FSB/Memory Overclocking

[Vehementi]

I wouldn't mess with your multiplier as of now, that seems to have consequences of it's own. Before you sort that out I'd just reccomend boosting your FSB. Isn't your default multiplier supposed to be 10.5? If so, set it back down to that. Just overclock your FSB - you'll get better gains.

Once you reach your overclocking limit while boosting only the FSB, this is the time to up the voltage. And damn, I just noticed that your default voltage is 1.65V. That is insane, my P4 won't boot with a Vcore over 1.6V, and it's default is 1.525V But, anyway, by upping it I mean raise it in small increments, 1.675V or even 1.7V for now, then overclock more. Not 1.8V right off the bat, even though your CPU and heatsink should (better) be able to handle it and give you better gains.

 

[Nic]

You can safely raise your Vcore to 1.85v, but be sure to check your CPU temps don't get too high. Maybe try 1.75v first. I would leave your memory timing the way they were and simply raise the CPU clock multiplier in your case. The Athlon XP does benefit from tight timings as much as from high fsb, unlike Pentiums, which prefer high fsb. Once you are happy with how much you can OC your CPU, then you can work on increasing fsb if you wish. I think you'll manage to get your CPU up to 2.4GHz no problem, but any higher will depend a lot on whether you were lucky and got a good CPU. I have an Athlon XP 2600+ Barton that needs 1.85v to get to 2.3GHz and won't go any higher regardless of any voltage increase. I also have another Athlon XP 2500+ Barton that will run stable at 2.5GHz with Vcore at 1.8v. It comes down to luck of the draw, but from what you've described 2.4GHz seems like a good bet. I personally would not raise Vcore higher than 1.85v as it would be very risky though I know of some users that have gone as high as 1.95v.

 

[Mictlantecuhtli]

If you read Tom's Hardware Guide's article about memory timings, you'll see what kind of effect these "3-4-4-8" and "2-2-2-5" things have.

I would say it matters when you've reached a point in overclocking where you have to slow down these memory timings in order to gain a MHz or two more - then the result will probably be slower than having less MHz and more aggressive timings.

 

[erickdj]

XtR-X, I have the exact same motherboard/cpu/ram combination you have. I saw the thread and started to follow your steps and some of the suggestions by Vehementi and some others. I have my cpu set to 200Mhz x 11.5 just like your last post here. My ram timings are 2.5 3 3 6 which is somewhat different to what you have and what Vehementi recommended. at 2300mhz the system is just as stable as it was with the 10.5 multiplier. My vcore voltage is set at 1.775 in the BIOS but CPU-Z said it was at 1.790 I'd like to know if you get it to run stable at anything higher than 2300Mhz.

 

[bedlam_4]

Um no, I'm sorry but you are completely wrong. Upping the voltage is proven by thousands of overclockers to yield better overclocking speed. Why do you think people need to get their CPUs to -80C for them to overclock fully? Do you think frequency produces heat? Do you think they need to get the processor cooler only to raise the MHz? Voltage produces the extra heat, and they need to make the processor cooler to up the voltage, which then allows for more MHz headroom.

Vehementi, with all respect for your knowledge due, the former statement is a very common misconception. I also thought that higher frequencies caused more heat when said frequency is measuring mechanical vibration. The following is a quote from Per Hanson's guide to over clocking ATI cards. After reading that it would be very logical to assume that a higher frequency = more heat output. After reading your post though it is easy to relate upping clock speed as akin to turning up an electric heater or spinning a rheostat. Then spinning the rheostat up a little on my brain (I needed to square what you stated with my own misconceptions) I came to the realization that the modules Per refers to below HAVE NO MOVING PARTS. I suppose there are many 'cookbook' over-clockers out there who really have no idea what is actually going on with their boards and systems.

" Our victim was an Atlantis Radeon 9800 Pro board that we reviewed sometime ago, Sapphire was kind enough of leaving the card for us to play around. Right out of the box with no additional cooling this card’s memory would overclock from 338 MHz (default) to 366 MHz.

At this point it is quite reasonable to assume that the memory is running a bit hot, so I bought some standard sized Alpha heatsinks and cut them to fit and put on the memory modules. After I putting them on with some Arctic Silver Thermal Adhesive and let it set for the night my maximum memory overclock was up from 366 to 372 MHz." from Per Hanson's ATI card mod guide.