may have reached this chip's limit

[Stratus_ss]

ok so the parts

820D
P5WD2-P
PQI 667 in dual channel mode
600W OCZ Powerstream SLI

Cooling
6002-775
D5
Maze 4 Chip
Maze 4 gpu
bip II
4 delta 7v

ok so my question is this. I am now stable @3.82ghz but I cant get it any higher. I have not done the vdroop mod to the board. I am wondering if that is the limit for this chip or is there some other bottle neck?
I am thinking about yanking the PSU eventually to the 520 Powerstream but will that make any difference?
any help is appriciated

I'm running the 0606 bios with the following settings

OVerclock=manual
ram speed set to auto
vcore= 1.5125
vdimm=auto


I should mention also
temps are @42 idle 55 under both core loads

as well I can get the board to post upto 4.13ghz but windows just wont go

[Mr. $T$]

Your reaching the limits of the dual 20a rail on your PSU, however it's also close on the processor. How much is your corevoltage droping on your processor between load and idle?

[thegreek]

I am thinking about yanking the PSU eventually to the 520 Powerstream but will that make any difference?

Yes, also why is your temp so high on water?

[Stratus_ss]

well to the best of my knowledge for this chip that is overclocked that isnt that high.

Under load it bounces between 0.03 and 0.05 below what it is set at.
On Idle its at 0.02 to 0.03 above what it is set in the bios

[Mr. $T$]

Alright, try running your fans at a higher speed. Maybe 9v or so, see if the heat isn't the cause.

[Stratus_ss]

ok i have done that... I have the air being pulled through the rad entering the case and then being sucked out the back by both the psu and a 120 mm exhaust

dropped the temp to 47-48 load

[pscout]

i haven't tried extreme oc'ing on my smithfields, but none of my 830's will go past 275 fsb on air. Never tried water on em yet (till my goethermal is running) but my suspicion is that smithfields won't run 724 fully loaded much higher. You might get higher suicides?

[Stratus_ss]

well i have been running stability tests for at least 30 minutes so far got up to 3.95ghz cant seem to get into windows with any higher... dont know if its the board or the chip

[pscout]

I ran 1.5V on my 920 for a week or so dual folding ... sort of by mistake cuz i forgot to lower when i lowered my fsb to 295 from 305 to get folding 100% stable 724. 295 needs 1.4375 on my p5wd2 with the 920. Temps didn't really alert me since i was on water.
It has been running like that for a few months now.

When I first got the 920, i tried it all the way up to 1.5625 vc to try to hiit 4.5 ghz. I could get a cpu-z verified there but nothing useful was stable.

In all my 5 other dc rigs on air , only 1 runs over 1.425 vc, (the latest 930 at 1.4375 just to get 270 fsb :eek: ) but any slight 5-10 bump in fsb isn't worth all the extra vc it takes, and i find in all cases, it really isn't 100% error free running at the apparent oc limit.

My guess is that even on water none of my 830's will be 100% stable at 275.

But as we all know, every chip is different. But your VC is already up in the high range.

From what I understand, the vdroop will help you from overvolting at idle and so is useful when going for very high oc when you need to raise vc a lot. Not sure how much it helps with stability on the oc itself ... the vc on all my asus 945/955/975 mobo's seems pretty stable under full load (which they always are folding 724).

I would expect that running at the very max fsb you can get, despite all the stability tests, you will still get errors from time to time and depending on what you are running, you may not notice them.

Since mine fold all the time, they show up even the infrequent errors, and my response is to back off the fsb by 5 unless it is obviously a temp problem. Folding can produce Early Unit ends which are sort of like errors. very few of them will end that way due to the simulation, at least outside of the beta testing i do, and so early unit end are generally caused by cpu or memory errors. When a rig returns an early unit end, stanford reaasigns the work unit it to several other machines and if they all end at the same point, it is a simulation problem, not a machine error.

The point of all this is to describe how imperfect the hardware we run on is. On mission critical systems, ECC memory is always used since we can't afford to have 'normal' memory errors corrupt data.

Of course, none of the above stops me from finding out how fast a chip will go. I had my first 930 almost striaight to 300 fsb and able to boot windows.
By the time i got a clean pcmark04 and p95, it was back down to 285 and has been folding error free there for nearly a week.

I suspect that getting past the barrier you are now at will take more cooling than water can provide.