REPORT WU-SPEEDS HERE !

Bill,

You may want to go into the BIOS and adjust some settings - overclock it. By my calculation, Gary's machine is running at about 2.08GHz to finish WUs at 21700 sec. If I remember correctly, the 2500 default speed was 1.867GHz.
As the Barton core grew mature ( became available at 3200 speeds), AMD "speed-binned" them for their lower speed models, meaning that the CPUs that would not test reliably @3200 speed were stamped with lower speeds, somewhat randomly, and sold as such, to keep AMD supplying a complete product line. When they made those large wafers of cores, they didn't make one wafer for 3200 speeds and another for 2800, another for 2600, etc. The Barton core matured about 2 years ago, so if your 2500 is newer than that, it will accept an overclock, most reliably through upping the FrontSide Bus, or FSB setting. You can OC it mildly (5-7%) with few or no other alterations. Should you wish to go further, you may have to adjust the core voltage upward very slightly, and install a nore efficient heatsink.
If you're interested in going this route, post back and I will provide a link or two to overclocking guides, or you can search for them on your own.
If your budget permits, and your motherboard chipset is compatible with 200 FSB speeds, for DDR400 memory, I would suggest this processor http://www.newegg.com/Product/Product.asp?Item=N82E16819103436
It puts out very little heat, so is extremely good for OCing, and costs less than $100US.

Your WU times for the 2500 are right about where they should be for stock speeds, if you were doing some other work simultaneously that "distracted" your processor. I calculate that you might see optimum times around 24300 sec/WU at your default settings.

Thank you for your contributions to E@H.

microcraft

Bill,

I must apologise for my previous tongue-in-cheek comments about your machine loafing. In my defence, I did insert a couple of smilies but I did omit to mention that my XP2500+ was running overclocked!! Michael is too smart not to work that out straight away. He gave a very good description of what happens when the chip manufacturing process matures. Suddenly most of the production is capable of 3000+ or 3200+ speeds but most of the market is at the cheaper end - eg 2500+. So you can imagine what the manufacturer must do. If it would make you happier we could simply erase the 2500 marking on the chip and replace it with 3200 and the chip can perform at that level quite happily.

To prevent this happening too easily, AMD locks the multiplier. A standard Barton 2500+ runs at (multiplier x FSB) 11 x 166.66 = 1833 mHz. To produce a higher rated chip, AMD would set the multiplier higher eg 12 x 166.66 = 2.0 gHz and they would call it a XP2700+ (I think) or something like that. When they want to produce a 3200+ it's usually done by running at 200FSB with the multiplier back at 11. So if you are lucky enough to have a 2500+ which has just been marked at that because AMD needed the most popular speed, why not run it at 3200+ speed and see if it works. Many people have reported success with this and it really represents minimal danger to anything.

That's exactly what I have done with a couple of my 2500+s. The secret is to put them in a good mobo where you can set the FSB in the BIOS to 200 and see if it will run. A lot do. Some may need a small tweak in Vcore to make them absolutely stable but it works a treat with reasonably low temperatures just on stock cooling.

There are a lot of Urban Myths around about needing expensive third party coolers if you want to overclock. Like all things, commonsense should be your guide. It always amazes me that people will bolt on an expensive cooler and think that is all they need to do. In fact the most important thing is not the expensive cooler but rather the ambient temperature inside your case. Blowing already hot air with a screaming high velocity fan throught the fancy fins of an expensive custom heatsink is simply not going to work. The secret is making sure you have cool air inside your case - as cool as possible. Then you will get a big effect by blowing it a bit faster through the fins of a standard elcheapo aluminium heatsink. Very simply, I just remove the standard AMD 60mm fan and replace it with a 60mm x 90mm adapter and bolt on a 90mm fan. That fan at low speeds (nice and quiet) will shift more air anyway. By the time this is all bolted on top of the standard heat sink, the inlet to the 90mm fan is quite close to the side of the case. So I cut a nice circular hole in the side panel and put a filter and wire mesh finger screen to cover the hole. The cpu fan is then drawing the coolest possible air from outside the case and the cpu temperature stays nice and cool even at full loads while crunching 24/7. The other benefit is that because the air temperature inside the case is kept cooler, the PSU inlet air is also cooler and hence the whole box runs cooler.

I have been running a large number of cpus ranging from Duron 1600s to Barton 2500+s and all are running between 2.1gHz and 2.25gHz on stock heatsinks with not a single problem for 2 years now. Of course I test every cpu for absolutely stable operation and acceptable temperatures. Today it's been 31C outside the building but a 2500+ crunching away at 2.2gHz has a cpu temperature of 47C and a case temperature of 30C. This box contributes to both EAH and LHC. With LHC, the validation is very strict and a result will be invalidated if it is not absolutely correct. I have no validation problems with either LHC or EAH.

I'm not trying to get you to overclock your machine. You must do exactly what you are comfortable with. All I'm trying to do is to suggest that if someone must overclock, they should think carefully about the best way to keep the cpu cool and that doesn't necessarily mean spending a lot of money. People just need to do their research carefully and if you aren't fully sure of what you are doing then don't overclock.

Ran just fine with the 200 fsb, ran 20,901.42 sec for the last wu, thanks for the tip!!! BTW me 3700+ does wu's in hight 18,000's to mid 19,000's, not that much faster!!!

Bill

Your multiplier is locked (fortunately!!!) by AMD and is still 11 even though you tried to set it to 12. That setting is simply ignored by the chip. What you did was a bit crazy!! Think about it ... You were trying to run the chip at 2.4 gHz (12 x 200) and that's way too big an increase in just one step. Put it back to 11 and be happy!! Looks like you have a good one but the fact it boots and seems to run OK is no guarantee that it is perfectly stable. You don't mention changing the cpu voltage (Vcore) so I assume it is still stock (1.65 volts). If it is then things are great and you have a good chip that is capable of XP3200+ speeds as they are really designed to do anyway.

There are a whole lot more things you should check before walking away and forgetting about it. You need a temperature monitoring program running in Windows to check the cpu temp under full load. It is going to be running hotter at 2.2gHz than before. If you havn't had to adjust the Vcore then temperature shouldn't be a problem. There are good free programs out there to do the job. Google is your friend!! :). I would also seriously recommend doing whatever it takes to keep the internal case temperature as cool as possible. You don't have to go as far as I mentioned in my previous post but reread it anyway and understand the advice. You also need to do the "Basic Overclocking 101" course :). By that I mean - use Google to research websites that have basic overclocking guides that spell it all out. There's a lot to learn and understand if you want to know you aren't doing something stupid. Overclocking is risky if you don't do the basic research and devekop an understanding. Overclocking is safe if done with understanding and care and caution!!!

You also need a good stability checking program that stresses your system and reports if it breaks. Prime95 is regarded by many as a good choice and if a chip is reported as "Prime Stable" then the overclock is good.

There are a whole lot more things you should check on, things that are way beyond the scope of this thread. Very good guides and "howto"s litter the internet and are easily found with Google. It's a fascinating area for study.

You have a good overclocking board and you have the potential to get more out of that chip through FSB overclocking but take it easy and check temperatures and check stability before you start submitting invalid results :). I have noticed that LHC is one of the best for giving invalid results as soon as there is the slightest hint of instability in the cpu. Their validation procedures must be pretty strict (ie an exact match is required, I think).

Good luck!! and do the research!! :).

LHC@Home, an exact match is required ... is a true statement. THis is why there is no OS-X version ... also, they have had some off and on problems with some chips against other chips ... VERY little tolerance for errors ...