If you wax it, washing becomes completely useless..a waxed surface repels water and detergent..

Yeah, due to the bug AMD implemented a workaround which degrades performance. A upcoming AMD Overdrive version will allow the workaround to be turned off for maximum performance, but when turned off the system may sometimes crash on full load.

No the errata will not cause problems. Actually its been fixed, but using a long workaround that sacrifices performance. With the next upcoming version of AMD Overdrive (or is it out already) it will allow you to turn off this workaround for full performance, but will result in occasional random hangs when at full load. The real reason why AMD did not want to release the 2.4Ghz proc is because its supposed to be a relatively high end proc, but because of the performance penalty of the workaround, it will not give the expected performance, plus if you turn off the workaround,

It affects all Agena/Barcelona processors revision B2 and earlier regardless of speed. It will be fixed in revision B3. Thing should get interesting after that, since the main impact of the workaround is that it slows down cache. Cache speed, and to a smaller extent size, is the main reason why the Stars arcitecture is losing out to C2D.

Any mobo can be voltmodded, its just a matter of understanding and manipulating the voltage regulation circuit. Google up the datasheet of the voltage regulator IC for a start. Btw, make sure you know what you're doing, most voltmods will void warranty, and a small mistake can fry a lot of stuff. Always check with a digital multimeter and take things one tiny step at a time.

Dunno why you're PWM so hot..even at 1.5v and full load mine is only roughly the processor's temp, which is around 60-65C.

Here's the rest of my OC log. For those who haven't read the first part, its here -> Link

I actually did something interesting first before I started OCing the F3..those of you who COD our F3s together would know quite well what..frankly, I had more fun doing this than actually OCing the F3..
http://forum.lowyat.net/index.php?showtopi...post&p=14653750

To be honest, there's not much to say this time, just a couple of screenies only.

My initial OC with the F3


And for my OC record, which I did not have any time to optimise settings (I was in a hurry),


At long last, add me into the list of AM2 overclockers
3800+ X2 Windsor F3 | lohwenli | Abit NF-M2 nview | Crucial 10th Anniversary | Thermalright XP-120

Superpi 32m screenshot in spoiler above

This post has been edited by lohwenli: Dec 14 2007, 08:35 AM

Yup, thats the main reason most people OC-get high end power for cheap. Only hardcore OCers go further than that, spending money and even sacrificing hardware if they have to to get nothing less than the best performance that money alone cannot buy.

Wait for a new version of clockgen, eventually your board will be supported. Another way would be to use nTune (download from nvidia's website) but its only limited to 250HTT for most boards.

Its ok, your temp is fine. Only if it goes beyond 65C then its worrying. Using valueram to OC is fine, it will still hold you back, but only at high OC. For example, you're using 667 valueram-if you set the ram speed manually to 400 while mantaining the timings used at 667, and perform a 50% overclock resulting in a processor speed of 3Ghz, then the ram will also be overclocked 50% from 400, resulting in 600, which mean you still haven't overclocked the ram. Keep in mind though that this is just in theory, while the numbers are correct, you may not be able to reach 3Ghz or will find it difficult, depending o luck and experience. You're using a Brisbane core, revision G1, which typically overclocks to 2.7-3.0Ghz with minimal voltage increases.

Just remembered that you're using a Brisbane, the temp sensor does not work properly, there is no sure way of knowing the exaxt temp of the proc. Best way remaining is to feel the base of the heatsink, if it feels warm but not hot then its still ok. For other processors where the temp sensor works properly, make sure temp of either core does not go above 65 (don't add up the temp). If it does reach that high for a short while (like a few hours), no problem, but running for weeks at that temp is risky. 70C is about the limit before the processor lifespan get shortened noticably, though it will not affect the proc if its just for a while, but it recommended that you stop immediately. Processors in general don't die that easily from overheating anymore because there are safety trip-offs, its usually voltage that kills processors nowadays. Always be careful when playing with voltage, take advantage of others' experience. Voltage tweaking is particularly tempting to use despite the risk because its the one of the most sure ways to allow for higher overclocking.

For a start keep the multiplier high while increasing the HTT, don't touch the voltage yet. Spend time tweaking around other settings, none of them are particularly dangerous. Anything wrong, just reset the BIOS. AM2 is particularly sensitive to memory settings, keep that in mind.

The temp sensor built into the processor is messed up..what to do..even AMD can only admit they screwed it up..

About HTT, for the time being, keep it to 1000 or below, its calculated as the base clock x HTT multiplier. HTT overclocks vary widely from system to system, so you can never be sure. Its always the last thing to be overclocked because it doesn't give much difference in most applications. Only recently have I heard that it can improve gaming performance, but ONLY if everything else is already overclocked.

The changing multiplier is because of AMD's Cool n Quiet function, where it drops the multiplier and voltage when the PC is not at full load to save power consumption and reduce heat. Turn it off when overclocking (look in BIOS), as the voltage it sets when it drops the multiplier during idle will no longer be sufficient to keep the system stable. If you still want the system to drop voltage & underclock to save power, you'll need to use CrystalCPUID, but how to use that I'll save for another time; its a bit long to explain. Or you can just go read though the guide on the website

Here's how..but must be rajin..go download the datasheets for the ICs near the voltage regulator modules and read though them..from there you need to find the line where the IC reads the voltage of the component in question (vCore, vDimm, vNB, ect) then start making small adjustments to the resistors on that line (and I mean small, like less than 10 ohm at a time-digital multimeter is compulsory). Then check and see if there's any difference to the voltage. The whole idea is to fool the regulator IC to think the voltage is lower than it really is, so it will push up the actual voltage to compensate.

Be warned-voltmodding is not for the faint hearted or those with shaky hands. Messing around with solder, pencil mods or even just measuring voltages on tiny board components with a multimeter takes a steady hand. I've already messed up a few times myself, I'll post my story on the POC thread when I can.

Tapirus, you've been warned..do hard-modding at your own risk..any hard modding project, you better consider first if you can accept accidentally frying something..

Overclocking won't kill your HDD directly, but possible for it to kill it indirectly. If your PSU can't handle the overclocking and blows up, it can also take out some of your hardware. Theoretically, its impossible to kill a hard disk via the data lines as long as the power is stable, but I had a similar experience to yours which left me confounded-I was using a reliable PSU (which rules out the PSU blowing up scenario) and I was trying to find my max overclock when I forgot to turn down the HT multiplier, after that windows got corrupted and then the hard disk would experience errors every now and then, enough to make windows crash consistently at boot after about a week or 2 or usage. I checked the hard disk and it turned out to be suffering from read errors, fortunately almost all of my data could still be copied out. Sent the HDD directly back to seagate for RMA and got back a replacement unit a few days later, by then I bought another HDD and sold the Seagate to verz84.

I'm completely at a loss to describe how that HDD failed, it was never allowed to overheat and until the overclocking incident it had passed all HDD tests with no problems. The only explaination that I found reasonable is that is happened by sheer coincidence, after all hard disks fail eventually, some earlier and some later.

Windsor F3 is easier to OC, but takes up far more power and puts out more heat than Brisbane G2. Both will reach 3.2-3.3Ghz at least on good aircooling without too much trouble, with extreme cooling 3.5-3.7 is possible. At the same clockspeed, Brisbane G2 loses out to Windsor F3 by roughly 0.1Ghz, meaning Brisbane must clock an extra 0.1Ghz to be even with a Windsor.

Wat kind of hard disks?

I haven't seen proper performance comparisions of the Brisbane G2 against 512x2 L2 Windsor, only against the 1024x2 windsor. However, when considering the benchmarks, benchmarks which are not really affected by cache also show brisbane at a 0.1GHz disadvantage vs winsdor. Those which are affected by cache, the results really

Not necessaarily, Windsor comes in 3 varieties, 256x2, 512x2 and 1024x2. The smallest cache is only for the lowest end processor, and the 1024x2 is only present on some of the highest end processors.

Not really. In gaming and multimedia, cache can make a 10% difference or more when comparing at the same clockspeed. This is why C2D is murdering AMD when it comes to benchmarks. In every other area, it doesn't look so bad.