You forgot to link BIOS 411. Stable BIOs so far but can't OC it yet since I don't have D9's with me. So far on Elpida's boot stable at 400Mhz

411 seems stable so far. Currently playing with Elpida's so trying to get the best bandwidth timings even at 333Mhz. Will push up slowly later. Older Elpida's aren't as high flyers as nowadays

Dang nice clocks. You're lucky using D9's. Can't seem to stabilise mine on Elpida's. Currently doing 8*333 instead stock volts but adjusted timings

EDIT: BTW, would appereciate if you guys posted results including which BIOS you're using. Thanks

This post has been edited by almostthere: Apr 20 2007, 01:33 PM

It's not actually strap recalculating. It's a safe method on ensuring that the BIOS does not get corrupt actually. Those who've frequently OC'ed would realise that too many times in a row changing the clock frequency would lead to some corruption in the BIOS. Sometimes it would mean a cold boot or worse (back when customised BIOS was still in it's infancy) a dead BIOS requiring a hot flash

EDIT: In fact for 965-S users, if you dare to risk a corrupt BIOS, you can opt for a soft reset option by selecting Mode 1 for Exit Setup Shutdown under Genie BIOS section

This post has been edited by almostthere: Apr 23 2007, 09:58 AM

VTT is if I'm not mistaken is your FSB voltage. If you've used 975XBX2, it corresponds accordingly. Based on what other have done, if at high clocks unstable, you'd need to boost it up a bit.

Currently doing 8*400 at 1.47v. Seems it's true what they say about the vdroop. A consistent 0.07v drop at load. I can boot stably at 1.4v but when ORTHOS loads up, reboots. Once corrected the vdroop, 1 hour stable so far. But for the time being playing with RAM timings. Tip for you guys, if your rig can't get past POST but everything else runs normally (i.e. you reset your PC and doesn't hang) adjust the tRFC. I'm setting mine at 22 which is the lowest I can go.

depends, it's a very subjective matter dependant on RAM chip and other variables. But yeah, should be about half of VDIMM

You should refer to Raja's thread in dficlub to understand all the options. But if I were you, I'd leave that startup clock alone TBH. And are you sure 0.02v? I'm detecting 0.07v using multimeter

http://csd.dficlub.org/forum/showthread.php?p=18796

aiyak...forgot to bookmark it and some more my router tengah buat hal. Gives problems whenever it heats up. But if you checked the XS sites, quite a few have tested it and reported those vdroops

This post has been edited by almostthere: Apr 24 2007, 01:16 AM

how you detected the vdroop? Using Smartguardian isit? Mine C1e and PPM enabled the voltage naik to BIOS settings got la. But the problem is that the real vdroop is not shown due to poor calibration and resolution.

EDIT: However I'm reading that the new beta 418 BIOS causes a lower vdroop. About 0.01v if upto 1.5v VCORE

I'll try my best to find that post of the vdroop measure. But do bear in mind that Intel's spec allows vdroop of between 0.01 to 0.05v

This post has been edited by almostthere: Apr 24 2007, 02:22 PM

anything reputable which has got a native 24 pin + 8 pin connector on it. Mine is a HEC dual Xeon rated PSU. Stable like anything

NOTE: Guys wanna share you all one finding. Seems the MBM5 mod by FCG of XS isn't working properly for the Dark. Repeated this 3 times to confirm. Every time if I load the MBM5 with the C2D plugin, CPU usage shoots up whereby System process (sub process DPC and H/w IRQ) takes upto 70-80% usage. Suspecting because the IRQ access is causing massive delays and increased queued process

This post has been edited by almostthere: Apr 24 2007, 09:59 PM

If I were you, don't even think about using converters. There was a reason why I emphasized using native 24pin + 8 pin PSU's. It's by no mistake my PC is so far the least problematic among the Dark users here. This is because this mobo is very sensitive to voltage stability like all past DFi's great clockers, so therefore you shouldn't even begin to cut cost there. If you're planning to use a mid range GPU, then just get a good server based PSU which has the native connection I mentioned earlier, stable 12v amperage of about 30A's and above. Brands like for example Delta and HEC. I'm using this for example

http://www.cardigan-computer.co.uk/product...atid=40&pid=428

And rails is like voltage lines. Good PSU's have 3 individual rails, particularly +3.3v, +5v and +12v. After that there are PSU's which even have sub rails meaning the +12v would be split down even further meaning there would be +12v1, +12v2, and so on

P.S.: Even if you don't use a G80 GC, plugging in that floppy power connector is a good idea. Made my rig more stable. Last time needed 1.48v like that for my current clocks but now does it at 1.42v with C1e enabled

Coming from a 3 post poster, doesn't this sound like spam?

This post has been edited by almostthere: Apr 25 2007, 08:39 AM

FSP alias Fortron's are good man. That was the choice of overclockers during the crazy nForce 2/ BH5 OC days because of it's stability. IF your's is budget then mine leh?

I've elaborated in post #91 about it and in fact if you really want to have a fast reboot, there's an option there which is a gamble if you ask me.

the mobo would be able to find the basic boot stable boot settings which is why so many have not faced problems starting theirs up for the first time. Not to mention that the mobo has a very good compromise settinsg wise in terms of latencies vs. bandwidth. But might I suggest that you adjust your advanced timings to take advantage of your RAM. I gained a reasonable increase by adjusting timings TBH

 mbench4.txt ( 438bytes ) Number of downloads: 10

Majority auto settings

 mbench.txt ( 438bytes ) Number of downloads: 6

All manually adjusted

NOTE: This is a simple test using the mbench program

This post has been edited by almostthere: Apr 25 2007, 01:51 PM

Not yet found marny to buy Delta. Want their 600W monsters. Only got a HEC server class now. 460W nia

Nah, HEC's are damned reliable since last time. And the ripple isn't as bad as some branded PSU's which I shall not name anyway. And since running a 7900GS, more then enough for me

Should be as long the amperage are good. Single rail by marketing terms would mean that there would be only one +12v, instead of +12v1, +12v2 and bla bla bla. Good PSU's would have at least the 3.3/5/12v all running on individual rails. It's recommended that at all times the floppy is plugged in for increased stability. My PC now can run at lower voltages due to the increased voltage stability.

try loosening up your advanced timings, usually that's the limiting factor there. D9's comfortable working range should be about the following

tRFC - 21-30

All Precharge to ACT = use 4

Write to PRE TWR = 10 (upto 840) 11 for upto 1000

TWTR - 9/10

This post has been edited by almostthere: Apr 25 2007, 08:16 PM

auto might cause 2 things. Either locking up or increased lag. Noticed that the 1st time I touched 400Mhz on auto. Was slower booting then at stock clocks....

according to Shawty, it doesn't. Maybe I'll ask around for you

Yep confirmed B0 no jalan, B1's will boot up on the 965-S Dark. Sorry Shawty, you're out of luck

If I recall correctly, B0 had some functions disabled or working incorrectly such as C1e and PPM which meant that mobo's which boot up based on detection of these functions would not be able to detect correctly everything

To be sure you need to read the sensor's using TAT which would measure based on core not Tjunction which at times might be miscalculated. Besides that, have you made sure your WB is making 100% contact with proc?

chill la bro..some people are still noobs leh.

and syapl, the mobo you can use 4 pin ATX but for better stability, it's recommended to plug in an 8 pin instead

anyway this is the final word about B0's from Raju of XS




This post has been edited by almostthere: Apr 26 2007, 10:25 PM