Is OCCT reliable at all???

[dust]

I managed to bring the fsb to 440 x 8 with the lowest voltage I could and use my system without any issues yesterday, even I played Warhead with maximum specs for nearly three hours without any issue(1280x1024).

I even run the test on Everest for nearly an hour just to see wht's that all abt and the temp graphs were fluctuating between 61 - 65 c on the fourth core (the highest temp of all).

I also installed PC Probe( heard it wasn't the best temp monitoring soft but I just wanted to compare the results) and when I ran it just before I was going to shut down the comp it gave me the MB TEMP at 191c. The NB was at 45c, SB abt the same perahps higher, I don't remember exactly, point is all the other records were below their threshold.

This morning I ran OCCT to test the stability and after five minutes it stopped by itself ( CPU too hot) and the temperature graph was showing two cores reaching above 70 degrees and in continuous growth.

All is fine but at the same time the Core Temp was reporting a maximum of 64 degrees on the fourth core, all the others being lower than that.

I ran Real Temp to check the sensors and they were all at 13 except for the second one which was at nine( could be stuck, it's always reporting 8-10 degrees lower temp than the others, under load).

My ambient temp is ~24c and I removed one side from the case several days back when I started playing with the voltage

My voltage values:



Vcore - 1,4 v( 1,38 v in CPU-Z)

NB - 1,41 v

CPU PLL - 1,55 v

Vtt - 1,3 v

DRAM - 2,1 v

SB - 1.125 V

SB 1,5V - 1,6

CPU GTL - 0.67X

NB GTL - 0.67X

DRAM C REF - auto

LLC - enabled( I was expecting a lower vdroop than this, maybe I'm wrong, I didn't do a pencil mod)

DRAM Static control - dis

 

[LOUISSSSS]

OCCT is reliable, it stopped it because it felt your CPU was too hot, in that case (given the set TJ max in OCCT) it stopped bc your CPU was 70C. It is a safety precaution.

 

[VirtualLarry]

Originally posted by: LOUISSSSS
OCCT is reliable, it stopped it because it felt your CPU was too hot, in that case (given the set TJ max in OCCT) it stopped bc your CPU was 70C. It is a safety precaution.

That's a bit conservative. C2D can run up to 90C or more perfectly stable. They don't burn themselves up.

 

[Tempered81]

yes they shut themselves off before killing themselves because of heat.

 

[dust]

all right but how abt the core temp showing nearly 10 degrees lower temp at its highest and the Pc probe showing 191c for the MB whilst the NB and SB sit comfortably at 45?
Yesterday I had nothing better to do but to install whatever I could find regarding the temp, Real Temp, Speed Fan, Pc Probe, just to compare the data but this is just strange.

I could accept the real temp shown by OCCT but I found better reviews on Core Temp and I tend to consider it more accurate. A friend of mine is fighting with a q6600 at 3,4 and had exactly the same problem whilst his entire system is like a fridge. After ~ 5 mins he gets the same message and he tried with the stock settings as well getting exactly the same result.

The funny part is that I ran the same test not two days back with nearly the same voltage but at 425 FSB and it ran fine for an hour , giving me a correct graph for the temp oscilation, more or less 60c troughout the test.

I didn't change anything, I'll reinstall both OCCT and Pc Probe and post the res.

BTW during this fiasco the Speed Fan was showing minus 65c for the system temp, how abt that

 

[LOUISSSSS]

Pc probe is not reliable, don't use it.

The temp you're getting is a GUESS. it is calculated by a preset value caleld the TJ MAX. Intel did not release any word on that so all the programmers can do is guess.

u can trust coretemp, realtemp, hwMonitor, and the latest speedfan. they may have 5c core temp reading differences between each other.

I'm not sure if you're calling my statement conservative or the programming of OCCT to shut down itself when it reaches 70c or so. i'm not the one telling it to shut down.. lol

 

[dust]

Originally posted by: LOUISSSSS
Pc probe is not reliable, don't use it.


I'm not sure if you're calling my statement conservative or the programming of OCCT to shut down itself when it reaches 70c or so. i'm not the one telling it to shut down.. lol

LOL

 

[dust]

All right, I unistalled everything and reinstalled only OCCT and Pc Probe( I'll throw it as soon as I get home).

I also reduced the VCC by one notch so it's at 1,39 in Bios and 1,376 in CPU-z. I ran the standard test of one hour on OCCT and the temps were fluctuating between 63-69C on the fourth core, the highest rec temp of all the cores. The test completed and the sys stable message appeared at the end. The temps were exactly like in the Core Temp report throughout the test but of course the PC Probe saw some voltage values at 0 for several secs every now and then on the SB, NB and whatever else it could think of. I agree the soft sucks( Pc Probe) and the OCCT is fine.

Further more it is likely that the previous test result made sense since I reduced the CPU voltage and the temps still reached 69c.( My threshold for nearly everything is at 70c in Bios)

I did notice the test was actually a mix of the other three types of tests the OCCT can run, more or less Prime95 tests. I think that's the main reason for the temp fluctuating within 6 c range and I still have to know what happens if I run the large fft test since it raises the temps a lot. Will see over the week-end.

Thanks guys for the support

 

[amenx]

You can adjust the OCCT temp threshold in the settings tab. Its conservatively set at 70, I put it at 75.

 

[dust]

I'll probably have to do it to reach 3,6 MHZ since I doubt I can get it without more vcore.

 

[Salaminizer]

I don't think you can run all those different temp monitoring programs at the same time. They use the same hardware resources and will get conflicts or odd data if you have everything going at once. This includes using a monitor like pc probe at the same time as OCCT, because again, they use the same hardware points for reading temps. The one with the higher priority usually will 'win' and the other will probably get dips/odd data.

 

[dust]

you're probably right, just before my last test I ran CPU-Z and Hw Monitor and as soon as I launched OCCT it warned me of the same thing exactly. It didn't seem to mind Core Temp though. I'll run the test again without any "audience" and I'll compare the graphs. If I get different values I'll post again.

For long sessions of OCCT and Prime95 which is the recommended test,the small fft or the large fft?

 

[supastar1568]

I have failed Prime and passed OCCT, and also passed Prime and failed OCCT. Maybe both?

I personally do 100 passes of IntelBurn, 24 hours Prime Blend, 12 hours prime small fft, 8 hours occt (mix), 10 hours memtest. Thats just me, long process.

 

[LOUISSSSS]

is your cpu already stable at 3.52ghz? and you're doing all this extra work to squeeze out an extra 8mhz to get 3.6? you know you wont notice ANY difference right?

^^ you're probably the type to stay up late at night playing 3dmark06 and anxiously waiting 09 to come out

 

[dust]

Originally posted by: LOUISSSSS
is your cpu already stable at 3.52ghz? and you're doing all this extra work to squeeze out an extra 8mhz to get 3.6? you know you wont notice ANY difference right?

^^ you're probably the type to stay up late at night playing 3dmark06 and anxiously waiting 09 to come out

Well that's correct actually, but what I'm trying to do is to reach the max FSB with air cooling only and of course within the safe temp limits, once I'll get there I'll probably return to 3,4 MHZ since it gives me more than enough for my needs. The point is actually to understand the whole thing since I'm new to this.

Regarding the 3dmark 06, I ran it several times when I had 3,65 MHZ and got 20700 points, but the high voltage for the CPU and NB I had to use to get there is not something I'm willing to trade for the performance.

 

[dust]

Originally posted by: supastar1568
I have failed Prime and passed OCCT, and also passed Prime and failed OCCT. Maybe both?

I personally do 100 passes of IntelBurn, 24 hours Prime Blend, 12 hours prime small fft, 8 hours occt (mix), 10 hours memtest. Thats just me, long process.

I'll try the same, but I hate long waits, that's usually the time when I screw my phone and lock it. Thanks for the input!

 

[supastar1568]

Originally posted by: LOUISSSSS
is your cpu already stable at 3.52ghz? and you're doing all this extra work to squeeze out an extra 8mhz to get 3.6? you know you wont notice ANY difference right?

^^ you're probably the type to stay up late at night playing 3dmark06 and anxiously waiting 09 to come out

you're probably the type that considers an overclock stable at 30 minutes of prime