 12% |
While the 3.6 ES was also able to reach around 4GHz when overclocked, this represents only a 12% overclock for this chip. It is not unusual for top-of-the-line processors to achieve a more modest overclock percentage than lower clocked CPUs that probably are off the same line. The advantage here is that the ES chip allows the setting of lower multipliers to determine the maximum overclock potential of this motherboard.
We should be able to achieve the same approximate 4GHz at lower multipliers - but with higher FSB. At a 14 multiplier, 288x14 would give us about the same CPU speed as 18x224. However, we find that the highest FSB that we can achieve is the same 248 maximum we found on the 3.2GHz. In fact, whatever the multiplier, 248 was the highest FSB setting that we could achieve.
While we can't be certain whether 248 is the limit of the Asus P5AD2 motherboard or an artificial limit of the chipset, we suspect that we are dealing with chipset limitations. It appears that Asus has effectively bypassed the Intel Overclock Lock, but the design of the lock cannot be completely bypassed. Even a clever design like the Asus still seems to be overclocking limited to around 25% above rated speed.
Facebook
Twitter
RSS
36 Comments
View All Comments
Fr0zeN2 - Wednesday, June 23, 2004 - link
#2 - An all-out clock lock? It'll be funny to see how many people actually buy Intel after that. Actually, it'll be funny to see how many people still buy Intel after THIS.MAME - Wednesday, June 23, 2004 - link
Intel sucks like that but mobo manufacturers have no choice; Intel is #1 by a long shotoverclockingoodness - Wednesday, June 23, 2004 - link
dvinnen,The motherboard companies have to support Intel. The motherboard makers know that they are useless without Intel. Intel is the top chip maker in the world so they can't just leave Intel even though they hate their starategies so much. It's called business. As much as you hate one of the key players in the industry, you still have to work with them. These motherboard makers might lose close to 50 percent (just guessing) of their profit, which is not good for them.
When AMD did not introduce new processors for 1.5 years (except for new models of Athlon XP), motherboard makers still supported them with new motherboards. Of course, after a while mobo makers don't have much to add as they already have all the new features out so they develop something unique.
That's how the industry lives on. :-)
To answer the above two posts, Intel is definitely a loser this year in the desktop market. Their mobile market is secure, BTW. The prescott core is a failure, the new chipsets and technologies are a failure. And I am already scared about the BTX technology. It would be funny if BTX technology doesn't show high performance margins.
To conclude...Intel has failed in the year 2004.
dvinnen - Wednesday, June 23, 2004 - link
What I don't understand is why the mobo makers take so much crap from Intel. First intel keeps charging more, then R&D for BTX, R&D for this new socket, they are going to have more returns because of this new socket, now the new crap that got to spend R&D to figure out how to get around. I keep thinking one of the big boys over there will grow a pair and stand up to intel and stop making their boards. I would inmagin their margins on intel boards are more than a 5th of what their margins are on a SIS or VIA board.ViRGE - Wednesday, June 23, 2004 - link
A thought: Last time we got "clever" and bypassed an Intel clock-lock mechanism, it was the B21 trick, which resulted in Intel removing the 66/100 multiplier lock, in favor of the constant multiplier lock we deal with today. If we get "clever" again, who's to say Intel won't repeat history, and implement an all-out clock-lock?Zebo - Wednesday, June 23, 2004 - link
Slower, generates more heat, uses more power, enthusiast unfreindly, no memory controller, no 64 bit future....why are people catering to this company again?