Long day's journey...

[[TPG]Grinder]

I promised a post regarding my need for assistance with my RAM woes, so here it must be, although I seem to have fixed the problem.
My recent first build consists of the following:

E6600 C2D (oc'd to 3.0ghz @ 1.325v)
EVGA 680i version A1 mobo (bios updated to P28)
Patriot PC2-8500 DDR2-1066 memory 2x1gb
EVGA 8800GTS 640mb (oc'd to 586/1.7)
Seagate 7200.10 NCQ 320gb HDD
850w OCZ GameXStream psu
Scythe Ninja Rev.2 (w/ 79.8cfm 120mm fan)
Apevia X-Plorer Case incl. 2x80mm intake fans, 1x80mm dorsal exhaust, and 1x120mm rear exhaust, plus an 80mm Antec Spot Cool fan hovering over the DIMMs.

After putting it all together, I just had to get into the BIOS. I'd read a ton of info regarding overclocking the 680i, the e6600, and the Patriot RAM, and couldn't wait to crank up my new rig.

Initially all went well, I bumped all the voltages, killed the thermal controls, unlinked the cpu and memory and set the fsb to 1333 and the RAM to 1066 (5-5-5-9 2T). It all ran fine, although the only stability test I used was the one in nTune. I gamed on it for a few nights, then clocked the memory to 1200, rather recklessly in retrospect, but it kept the wheels on for another long night of CoD2. I was very pleased.

The next night things got hairy. I crashed repeatedly while gaming, and again the next night during gaming and video playback. I tried bumping the vdimm to 2.4 in order to hold the high frequencies, but got bars on post and backed it immediately off. The RAM would no longer boot Windows at 1200, and indeed, would not go to 1066 any more and hold on for more than a few minutes in Orthos. I finally had to settle for 888.5megs as my highest stable frequency, accepting the 3-4-4-7 timings as a reasonable consolation. I also bumped the fsb to 1520x8 (3.0ghz) during this period btw.

Then tonight I thought I'd try one more thing.

As you may know, the Scythe Ninja is no small heatsink. And with a rearward facing 120mm fan mounted on it, the first memory slot is blocked by the fan. I had been running the two memory sticks in the 1 & 3 slots since build. I finally openned the case up and removed the fan, installed the RAM underneath and reattached the fan high enough to clear. And Voila!

My system booted like a charm to 1066megs and 1333fsbx9, and long and short of it is that while I'm posting this, Orthos blend shows my system stable over the last 1hr and 6mins and counting.

I'm really hoping this may be the very mundane solution to my weeks long troubles. I have done and read everything I could regarding RAM issues with the 680i, and had myself convinced it was lacking in support for memory above 1000mgz, but here I sit at 1066mgs as I write! My fingers are crossed.

Thanks for taking the time to read through all this. You guys are great, but sorry for the long post.

1hr 21 mins, and counting...

Now to find out just how far this RAM will OVERclock!

[Dr. V]

LoL I look forward to your results.

One quick thing though - posts that are all in bold are very striking and quite annoying to look at. Stick to emphasizing with it - it makes for a more enjoyable and mellow canvas to read.

[Hitman]

Originally Posted by ST!X

LoL I look forward to your results.

One quick thing though - posts that are all in bold are very striking and quite annoying to look at. Stick to emphasizing with it - it makes for a more enjoyable and mellow canvas to read.

Doesn't bother me, as long as you attempt to use the English language properly then I am good. Bold whatever you want, its your post.

[Elysium]

Keep us updated.

[One4yu2c]

Originally Posted by [TPG]Grinder

EVGA 680i version A1 mobo (bios updated to P28)

I tried bumping the vdimm to 2.4 in order to hold the high frequencies

Let me caution heavily against raising your RAM voltage that high. When stressed, 2.4v can get those chips pretty damn hot, and you're at a greater risk of killing your RAM. This has been a particular issue when coupled with the 680i chipset, and specifically EVGA's iteration. I covered this a few months back, which you can read more HERE.

[[TPG]Grinder]

Aaah...what a welcoming group!!

Thanks for your ongoing support and reassurance.

I had Orthos run blend while I slept, and it reported no errors for 6hrs (until the wife arose and turned it off). I had the system running at 1333/1066 5-5-5-9 2T for the test and, lo and behold, stability!

Apologies for the bold print in my earlier post; I wasn't aware of having done that. Assume nothing of my computer skills. I am 42 yrs of age, and my first home computer came into my possession less than 3 yrs ago.

Can you say green?

That first system was a pre-built from a clone manufacturer, and with the exception of adding a sound card, and updating the vid card (AGP), I hadn't opened the box in all that time. I have researched many hours of info online for this current build, but know that there are a great many neophyte errors inside me awaiting their turns to be had.

As for the 2.4vdimm issue, I have read your RAM frying entry during my ongoing memory issues, and found it quite enlightening. That was after my post bars experience, and I haven't run that voltage since, as I also found a comment from EVGA that 2.4v IS frying RAM in the 680i...but don't blame them. Thanks for pointing it out to me, One4u.

All that I know has been gained from reading articles and forums, like this one. I consider all your efforts a very generous contribution to the collective wisdom, and offer you all a tip of the hat.

I'll be reaching for higher overclocks in the very near future. As soon as I recover from my most recent experiences of it.

But until then, here is a question:
If I am not bumping my voltages on cpu or ram, am I shortening their life expectancies by increasing their respective frequencies?

Till later, thanks again for everything!

[One4yu2c]

Originally Posted by [TPG]Grinder

But until then, here is a question:
If I am not bumping my voltages on cpu or ram, am I shortening their life expectancies by increasing their respective frequencies?

In theory, you potentially could be. In reality, not likely. Consider that some of these lower clocked processors are sold at their respective frequency to meet a pricing demand, and not necessarily because they've been invalidated to run faster. The market for high end (and pricey) chips is relatively small compared to the mainstream, and many of these lower clocked chips ship at their respective frequency to supply the demand. Same concept applies to RAM.

The disclaimer here is that even without increasing voltage, there is a risk of damaging a component due to overclocking. Especially with the processor, the risk is pretty minor (a system will shutdown or refuse to boot at too high a frequency), but it still exists.

As a rule of thumb, proceed in baby steps. Keep a close tab on your temperatures, which should be a habit throughout the life of your system, and utilize stress testing programs to test for stability (Memtest68+, Prime95, Orthos, 3DMark06, etc). While the risk can never be eliminated completely, you can reduce it to almost nil with intelligent and patient overclocking.

[[TPG]Grinder]

Thanks for the speedy response. I will proceed slowly to milk these components at modest volts and careful increments, and keep you posted.
Cya soon.

Grinder

[[TPG]Grinder]

Praise goodness!

Been enjoying some stability since the RAM slot change. I ran the Orthos blend test for 6hrs/41mins. last nite at 1400fsb (350 x 9 = 3.15GHz), and 1120 memory frequency; no errors.

Had the memory timings relaxed to 5-7-7-15 as a precaution. Raised the vcore to 1.4v and the FSB to 1.5v to be sure it would hold, but I will work on reducing these voltages, along with the memory timings, tonight.

Question:

Is the gain in frequency a fair trade off to the relaxing of the timings? Or, will I need to run 3Dmark in order to know? I have yet to do so.

Thanks again!

[One4yu2c]

Originally Posted by [TPG]Grinder

Question:

Is the gain in frequency a fair trade off to the relaxing of the timings? Or, will I need to run 3Dmark in order to know? I have yet to do so.

Thanks again!

The answer is almost always 'yes.' A processor's clockspeed trumps RAM timings in most scenarios, synthetic memory benchmarking notwithstanding.

Going from 3.0GHz to 3.15GHz represents an additional 150MHz overclock, which isn't a tremendous boost over where you were previously at. But the Core 2 architecture scales well from a performance standpoint, so I still suspect the small boost will be worth the latency trade off. What I'd suggest is in addition to 3DMark06 is to benchmark the games/apps you use most to see which setting is yielding better performance. Don't get caught up in specialized synthentic benches, as the results can be misleading.