Lucid155

Distinguished
Mar 17, 2011
25
0
18,530
Been trying to overclock my 955, i have an antec 900 v2 case, hyper 212 cooler, 8gb of Ripjaw RAM.
I have it set to 3.8ghz, multiplier at x19, 1.4v. idle temp: 38C Load:49C. Does this seem like the peak in terms of temps? Also my NB freq is at 2000MHZ, i have this mobo (http://www.newegg.com/Product/Product.aspx?Item=N82E16813128435) should i be setting that freq to something higher like 2.4 or 2.8? Will i see a speed increase?
 


In my opinion, those are nice temps for 3.8 Ghz. Is that CPU or core? I'm guessing CPU temps, what you want to avoid is going over 60C load on the cores. Its possible to get the 955 to 4.0 Ghz, but I'm not sure you'd want to go too much higher as core tmeps on Phneom II are right around 10C higher than CPU temps report.

As far as the NB goes, you don't want to clock it higher than the HT speed. This wont be an issue unless you start to mess with the reference clock. So, 60C max on core load temps and don't go higher than 1.45V unless you obtain more exotic cooling. Really though, 3.8Ghz is nothing to sneeze at.
buzznut
 

JofaMang

Distinguished
Jun 14, 2009
1,939
0
19,960
NB only becomes a bottleneck when it is less than 1.5x your memory speed. If you are running 1333mhz memory, then nothing above 2000mhz will make much of a difference. If you are running 1600mhz memory, then OCing the NB to 2400mhz will show significant gains in performance. You should be able to run 2400mhz NB with only a small pump in NB voltage, though some boards can run that speed at stock voltage.
 

Lucid155

Distinguished
Mar 17, 2011
25
0
18,530

Im running the 1600mhz ripjaws, am trying to bump up the NB freq to 2400mhz and still messing with the voltage, im at 1.4v and still getting an error on worker #4 during prime95 testing. Do i need to also bump up the HT freq while im adjusting the NB freq and voltage? Should i not go any higher than the 1.4v on the NB?
 

JofaMang

Distinguished
Jun 14, 2009
1,939
0
19,960
Which Prim95 stress test are you running?
How long does it take for the error to occur on work #3?
How long have you stress tested it before raising NB?

NB OC will not cause individual workers to fail, generally an unstable NB OC will cause blue screens and sometimes even require cmos reset. That being said, a NB OC will open up your system and perhaps could be highlighting an unstable OC on the CPU that otherwise would have been dormant or at least sporadic. You may (probably) need more voltage on your CPU. Since you are still at only 49c right now at current voltage, a small bump shouldn't rocket your temps much at all. 62c is really bad, but I always tried to keep it below 55c load for that extra buffer.

HT can stay at 2000mhz, some hardcore benchers and overclockers will jack it up, but in my experience it has a very small effect on overall performance.
 


I've always heard you want to keep the HT and NB speeds equal for best stability.
 

Lucid155

Distinguished
Mar 17, 2011
25
0
18,530
set NB and HT to 2400mhz, no adjusted voltage. set my 955 to 1.5v at 3.8, that seem to do it. Ran for about 20 mins on prime 95 small fft's, no errors. Touched 56C at times but goes back to 55 and remains there under full load. just ran a 3d mark vantage benchmark and it BSOD during the second graphics test, set NB and HT back to normal. what should i do?
 

JofaMang

Distinguished
Jun 14, 2009
1,939
0
19,960
While that board has many great features and is decently built, it has one major flaw for OCing, and that is the 4+1 power system. This will limit the stability of the power delivery to the CPU. I owned this board and could benchmark some decent clocks with my 1090t, but only for very short periods, and usually with very cold winter air blowing over the board. With my 965 BE, I could only see a stable 3.8ish daily OC, but got as high as 4.4ghz benching outside at -30c. With my 1090t, i had the same daily limit, 3.825ghz, but even lower max of 4.2ghz, again, only with cold weather benchmarking.

Yes, you may be at the limit for your board.