PCA Forums

according to x-bit labs -


Overclockability of Athlon XP 2800+ Processors. They Cannot Be Overclocked
Posted 10/12/02 at 5:36 pm by Anton

We have already discussed the situation around availability of the newer Athlon XP processors earlier today (see this news-story), and I expressed some doubts about AMD’s ability to start the volume shipments of the 2800+ CPUs on time. By now I have received indirect confirmations to my concerns. Apparently, Athlon XP 2800+’s 2250Mhz frequency is very close to the highest core-clock the core can function at.

Unfortunately for AMD, I now have to admit that with Athlon XP 2800+, AMD has reached the point of the maximal frequency the Thoroughbred can achieve. Our friends from OverClockers.ru web-site managed to grab a sample of the mentioned processor and tried to overclock it. To their great disappointment, they only succeeded in launching the CPU at 172MHz FSB, while the nominal frequency is 166MHz. The resulting core-speed was 2327MHz. They did not test the processor on stability at this clock as there is no sense to overclock so fast CPU by 77MHz. The whole issue from our Russian friends can be founded here.

We can sum up that 2250MHz is the highest speed AMD K7-family processors can achieve. In early 2003 AMD will launch its Barton CPU that will contain 512KB of L2 cache, hence, will be clocked much lower due to increased number of transistors. Of course, AMD can still add another metal layer to the integrated circuits design in order to be able to increase the frequency, though, I seriously doubt it will help the company much. On the other hand, we now see that AMD will hardly be able to provide a lot of its 2800+ processors this year without adjusting their manufacturing technology. Since I cannot remember AMD to tune their processes so that the CPUs were more easy to overclock, the Athlon XP 2800+ processor may become a “sacred” device for AMD closest partners and will never reach the retail.

[printed in full]

link

Them sonsabitches sure do assume a lot. They test one sample of a just released CPU and claim it's the "limit." It isn't worth the disk space used to store it.

Yeah and what they fail to tell you is that they get a engineering sample... I'm sure by the time we actually see teh 2800+ at the beginning of 2003 the thing will be just fine... The bartons aren't gonna have SOI so they'll pretty much be a tbred with 512k L2 cache and 166 FSB... some dumbasses need to think before they type...

Nah, I think the core maxes out at around 2.5, because thats what a guy with a 2400+ and a vapochill system was able to reach, but nothing higher.

the plural of "anecdote" is not "data".

yeah and also when the 1500's and 1600's first came out you were doing good to overclock one past 1.7 gig... I've got a 1600+ right now that will overclock to 1.9 gig on air cooling. Only because of core revisions. Too early to specualate one how high they'll take the t-bred.