Opteron 165 Overclocking

Can anyone share experiences with increasing FSB voltage or chipset voltage on the A8N-SLI boards? I'm asking because I suspect that my overclock is being limited by motherboard's inability to reach HTT higher than 280. My Opty 170 is stable at 280x10 with just a hair over stock voltage (1.3625V), yet I can't get 285x10 stable even with 1.4125+ VCore. Would increasing FSB or chipset voltage possibly help stabilize higher overclocks? Maybe my intermittent failure to POST is also related to this?

I am really surprised that board wont do over 280 HTT.

These Opty's need up around 1.45-1.5 vcore to see 2.9-3.0GHz stable.
Things tend to start to get pretty hot up past 2.8 as well especially on air cooling.
What are you using for a CPU cooler? I didnt see you mention it anywhere.

To be honest that was about as high as I could get mine stable on air cooling.
Much higher then that I started having to add a lot of voltage... and heat.
That why I got this Monsoon TEC cooler....got me those last couple hundred MHz

Your 170 is far beyond most at that voltage and that speed. I agree w/ Wanna, you need to add a lot more Vcore to get higher. Most 170s never get past 2.7 regardless of Vcore so you are doing well but one thing about these chips is you hit a wall and the thing won't get stable after that unless you add some really exotic cooling system that drops the silicon's temperature. I think you have room there to maybe get to 2.9 but it will take better cooling and a lot more Vcore.

looks like my flex tube idea may work out. right now the outside temp is 5c. ambient in the room about 19c. idles at 14c and orthos runs it to 38c 3006mhz 1.47v. got to do a better job sealing the duct through the window so i don't freeze. we're supposed to be real cold for the next few days so i can play a bit. course it won't do me any good in the summer here in the desert.

5c......brrrrr. I'd have to own pants if I lived there

Sounds like you have a fun little setup to play with there.
Just be careful as it may want to create condensation.
Keep an eye on things.....dont want any moisture buildup

the temp is now -10c and falling. supposed to be like this the next few days. that cold outside air never goes into the case. just goes straight into the radiator which is mounted on top sort of like a koolance unit. the case is closed and the room temp is about 18c. once i get the window sealed a little better i'll push this up.

QUOTE (WannaGoFast @ January 12, 2007 06:11 pm)

5c......brrrrr. I'd have to own pants if I lived there

KillerBuckeye:

My guess is that you've reached the point where you need to start adding vcore aggressively in order to move up. The AMD dual cores are rather strange beasts in that they will overclock 300 - 400 Mhz on next to no additional voltage and then ask for nearly 0.1v for every 100 Mhz above that point. If your cooling will handle it, I'd goose the vcore to 1.45 for a 285 x 10 set and see if that's stable. If it is, walk the FSB up in increments of one until it becomes unstable. If it's not, walk it down in increments of one until it stabilizes.

looks like i've hit the end on this chip. 345x9=3105. i've tried 133,166,183 dividers. 2x & 3x multi's. 2.9vdimm. up to 1.65vcore. overvolted the nb, sb, and ht. looser ram timings, tighter timings in short everything i can think of it just won't do 346x9 or above. guess i'll have to settle for a lousy 73% oc. ha ha still an amazing chip though. really wanted to beat wanna at the 32m super pi but it ain't gonna happen. did do a 24m08.281s run though. but since i didn't beat him i didn't bother to post it.

Post it anyways. We need to gather data. Beating Wanna is a pretty tough nut to crack, he is one of the best. Best thing is he'll teach you how to do it.

Has anyone had any experience with this memory and their opty/asus rig?

Patriot PDC2G3500LLK
The memory is based on Infineon CE-5 memoy, the review I read says it can go to ddr500 @2.5-3-4-3-7 1T. Reason I am asking is because I can get this new from a local shop for 179.99. Here's a link to the review.

3500LL review

I will move this to the ram section.

I have an update to my Opty 170 overclocking experience. In previous posts, I mentioned that I just could not stabilize the CPU at anything over 2.8 GHz, and that the computer would sometimes fail to POST even at settings that are Orthos stable. Well, it turns out that it's actually my memory causing the problem.

My first clue came a few weeks ago when the computer refused to POST at 280x10 anymore. I had to settle for 275x10 for a while. Then, yesterday I couldn't get a POST at 275x10! Based on my experience with Ballistix memory that degrades over time, I changed the divider from the DDR333 setting to the DDR200 setting (remember, my DDR266 divider doesn't work

). Sure enough, the system had no problem booting at 280x10 with the memory at only 140 MHz. I now have worked my way up to 290x10, which I am currently -*test*-('")ing for stability. It seems that 290x10 with memory at 145 MHz, 2-2-2-6, 1T, gets me roughly the same benchmark scores as 280x10 with memory at 234 MHz, 2.5-3-3-8, 2T.

Hi guys, I haven't posted in quite some time. I just wanted to update you on the status of my rig, and confirm that many of the quirky issues I was experiencing before with unstable memory dividers were caused by the mobo's apparent incompatibility with the Crucial Ballistix memory.

My rig had become very unstable over the past few months, with BSODs becoming more and more frequent. I had been running my rig at 290x10 with the DDR200 divider (yes, the memory was underclocked at 145 MHz!). To see what the heck was going on, I returned everything to stock settings (which actually increased the memory clock), and quickly found that the system was extremely unstable--I couldn't even boot to Windows. A few Mem-*test*-('")86 runs later and I found that I had yet another bad Ballistix module!!! This is faulty stick #4 for me! And this time, I had the stinkin' things UNDERCLOCKED the whole time.

I am convinced there is something really funny with the Ballistix RAM and the A8N-SLI series of mobos... My memory seemed to die every 4-6 months, while all other components were fine. This also happened on 2 different mobos (a Deluxe and a Premium). Heat was not an issue as the modules weren't even hot to the touch during torture -*test*-('")ing. Add to that the fact that my memory dividers never worked right: with my Venice CPU, the DDR333 divider didn't work, but all the other dividers were fine; with my Opteron, the DDR266 divider didn't work, but all others were okay.

Anyhow, I purchased 2x512MB of Corsair VS PC3200 at the bargain price of $60 shipped! (I just couldn't bring myself to spend the money for 2 GB when I'm hoping to upgrade to Conroe within the next year or so). I just popped in these modules and all the dividers seem to work just fine now. I'm currently Orthos -*test*-('")ing 285x10 with the DDR266 divider, which puts the memory at 190 MHz. The only bad thing is that I have to run these modules at relaxed latencies compared to the Ballistix. Oh well, I'm sure the 45-55 Hz increase in memory speed will more than make up for that difference.

Sorry for the babbling, but I thought some of you might find this interesting, if not reassuring. I wish I would have purchased high-end Corsair memory instead of Crucial in the very beginning... Oh well. Live and learn.