UPDATE: E6400 @3.5Ghz/1.42vcore Orthos small FFT's 8hr

[NoStateofMind]

Well I got my E6400 OC to 3.520Ghz finally. But, it fails Orthos and Prime95 within a minute. I played about an hour of DOD:S and not one thing went wrong. My load temps during the game never exceded 52c (Asus A.I.). While running Orthos (small FFT's) it spikes up to 63c (Asus A.I.) and 67-68c in Core Temp. I have vcore in BIOS set to 1.46 but in windows it shows 1.416. Bad vdroop. What I want to know, is Orthos and Prime95 really the tell all on system stability?

EDIT: IDLE Temps- 47c Asus A.I.
53c Core Temp

UPDATE: Got it to 3.4Ghz. Screen Shot here.


Orthos small FFT's 6hrs

EDIT #2: Have it @ 3.5 with 1.42vcore. Orthos 45 minutes so far. Will post screen shot when I have 6+ hours.

Orthos small FFT's 8 hours 30 minutes 3.504Ghz screen shot

 

[MADMAX23]

Yeah, you've already answered your own question....If your system is not stable trough the famous BLEND test in Prime95 or ORTHOS....then your system is not stable, it may seem to be stable, but it isn't....try running BF2 v1.4...and see if it crashes during gameplay...
About CPU temps: You should already know Prime95 stresses the Cpu like no other program, that's why your temps get higher...like mines...it's absolutely normal.
Test, test and retest until you get your system Prime stable....sorry...there's no other way..
Have a nice day!

 

[NoStateofMind]

I understand that the temps will shoot up during CPU stress testing, that's obvious. I can add more vcore and it will get more stable, but, my temps increase a bunch. When I increased the vcore to 1.5 in BIOS it showed something like 1.46 in windows and CoreTemp was at 67c running dual Prime95. I got scared and backed off. During that test Asus A.I. reported 63c.

EDIT: And I dont own BF2, I hate it

 

[MADMAX23]

And...oh, 1.416v (real vcore) for 3.520 Ghz is quite low if you make an average of what others are reporting trough different forums...you should try 1.445v or higher in Windows.
Also note that when you run Prime95 or ORTHOS cpu stress test, the vcore drops even a little more than under any other loading process....check that out with Cpu-Z and ASUS Probe.
For ex; if your Windows vcore is 1.296v, if you run Orthos Cpu stress test, vcore drops to 1.288 or maybe 1.280v....so figure out, always take this lower value as the real vcore.

 

[MADMAX23]

What board do you have? P5B-D?
The sweet spot for the 6400 is 3200 Mhz (400x8)....if you own the P5B-D you should try taht first...

 

[NoStateofMind]

Originally posted by: MADMAX23
What board do you have? P5B-D?

Yep. Just worried about my temps being too high

EDIT: And which do you believe? Asus A.I. or CoreTemp? I relize they are both right, in effect of CoreTemp being the CPU die and Asus A.I being a probe outside of the CPU.

 

[NoStateofMind]

Originally posted by: MADMAX23
What board do you have? P5B-D?
The sweet spot for the 6400 is 3200 Mhz (400x8)....if you own the P5B-D you should try taht first...

I have and it runs great at that speed. 3.2 forever and a day with Orthos or Prime95. I just want that 3.5Ghz!!!!!

 

[MADMAX23]

I trust both ASUS PC Probe and Core Temp...very close temps to each other

 

[NoStateofMind]

CoreTemp is always 6c higher than Asus A.I. for me.

 

[MADMAX23]

I see your needs...3500Mhz...that's going to need some vcore...as I told before...and temps may be hitting 70ºs....try 3.4Ghz...I tested that speed and could achieve it much easier than 3.5Ghz...and with less volts of course...

 

[Thor86]

Stability is a relative term. If you don't care about corrupting your data and just play games, then 3.5Ghz gaming stable is great for you. But most people do not want data corruption, therefore test cpu/memory stability using these utilities so that over time they don't have to worry about losing important data or having to worry about software glitches.

 

[NoStateofMind]

Well what temp is too high? I've seen Intel state that it shouldn't go over 61c. But is that for CoreTemp or Motherboard probes?

EDIT: Running Orthos for 30 minutes @3.4Ghz, 1.368 vcore, Coretemp max 69C, Asus max 65C, motherboard max 45C. Still running.

EDIT #2: It errored after 35 minutes and 5 sec. Gonna raise vcore a notch to see if that solves it :disgust:

EDIT #3: Passed the 35minute mark, no errors yet All max temps increased 1C.

 

[NoStateofMind]

Updated

 

[cmdrdredd]

I just stuck with 3.2Ghz. I can do 3.4Ghz stable but the temps above 65c are too much for me (using core temp or Intel's TAT)

 

[MADMAX23]

My congratulations!....Hope you test your system at that speed for at least 6 hours under ORTHOS blend test...
Very nice again, I see you followed some of my suggestions hehe, nice dude!!
About temps; If your Cpu temps don't get any higher than 65º under heavy load (not ORTHOS or Prime loading) with real world programs or games, you can feel happy and safe.

 

[MADMAX23]

Another thing you should take into account is Ram memory performance when FSB is set higher than 400FSB (like 401 or 425) and lower than 450-470....The reason is the straps...when you set your FSB to 401 your memory subsystem slows down rather noticeably....and you lose bandwidth, that means poorer communication between the Cpu and the Ram mainly...check it out....and tell me what you think later.
However to obtain the same performance when 400Fsb is set, you can up the FSB past 450 and little by little (increasing FSB freq.) regain the same Ram memory performance/bandwidth/speed.
Cheers!

 

[cmdrdredd]

Originally posted by: MADMAX23
My congratulations!....Hope you test your system at that speed for at least 6 hours under ORTHOS blend test...
Very nice again, I see you followed some of my suggestions hehe, nice dude!!
About temps; If your Cpu temps don't get any higher than 65º under heavy load (not ORTHOS or Prime loading) with real world programs or games, you can feel happy and safe.

Blend test is bad for 1 reason. It does not isolate the cause of instability.

For example. Run blend for 5 hours and get an error. Since you're testing your memory and CPU you never know where the error came from.

Test smallFFT for about 12hours by itself. Then run memtest for 12+hours by itself. That way you know whether your CPU can do the speed or whether your memory speed or timings are causing problems.

 

[NoStateofMind]

Originally posted by: MADMAX23
My congratulations!....Hope you test your system at that speed for at least 6 hours under ORTHOS blend test...
Very nice again, I see you followed some of my suggestions hehe, nice dude!!
About temps; If your Cpu temps don't get any higher than 65º under heavy load (not ORTHOS or Prime loading) with real world programs or games, you can feel happy and safe.

After about an hour of DOD:S, it is only at 52C. Thats about as intensive as it gets with my system. I think I got the OC'ing thing down now

 

[NoStateofMind]

Originally posted by: cmdrdredd

Originally posted by: MADMAX23
My congratulations!....Hope you test your system at that speed for at least 6 hours under ORTHOS blend test...
Very nice again, I see you followed some of my suggestions hehe, nice dude!!
About temps; If your Cpu temps don't get any higher than 65º under heavy load (not ORTHOS or Prime loading) with real world programs or games, you can feel happy and safe.

Blend test is bad for 1 reason. It does not isolate the cause of instability.

For example. Run blend for 5 hours and get an error. Since you're testing your memory and CPU you never know where the error came from.

Test smallFFT for about 12hours by itself. Then run memtest for 12+hours by itself. That way you know whether your CPU can do the speed or whether your memory speed or timings are causing problems.

I wouldnt mind that except that many hours waiting would simply kill me

 

[cmdrdredd]

Originally posted by: PC Surgeon

Originally posted by: cmdrdredd

Originally posted by: MADMAX23
My congratulations!....Hope you test your system at that speed for at least 6 hours under ORTHOS blend test...
Very nice again, I see you followed some of my suggestions hehe, nice dude!!
About temps; If your Cpu temps don't get any higher than 65º under heavy load (not ORTHOS or Prime loading) with real world programs or games, you can feel happy and safe.

Blend test is bad for 1 reason. It does not isolate the cause of instability.

For example. Run blend for 5 hours and get an error. Since you're testing your memory and CPU you never know where the error came from.

Test smallFFT for about 12hours by itself. Then run memtest for 12+hours by itself. That way you know whether your CPU can do the speed or whether your memory speed or timings are causing problems.

I wouldnt mind that except that many hours waiting would simply kill me

Run it while you're sleeping and that's 8 hours. Run it when you're at work, that's 8 hours probably.

 

[Technonut]

Originally posted by: PC Surgeon

Originally posted by: cmdrdredd

Originally posted by: MADMAX23
My congratulations!....Hope you test your system at that speed for at least 6 hours under ORTHOS blend test...
Very nice again, I see you followed some of my suggestions hehe, nice dude!!
About temps; If your Cpu temps don't get any higher than 65º under heavy load (not ORTHOS or Prime loading) with real world programs or games, you can feel happy and safe.

Blend test is bad for 1 reason. It does not isolate the cause of instability.

For example. Run blend for 5 hours and get an error. Since you're testing your memory and CPU you never know where the error came from.

Test smallFFT for about 12hours by itself. Then run memtest for 12+hours by itself. That way you know whether your CPU can do the speed or whether your memory speed or timings are causing problems.

I wouldnt mind that except that many hours waiting would simply kill me

That is why you run Memtest86+ or Orthos overnight... I run Orthos overnight, then finish-up to the 24 hour mark (see 3.6GHz link in my sig) while doing other tasks...

 

[MADMAX23]

Originally posted by: cmdrdredd

Originally posted by: MADMAX23
My congratulations!....Hope you test your system at that speed for at least 6 hours under ORTHOS blend test...
Very nice again, I see you followed some of my suggestions hehe, nice dude!!
About temps; If your Cpu temps don't get any higher than 65º under heavy load (not ORTHOS or Prime loading) with real world programs or games, you can feel happy and safe.

Blend test is bad for 1 reason. It does not isolate the cause of instability.

For example. Run blend for 5 hours and get an error. Since you're testing your memory and CPU you never know where the error came from.

Test smallFFT for about 12hours by itself. Then run memtest for 12+hours by itself. That way you know whether your CPU can do the speed or whether your memory speed or timings are causing problems.

BLEND test is not bad at all, in fact it is the most reliable stress test that exists.
My method for overclocking:
1) Only use Bios, not any software like ASUS A.I. Tuning or ClockGen or SETFSB...
2) Decide the Max Vcore I'll ever to apply to my Cpu (C2D: 1.5v for ex.)
3) Let Ram memory out of equation; How? Running a lower divider (if possible, not with the P5B-D for ex.), increasing Ram voltage (2.2v-2.3v), relaxing all timings (5-6-6-18), testing 2 hours with Memtest 86 1.65 specially tests 5 and 6.
4) With that Max Vcore (1.5v for C2D), find my Max Cpu speed. Once Ram memory has succesfully passed 2 hours of Memtest, then I start overclocking my Cpu, raising the FSB and so...and I test its stability running the ORTHOS CPU stress test for at least 6 hours to consider a certain speed (in Mhz) stable.
5) Once I get a reasonable CPU Oc and Voltage ratio....and STABLE (12 hours priming under Cpu stress test) I begin with Ram memory, tightening its timings etc...you know..then testing with Memtest again until you get what you want.
6) TESTING with ORTHOS your CPU and Ram....using BLEND TEST.

This is a short description of what Oc the right way implies....
Cheers!!!!

 

[NoStateofMind]

Originally posted by: MADMAX23

Originally posted by: cmdrdredd

Originally posted by: MADMAX23
My congratulations!....Hope you test your system at that speed for at least 6 hours under ORTHOS blend test...
Very nice again, I see you followed some of my suggestions hehe, nice dude!!
About temps; If your Cpu temps don't get any higher than 65º under heavy load (not ORTHOS or Prime loading) with real world programs or games, you can feel happy and safe.

Blend test is bad for 1 reason. It does not isolate the cause of instability.

For example. Run blend for 5 hours and get an error. Since you're testing your memory and CPU you never know where the error came from.

Test smallFFT for about 12hours by itself. Then run memtest for 12+hours by itself. That way you know whether your CPU can do the speed or whether your memory speed or timings are causing problems.

BLEND test is not bad at all, in fact it is the most reliable stress test that exists.
My method for overclocking:
1) Only use Bios, not any software like ASUS A.I. Tuning or ClockGen or SETFSB...
2) Decide the Max Vcore I'll ever to apply to my Cpu (C2D: 1.5v for ex.)
3) Let Ram memory out of equation; How? Running a lower divider (if possible, not with the P5B-D for ex.), increasing Ram voltage (2.2v-2.3v), relaxing all timings (5-6-6-18), testing 2 hours with Memtest 86 1.65 specially tests 5 and 6.
4) With that Max Vcore (1.5v for C2D), find my Max Cpu speed. Once Ram memory has succesfully passed 2 hours of Memtest, then I start overclocking my Cpu, raising the FSB and so...and I test its stability running the ORTHOS CPU stress test for at least 6 hours to consider a certain speed (in Mhz) stable.
5) Once I get a reasonable CPU Oc and Voltage ratio....and STABLE (12 hours priming under Cpu stress test) I begin with Ram memory, tightening its timings etc...you know..then testing with Memtest again until you get what you want.
6) TESTING with ORTHOS your CPU and Ram....using BLEND TEST.

This is a short description of what Oc the right way implies....
Cheers!!!!

Whew! I'm breathing hard just from reading it!

 

[cmdrdredd]

Originally posted by: MADMAX23

Originally posted by: cmdrdredd

Originally posted by: MADMAX23
My congratulations!....Hope you test your system at that speed for at least 6 hours under ORTHOS blend test...
Very nice again, I see you followed some of my suggestions hehe, nice dude!!
About temps; If your Cpu temps don't get any higher than 65º under heavy load (not ORTHOS or Prime loading) with real world programs or games, you can feel happy and safe.

Blend test is bad for 1 reason. It does not isolate the cause of instability.

For example. Run blend for 5 hours and get an error. Since you're testing your memory and CPU you never know where the error came from.

Test smallFFT for about 12hours by itself. Then run memtest for 12+hours by itself. That way you know whether your CPU can do the speed or whether your memory speed or timings are causing problems.

BLEND test is not bad at all, in fact it is the most reliable stress test that exists.
My method for overclocking:
1) Only use Bios, not any software like ASUS A.I. Tuning or ClockGen or SETFSB...
2) Decide the Max Vcore I'll ever to apply to my Cpu (C2D: 1.5v for ex.)
3) Let Ram memory out of equation; How? Running a lower divider (if possible, not with the P5B-D for ex.), increasing Ram voltage (2.2v-2.3v), relaxing all timings (5-6-6-18), testing 2 hours with Memtest 86 1.65 specially tests 5 and 6.
4) With that Max Vcore (1.5v for C2D), find my Max Cpu speed. Once Ram memory has succesfully passed 2 hours of Memtest, then I start overclocking my Cpu, raising the FSB and so...and I test its stability running the ORTHOS CPU stress test for at least 6 hours to consider a certain speed (in Mhz) stable.
5) Once I get a reasonable CPU Oc and Voltage ratio....and STABLE (12 hours priming under Cpu stress test) I begin with Ram memory, tightening its timings etc...you know..then testing with Memtest again until you get what you want.
6) TESTING with ORTHOS your CPU and Ram....using BLEND TEST.

This is a short description of what Oc the right way implies....
Cheers!!!!

You want to ALWAYS isolate the fastest stable clocks first. Blend never allows you to isolate the CPU from the memory.

 

[MADMAX23]

cmdrdredd: I think you haven't read my short description of my testing methodology!