The problem is obvious, reduce the HT multiplier. Also, its a Windsor F2, which means beyond 2.4GHz you'll need to tweak around a bit.

Brisbane core has a P/N ending with DD. CZ is the Windsor core, stepping F3, also another lovely overclocker. CU is the normal Windsor, which although can overclock, takes a lot of effort.

And yes, you can overclock to at least 2.8GHz even with the stock cooler if you're using a DD or CZ; keep an eye on temperature though. 3.0GHz is not difficult if you have a better cooler.

The cache latency is higher, so brisbane needs to be overclocked 100-200Mhz more than Windsor for the same perfomance. The different power ratings actually all use the same chip, but AMD just undervolts them-you can do it yourself too..really cool man, after that you can even turn off your CPU fan and it won't overheat (casing must be well ventilated though).


Windsor comes in F2 and F3 steppings, Brisbane comes only in a G1 stepping. Windsor F3 and Brisbane can clock at least 2.8GHz without much voltage increase.


Yeah, Windsor F2 needs a lot of volts to clock high. But Windsor F3 (P/N ends with CZ) can clock all the way to 2.8GHz with minimal voltage just like Brisbane.

Almost the same, but the fact that a brisbane's performance is slightly slower makes up for the difference. So at max overclock (~3.0GHz for Windsor F3, 3.2GHz for Brisbane), both chips perform roughly the same.

Hehe, but the Abit NF-M2 nview special...can put until 2.5Vdimm..Vcore can set until 2.0V..not careful maybe can have smoke come out..

Yup, silent OTES is crappy if you use it straight from the box, but with some modding it works damn well..1st dump the stupid "bubblegum" themal pad and use some proper thermal paste under the heatsinks..then make sure your chassis is well ventilated (a silent 120mm fan will do)..then the silent OTES ROCKS!!

I've got to admit, though the percentage increase is fantastic, the actual clockspeed isn't. But I think its because the motherboard is limiting the overclock, must be running almost 390HTT by then.

---

Added on May 26, 2007, 8:55 amThey're rams with Micron chips, labeled D9***, where the most popular for overclocking are the D9GMH and D9GKX. It doesn't matter what ram is it, whether its cheap or expensive, if it has those chips it will overclock to at least DDR1000. Particularly good batches can overclock to DDR1200.

This post has been edited by lohwenli: May 26 2007, 08:55 AM

CnQ can be unstable when overclocking because it also reduces processor voltage in addition to reducing the clock frequency. It does it in steps, no sudden jumps from 50% to full clock speed. After overclocking, sometimes some of the default voltage settings in the intermediate steps will not function well anymore-too low to be stable at overclocked settings.

However, sometimes you may just be lucky. In my case, not so. Games which do not fully load the processor but still do more than 50% load cause CnQ to switch the processor from full speed to partial (but still more than 50%) many times a second due to the game's constantly varying load. Eventually, the system crashed. Currently I use CystalCPUID to create a form of manual CnQ, which undervolts and underclocks the processor at idle, and restores it to normal speed (with slight undervolting) at moderate load, and overclocked (with overvolting) at full load.

This post has been edited by lohwenli: May 28 2007, 11:21 AM

Virtualisation is to allow multiple OS to run at the same time. Sort of like splitting up the processor to run Windows, Linux and Mac OS all simultaneously. Usually used in data centres to allow 1 server to handle multiple tasks at once. Of little use to end-users like us, as in-depth knowledge of several OS is required.

Got..AMD's one is called Pacifica..available in all AMD CPUs since mid 2006 (basically most of the AM2 procs).

Oh yes, another beneift of virtualization is that when one of the OS running on the system crashes, it the remaining OS running will not be affected. If I'm not mistaken, on top of that the you can run several instances of the same OS, don't need to install several different OS to use virtualization.

I've used a 130W with an XP-120 before, couldn't get it to run at 130W, only around 70W because can't find a 15V supply for it.

No kidding. The real problem is A64's put out damn low heat on idle, even fans are not required on idle when using stock heatsink. Things get real interesting when you add a peltier-think sub zero as a possiblity (for the 320W peltier sold in CPU, motheboards section, I think sub zero for full load may even be possible, but watercooling required for the peltier's hot side).

But of course, that comes with a catch..remember what you see on a glass with an ice drink? I had to pull the plug before any damage was done.

I didn't insulate because I though it would only be around 20-30C full load. It was when I stopped prime when temps took a major nosedive Then the IHS started 'sweating'...

My usual full load temp is around 53-55c, with the pelt it was still about the same (but idle was insanely low-15c and still dropping fast when I pulled the plug). Seems the pelt was being overloaded, more voltage needed to get its full cooling capability. But anyway, I'm going to test out with watercooling, hoping to see a difference-if none means the pelt is definitely overloaded.

Btw bro, your pelt also needs 15.2V. If you plug it into the 12v you'll probably get 50-60W only, not even enough to cool it at stock speed. At full 15.2v it might work out ok. A ultra high watt pelt is more useful in the sense that you can use voltage/PWM controllers to tailor it to your needs and running it below the rated voltage will still give sufficient cooling. My advice though, is to use speedfan to control the pelts via a power MOSFET relay.

I'm thinking of nabbing one of those 320w pelts from Garage Sales someday, but currently still busy with my first watercooling set. A bit of headache with it, because a fair bit of modding needed.

That's a 89W Windsor F2. Pretty hard to clock faster than 2.5GHz without a fair bit of voltage and good cooling. Price is nice for the spec though, but not worthwhile for overclockers-no thrill also cos it already starts at 2.2GHz, which is damned easy for any overclocker to achieve on that core.

Its up to you, but frankly, those chips are poor choice for overclocking due to cost effectiveness as cheaper, lower spec chips will also achieve the same overclocks. Sell them to non-overclockers, they will appreciate the lower price of RM300 or below (must be retail boxed units though). Btw, this is going off topic..discuss it elsewhere..

Either your motherboard chipset or the Hypertransport was overclocked too high, or the SATA clock for that particular socket is not locked-the problem does not appear immediately but takes a while before it can be noticed (more on this below). I had the same problem also-I was lucky the drive could still be read, though with frequent lockups. But it was definitely damaged-testing it on another system revealed numerous read problems even though the drive had never overheated and and no bad sectors.

In the case of an unlocked SATA clock, there is no choice but to try another SATA port. In early K8 days, motherboards were notorious for having unlocked SATA ports, which even a few DFI Lanparty's were plagued with. Sometimes some of the ports are locked, and some are not. If all are not locked, then too bad-the board is going to hold back overclocking or be unstable no matter what.

Here's a problem most overclockers here don't know-none, yes none, of the stability tests we use actually test things other than the CPU, ram, memory controller and graphic core and memory. On AMD's K8 processors (754/939/AM2) its particularly difficult to know if the motherboard chipset is unstable during overclocking until it is severe enough to cause a system lockup because the memory controller (which is usually the first component on an intel CPU-based chipset to give up) for K8 isn't on the motherboard chipset. If the chipset is slightly unstable, say the SATA controller isn't working 100% stable, it won't be obvious until it bungles up a crucial read/write operation(eg windows kernel, or worse, file table/partition table), and the usual non-fatal read/write errors will go unnoticed because of the uncountable bugs found in windows.

And this is only the beginning-the motherboard chipset also controls USB, PCI/PCI-E, LAN, Sound; pretty much everything else other than what the processor, graphics, memory does. Even if the PCI/PCI-E clock is locked, the chipset still is effected because it must relay the data over. So you can't say the chipset is 100% stable until you can be sure that all the motherboard functions work 100%, which to my knowledge is a pain to check everything.

About your data, I'm sorry to say this-if the partitions have disappeared it will probably not be possible to recover it without substantial effort (especially if you have done any write operations on the disk since the disaster). I personally have had to deal with 120+160+120+160+80GB worth of data recovery, and even with professional help I can tell you it wasn't easy nor was it a complete recovery.

To all overclockers, if this is discouraging-keep this in mind. If you keep backups and are willing to put up with occational system lock ups, overclocking is still rewarding-we are getting high end performance out of cheap stuff, in some cases even getting performance money alone can't buy . Not to mention all the fun we've had tweaking everything to perfection

I understand your problem about the sata ports, as I'm also faced with the same dilemma. Seriously I also have no idea how to find out if the chipset can't take it or the SATA clocked is unlocked until something goes wrong. And yes, the hard disk can be corrupted because of this, I have a Seagate 160GB here suffering the same fate. Mine even needs to be sent for RMA-having read errors even though no bad sectors. Before the incident it was 100% ok.

Its not likely to find any option to lock the SATA ports-I know very few motherboards had it, it was usually listed together with the PCI/AGP/PCI-E clock (which you should ALWAYS lock when overclocking). Currently, SATA usually shares the PCI-E clock, but it depends on board design.

About data recovery, there are a few programs you can try, but since you said the partition went missing, its probably not going to be simple. Most of the easily obtainable data recovery programs require that the partitions are still intact, and preferably not overwritten. There are more complex programs that can, but very hard to find, usually used only by data recovery services and police forensics. I have a contact of a data recovery company, but I'm not sure about the cost.