The Official OC (OverClocking) Thread!

Ok. Ive since tried bumping voltage to 2.05! It did effect its AUTO control. But it did not increase to 1066. Rather than post the image again, just know that its automatic control, matched the SPD side of the image. Am I right in assuming that SPD, is simply the Rams way of communicating with the Bios, as to its max specifications?

Ive also tried changing, "Minimum RAS Active Time(MRAT)" to 15T. That also had no effect on it. Im currently running under manual Mem/Ram settings. Both Command Timing(2T), and MRAT(15T) are currently set. And im still running at 800Mhz. Surely it doesnt need yet another voltage bump. And I also noticed, im not allowed to change the EPP settings. This could pose a problem. Im currently running windows 7, and bios flashing makes me nervous just in XP!!! And I know there has been at least 1 new bios release since I last flashed!

 

Well...im officially at a loss, until I can check each Module individually(unfortunately, Memtest bootable wont test 4 simultaneously)! Because as of now, I either have one or more faulty modules, the memory controller cant run 4 modules simultaneously at 1066, im not supplying enough voltage(Unlikely), the board has an issue, LOL!!! I suppose the list goes on. In all likely hood, given what ive read online, controlling 4 modules simultaneously is quite the challenge. It is no doubt the trade off for capacity, to run at a slightly slower speed. And who knows how high I can get it with the FSB! Im now back up to its typical OC! Ive had this setting with lesser ram, AND a lesser CPU, soo...this could get interesting, FOR ME anyway.

Is it also possible, since I am running a Black Edition processor, the Memory controller is configured in such a way, that will limit Ram Oc'ing??? I may be reaching here now

 

rubixcube,
As far as I know, no modern (socket AM2 and up) AMD motherboard will let you run at 1T using DDR2. If you set it to 1T, it will set itself to back 2T. As far as I know, high speed DDR2 won't run at 1T. It's a moot point anyway, as the performance increase is so slight, it's essentially not worth it. Here's an interesting analogy I found at Overclockers, that gives you an idea of what's going on.

"consider the cpu a Car factory and the mem bandwidth the steel supplier

imagine now the steel supplier is very very fast (1T) and brings tons of steel to the car factory.

but the factory can't produce so many cars per day, so most of the supplier capacity is wasted".

What I get out of that, is the CPU's ability to process the additional information, isn't enough to keep up with the supply, so that almost all the additional speed gained from using 1T is wasted because the CPU can't process it fast enough.

Synthetic benchmarks may show a big improvement, but they are there in theory only, based on the memory settings selected in the bios. In real world use, there's minimal improvement at all, and stability issues become far more likely. The CPC's timing is RAM to MB relative and not a timing effect of the RAM itself.

Best Regards,
Russ

 

Well...im even more confused now. Apparently Memtest86+ is designed for configurations such as mine(4 modules). However, it did not detect multiple things properly. It thought I was running an Athlon for instance, for the test it ran it at 1166! Yes 1166, not a typo! I think I like the layout of my older version more LOL! I let it run the test for 5 minutes with out error. Im thinking its a controller limitation. I guess I can remove 2 modules, and see how it reacts with only 2 modules. Unfortunately, its yard work time LOL!

 

Oman7,
It thinks you are running an Athlon because Memtest 86 has never been updated to include the Phenom or Phenom II! Essentially, an Athlon is all the software recognizes!

You might be able to run 1T with 2x512 or 2x1GB, but the improvement I saw with SuperPi was 3ms over the 2T score. In the socket 939 days with DDR, the effects of using 1T were measurable, with DDR2, it's almost non existent and causes stability issues depending on how many memory slots are populated.

Russ

 

It shows support for K10/phenom and i7 processors.
http://www.memtest.org/
But perhaps they still have bugs to work out.

 

Ok, now it seems to be stable at 227Fsb 15.5 Multi for an effective 908Mhz ram speed(in windows 7 no less). Its currently been running Prime95 for 6 minutes, without fault! Cpu Vcore at 1.4V! I shall attempt to raise the FSB until unstable. Though from what ive read, it may be necessary to drop the HT respectively. Though its currently stable @200+mhz OC'd

 

Ok. Now during the last 3 OC's, one core will fail to continue running. 2 tests, a single core almost completely quit within 15 seconds. The last test(230FSB), survived for 2 minutes, then a single core quit running. And then I noticed that the Voltage was low in the 12V range! Could this be my problem??? I saw it dip down to 11.88V!

 

Ideally volts should be the other side of 12V, but 11.88 is alright. Below 11.75 is where it gets dodgy.

 

Oman7,

People here can say what they want, but the only thing that changed that non-problem, was going to a different PSU.

I thought I had a problem at 11.88v as well, but when I asked on the thread, Sophocles said "what problem"? Still, the last three PSUs I bought for myself all maintain 12.3v to 12.6v no matter what the load! That also includes the one that came with my case, which was running the same MB as you are, as well as the UD4H. Considering the difference in CPUs and video cards between yours and mine, to my eyes, that makes the PSU you have a bit suspect! You might well need a bit more power than your PSU is providing you with!

P.S. I also read earlier today that the minimum wattage PSU for your Video Card is 700w, and yours is overclocked 40%!

Russ

 

Yah...im really wondering about it. This last test averaged 11.92, and stable for 7 minutes(5 tests). WONDERFUL(Sarcasm)!!! And actually, I thought 500-550 was minimum for my GPU. Ehh...I should have got a behemoth! LOL!

 

I returned it to its typical 206Fsb 17X multi. It affected the 12V value by a good .05 - .07 Volt average! It now averages closer to 11.95-11.99 and quite stable. Though I get the feeling, that my PSU is more stressed than I originally thought!!! Well...I guess I have my spare PSU! Now im in the market SOOOON for a supremo PSU.

 

As I say to many people, OCZ units aren't quite as powerful as they say they are. That system could run off a good 550W PSU, but a 700W OCZ? Not so sure.

 

depends on how old the unit is aswell. PSUs do loose power.

 

I WILL NOT skimp on my PSU in the future LOL! They are as many say, one of(IF NOT) the most critical components in ANY system!

Its been running since December of last year. This is the second rig its been in. I suspect it will be a VERY good PSU for my secondary Rig, which will more than likely be my offline rig!

 

Here's a little thought to consider. Which chip does your GTX have? The 65nm chip or the 55nm chip? I just happened to come across this comparison.



As you can see by the graph, if it's not the 55nm chip, the power demands can get pretty high, and these are for stock settings, no overclock! Given your CPU and everything else you have in your computer, I would say that if your 260GTX is the 65nm chip, you could be treading on the limits of your PSU!

Russ

 

Well...Newegg doesnt list its Gpu technology(nm)! LOL, not that I would trust them! They've been less than trustworthy in the past. However, Everest reports 65nm. Though that could be due to everest requiring an update. Im still searching.

Well...according to Gpu-z, my 260GTX is indeed 55nm technology. Can this report be trusted? Version 0.3.3 by the way

 

Oman7,
I would tend to believe Everest before I would believe GPUz simply because they seem to be very up to date on things like that! I had Everest 4.50 before the P35 even came out, yet it told me that the chipset was a "Bearlake", which was correct. It also identified my GPU correctly, even though it wasn't a top card, when I bought it! You could try using google, using the serial number to find out which chip it is. That might yield something!

If you give me the exact model number from the box, I can tell you from that!

Best Regards,
Russ

 

this is the EXACT model number:
BFGEGTX260MC896OCBE
All I can find for BFG is a 55nm GPU. So XFX and others may be the only manufacturers to deploy a 65nm?

 

Oman7,
BFG has at least 3 that I know of. I was on their product page just before I posted the question. I'll check it out! BRB!!

If it looks and reads just like this one, it's 55nm. If it doesn't say Maxcore 55 on it, it's a 65nm chip!



Russ