So much for getting to bed early tonight.

[General Septem]

I'd been running Angelina (that is, my 165, not the whole system) at 312*9 for so long, but whatever I did I could never get my RAM to run on anything other than the 1/02 100mhz divider. I tried everything from setting the timings looser than a knocked-out tooth to bumping volts up to levels that could kill a yeti, and whatever I did didn't make any difference as the board just wouldn't boot at 312 with anything other than a divider of 1/02, even though the next lowest divider would still not put my RAM over DDR400.

 

Now despite Angelina running at 2808 MHz, having the RAM at DDR312 was really bottlenecking performance. For instance, EVEREST memory read benchmark gave 2491 MB/s at my 2808 settings, whereas even stock outperformed this at 3131 MB/s. I'd been meaning to do something about this for a while, and decided to try something tonight.

 

My first idea was to lower the CPU multiplier to 8. This did allow me to boot higher than 312 FSB, but topped off at around 330 which really didn't improve anything significantly. So I decided to try something different.

 

(Here is where you can begin to read if you just want the abridged version.)

 

My second, and arguably better idea was to start over at stock settings and to attempt to raise the FSB as high as I could without having to put my RAM on a divider. It was working, too. I knew Angelina would never give me 2800 with this method, but I also knew I'd end up with better performance, and subsequent testing proved this to be the case.

 

You already know EVEREST memory read did better at stock speeds than at my 2808 settings. Well, at 223*9=2007, EVEREST memory read performed a whole 1000 MB/s better. 1M Super PI showed a tie, but a tie between 2007 and 2808 was enough to tell me that I was on the right track.

 

 

And then I did something rednecked stupid. I jumped the FSB from 223 to 245 without even really testing 223 beyond 1M Super PI and a few EVEREST memory read benches, let alone any kind of testing for any of the 22 settings I arbitrarily skipped in my greedy quest to hit the next 200 MHz mark. A brash move on my part, my only justification being I already knew Angie was more than capable of 2200MHz.

 

From what I remember, my first attempt was to boot at 245*9 with no divider and all stock voltages. Failure. Wouldn't get past the screen that says "Backing up CMOS data". I think it actually froze just before it said that and rebooted. So I went into the BIOS and changed the FSB to 240. Not as big of a jump from 223 but it still failed.

 

So instead of lowering the FSB until it booted (or lowering it all the way back down to 223 and worked up like I should've done in the first place), I changed some other things. I lowered the HTT/LDT ratio to X3, I set the CPU VID to 1.450V, and I raised NB VID to 1.6V. Still no boot. So I raised the NB VID to 1.7V.

 

I know there was also a blank screen with a couple random ASCII letters at the top involved in one failed boot, but I forget when. I know it was after I'd decided to boot at 245 but I don't know if it happened at 245 or 240, and at what voltage.

 

 

And then it happened. I got the same failure I'd always gotten when the board tried to boot up at too high an FSB with too low a divider, only this time DEBUG lights 1 and 3 lit up at that screen, and then it rebooted.

 

 

It never got past the POST screen. It would detect my DVD-ROM drive, detect that no other PATA drives were installed, and then get stuck with "Detecting IDE drives..." written beneath them. I figured there was a setting set too high but I couldn't get into the BIOS to change it.

 

So I cleared the CMOS. And I cleared it again. I cleared it a third time, each time for a longer period of time. It still wouldn't get past the POST screen. I'd leave it and it would do nothing. I'd press DEL and it would do nothing. And what was weird is it reported my RAM as having 2T timing. I think that was just a result of having cleared the CMOS. At this point I've prayed to God several times and done just about anything else I could think of.

 

So I decide just to go to bed and leave the damn thing stuck at the POST screen and see if by the time I'm awake it gets into the BIOS. I'm halfway out the room when I receive this uncontrollable urge to shut the system down and unplug my hard drives so I don't lose any data in the six hours of sleep I was about to get.

 

 

Wouldn't you know it? The board booted right up. Not to Windows, of course, because that was on my unplugged hard drive, but it got past the POST screen and it seemed it wasn't frozen anymore.

 

So I shut down and plugged my drives back in. I boot back up and it freezes at the POST screen in the same spot it did before (only this time at least my RAM timings are right). So I shut down, unplug all drives except my main drive, and it does the same thing. I'm beginning to see a pattern here...

 

It became evident to me that the board only freezes when there's a drive in one of the first four SATA II ports and, hence, that there's something screwed with the nF4's SATA II controller. Luckily I have an SLI-DR which has four extra ports on a separate RAID/SATA II controller, so I just plugged all my drives into there and was finally able to boot.

 

And now here I am. Running stock right now. Haven't been here in a while, I've forgotten how cool everything runs.

 

Any thoughts? What do you think went wrong? Did I fry something real good? Am I lucky to be alive?

 

If I did fry the SATA II controller, I guess I'm lucky for a lot of reasons. I'm lucky that's all I fried, and I'm lucky that given what I did fry, I didn't lose data. So if that's what happened, I guess that's what happened. I'll take my second chance graciously and OC more smartly in the future.

 

 

 

On a lighter note, I think I'm a much more interesting writer at 2:00 in the morning...

[dr_bowtie]

Here is what I would do....pull ALL power connectors from the board....ATX and extra power... move cmos jumper to clear and pull the battery..

 

Leave it a good long while..a good day or so will be good...

 

If you think you in some way corrupted your hard drive wipe it clean...and fresh install it again...

 

 

as for your overclocking meathod do this... Prepare a bootable CD with the Memtest image on it...you can either download the ISO from memtest.org or download the "UBCD" from the dot com....I choose to use the Ultimate Boot CD as it has all the other tools I use also...

 

Beings you know that the cpu is good for 2.8gs and the right voltage to need to work on your ram and finding the Correct seeting to get this thing overclocked at its best...

 

Here is what I do...I boot to memtest and select either test #5 or #8 (in the end I run both)

 

start clocking it up till you get just a couple errors per pass...dont leave it run... reboot into bios...Now here you can try giving it a little more voltage and see if that does it... if it does up the buss some more and re-try memtest..

 

if the voltage doesnt help then here is where the work is...try changing 1 and only 1 setting at a time and boot back into memtest and note wether the errors per pass are more or less...if less...boot back into bios and change that Same setting again and see if it helps more or less...

 

If it didnt help any more change it back...If you get error free then up the buss again till you get errors and try changing the same setting to see if it helps with the errors...if it doesnt move to another setting and try...and so on and so on...

 

I will run 6 passes of each test unless it errors...if it's erroring I will only finish that pass so I can count the errors per pass....then reboot...

 

Some settings may help if you change it 1 step while other may make it worse you'll just have to play with each setting...

 

do it this way is very time consuming and takes a few hours but on some ram that would clock well I did this and gain 50mhz so it was worth it

 

basically just just have to find the right timing(s) to relax or tighten to get the ram to play right with the controller...

 

I had to do this yesterday on my TForce 550 rig witt the 3600 Brisbane...I put in some OCZ 2x512 ddr533 and it would error in Orthos in 5 seconds at stock...

 

The ram was rated at 3-3-3-8-1T and I changed those to 4-4-4-12-1T no go still errored out...changed Async latency from 4ns to 8ns still no go...uppped volts to 2.1v no go... Changed tref from 3.9us to 7.8us no go...

 

All of these errored out in 5 seconds no matter what...at stock setting...then it hit me...(try the trfc....) so I did...I dropped it to 105ns from the 75ns and put everything else back to stock settings and what do you know....it's still priming away a day later...

 

just one single setting and botch a good clock...you just need to find it...

 

and beings you have OCZ...try their forum for some advice maybe you can get some good optimized settings..? enjoy

[General Septem]

Here is what I would do....pull ALL power connectors from the board....ATX and extra power... move cmos jumper to clear and pull the battery..

 

Leave it a good long while..a good day or so will be good...

 

If you think you in some way corrupted your hard drive wipe it clean...and fresh install it again...

 

I don't think I corrupted my hard drive since I'm in Windows right now. I can check for errors if you think it'll pick it up, though. I'll also try that long CMOS clear, thanks. One question though, you said to remove the jumper, do you put it in the up position or remove it completely?

 

 

as for your overclocking meathod do this... Prepare a bootable CD with the Memtest image on it...you can either download the ISO from memtest.org or download the "UBCD" from the dot com....I choose to use the Ultimate Boot CD as it has all the other tools I use also...

 

and beings you have OCZ...try their forum for some advice maybe you can get some good optimized settings..? enjoy

 

Thanks for the tip. I'll definitely give this a try.

 

So you gained 50MHz CPU or FSB?

[ReelFiles]

Maybe the SATA part of the bios got corrupted? Tmod would know more about that than I.

I would re-flash and start from scratch, without any shortcuts

[General Septem]

I ran CHKDSK and it found no errors.

 

Thanks Reel, I'll add reflashing to my list of things to try. Hmm, now the decision is, do I do a CMOS clear first or do I flash first?

[dr_bowtie]

Always clear the cmos 1st...you want to make sure it's clean/clear before flashing...

 

Yes.. I meant moving the jumper to the clear position...and yea I gained 50fsb alone...but it took 4-6hours of my time testing each and every setting...

 

Some one with alot more experience and understanding of how ram actually works would not have to go thru all that I bet...I learned 1 thing...patience...lots of time....I guess thats 2 things...

 

some ram will clock right up while others will struggle for 1mhz...it just doesnt seem fair but it is what it is...lol

[General Septem]

Always clear the cmos 1st...you want to make sure it's clean/clear before flashing...

 

I didn't end up clearing it first (although I'd cleared it a few times last night) but I did clear it after I flashed it. It fixed the problem though as I'm running from the main four SATA II ports now and everything is working great. Thanks for the help.

[General Septem]

Your idea worked, Doc. I'm currently running at 250 FSB, which means my RAM is at DDR500 and my CPU is running at 2250 MHz. I can probably go even further but I haven't tried yet. The last thing I did last night was raise the FSB to 250 and I let memtest run all night. When I woke up this morning it had made 25 passes and no errors.

 

I started at 2.5-3-2-8. I ran into two snags, the first time I bumped the voltage to 2.7 and the second time I changed Tcl and Trp to 3. At 3-3-3-8 and 2.7 volts it went up to 250 and I'm assuming beyond that with no problems. Thanks for the help.

[dr_bowtie]

see how easy that was...

 

the farther you push it the more the secondary timing will come into play...

 

Setting the secondaries correctly may enable you to run 250fsb at the 2.5-3-3-8 timings...

[General Septem]

see how easy that was...

 

the farther you push it the more the secondary timing will come into play...

 

Setting the secondaries correctly may enable you to run 250fsb at the 2.5-3-3-8 timings...

 

Thanks. Which timings comprise the secondary timings? Also, does a Tcl of 2.5 have a large impact on bandwidth as opposed to 3?

[General Septem]

So I continued today, and managed to get up to 265 FSB before my RAM ran into a snag that it wouldn't get past. When I booted into memtest I didn't even get a chance to tell it to run test #5 before it ended up with thousands of errors just from test #1. I bumped the volts a little and ended up with 28 errors. Beyond that I tried changing every timing I could think of, many of which in either way. Some did nothing, some made it worse.

 

These are Infineon chips according to Ryder though, and my sources tell me that Infineons usually reach 260 maximum, so I'm alright with that.

 

So I lowered the volts back down and the FSB back to 260. I'd run memtest at 260 prior and it passed with flying colors.

 

It wouldn't boot into Windows, however; it would get as far as the Windows XP splash screen with the bar thing, stay there for about 10 seconds, and then reboot. I tried raising CPU and NB volts a notch and that did nothing. I'm back to 250 right now. Anyone have any ideas? Should I try raising volts just a little more?