Thermal Fluctuation on D900K

I have been having lots of stability problems with my D900K since I got it. I undervolted the Opteron a little bit, but I never tested each multiplier for stability. I was starting to get really pissed off when I would be working and all of a sudden it would shut off. My cpu temps would usually be at 55C under full load.

Yesterday and today I started checking stability at each multiplier. I have set some of the intermediate multipliers down a little, but something strange is happening.

My peak cpu temperatures at full load at 2600Mhz per core is 117F or 47C. This is much lower than it used to be, and its a pretty hot day today.
I was just wondering what may be causing this immediate drop. I actually had to raise the core voltage at multipliers 10.5 - 13.
My idle temperature is also around 85F or 30C.
This is strange cause I also had to up my core voltage at 1000MHz which is the idle speed (to get it stable). At idle my cpu is using 1.075V x 82 Amps or 88.15WATTS of power. at max power the cpu is using just shy of 118watts of power usage. Still that is a ton of energy, and its stange to all of a sudden have much lower temperatures.

Any ideas?
The only difference is that I upgraded to the latest version of CPU rightmark

K-TRON

 

K-Tron, believe me when I tell you this.

The Opteron's were not made to work perfectly in this system. The BIOS may detect them, but the low-level Cool N' Quiet is the main issue that is messing with the Opterons.

I unsuccessfully made an attempt to re-write the BIOS to force Cool N Quiet to be disabled... its got to do with some firmware as well... I hate this VIA K8T800 hybrid chipset... too hard to find resources about compared to nForce4.

I actually have the same issues as you with the random restarts... but got it to stabilize with RMclock settings.

I found that any Athlon X2 and FX works perfect with no issues, while the dual-core Opterons do... which sucks.

I am actually running a Athlon X2 4800+ in my system now for stability tests... left everything the same with my Opteron 180.... all is good no issues. (even with no RMclock on)

Then when I put in the Opty 180, its back to some random issues (mainly crashing games to desktop).

I think I will be getting a FX-60 and end my issues with CPUs for this system forever... especially since the FX-60 is the fastest and fully supported CPU specified by Clevo.

... while Opterons are not specified.

 

I have force disabled cool and quiet, and it seems more stable now. I had to up the idle voltage from 0.875V to 1.075V to get it stable at idle.
What rmclock settings did you set?

I found that a lot of stability was gained by setting rightmark to use the AMD certified p-state transistors, and AMD half FID.

It is running stable now, and I have managed to get it stable on all of the 18 multipliers. Their is really 9, but with AMD half FID their are 18.

It was just very strange, that sometimes I would boot and it would just reboot itself like 3-4 times before running fine. Than sometimes it would crash randomly.
I thought it was my Raid 0 array acting up, but it does the same with different harddrives and in different raid setups.

I just spent literally 40 hours over the last few days clocking in 90 hour at full cpu power, for each multipler for testing with Orthos.

I think its running more stable now.

GOPHN, here are the Voltages I have found to be good with my Opty 185

5.0x = 1.0750 1000MHz
5.5x = 1.0750 1100MHz
6.0x = 1.0750 1200MHz
6.5x = 1.0750 1300MHz
7.0x = 1.0750 1400MHz
there is no 7.5x for some reason
8.0x = 1.0750 1600MHz
8.5x = 1.0750 1700MHz
9.0x = 1.0750 1800MHz
9.5x = 1.1000 1900MHz
10.0x = 1.1250 2000MHz
10.5x = 1.1500 2100MHz
11.0x = 1.2000 2200MHz
11.5x = 1.2250 2300MHz
12.0x = 1.2500 2400MHz
12.5x = 1.2750 2500MHz
13.0x = 1.3000 2600MHz

I hope that once I finish my last stability tests on the 11.5x multiplier, it should be much more stable. I hope it works better. It seems to be running much more stable, but the drop in temperature is concering me, cause core temp and get thermal, are both reporting lower temperatures despite the hot weather.

EDIT:
Its now ORTHOS approved stable at all of the multipliers. I am going to start doing some restart tests and see what happens. If it holds stable, than I would be really happy. If not, back to the drawing board.

K-TRON

 

It seems to be running stable now, I have rebooted it continuously and it seems to be working fine now. I think all the Opteron needs for stability is the AMD certified tranistors. It used to be unstable with the standard P-state transistors, and now its stable with the AMD ones, so maybe Gophn, try it and see if yours uns better. Mine seems to be working good now.

*crosses fingers and prays it continues to run stable*

K-TRON