Best (simplest) program to stress test new OC?

Mar 1, 2000
26,741
129
126
#1
I have a Phenom II X2 555 on delivery and plan on unlocking and OCing.
I'm not looking to squeeze every last ounce out of it, but I would like to get as decent of an OC out of it that I can.

The last OC that I did was an Opteron 165 @2.4ghz and I just tested with Prime95 and that has seemed to suffice for the past 4 years or so.

Is Prime still the defacto? Or should I be stressing the CPU with something else?

Thanks in advance
 

evolucion8

Platinum Member
Jun 17, 2005
2,872
0
61
#2
Try to download OCCT, it has Linpack built in and it will stress your processor better than any Prime 95. Prime 95 is good to check if there's any errors when overclocking the memory and th HT link/FSB etc.
 

jvroig

Platinum Member
Nov 4, 2009
2,399
0
76
#3
Try to download OCCT, it has Linpack built in and it will stress your processor better than any Prime 95.
I like OCCT (classic), but not OCCT linpack. Some have pointed out that its Linpack implementation is flawed, or at least not quite as effective as those of LinX.

I can't say if it is really flawed, but In my experience, compared to Prime95 Small FFT, it heated up the CPU (Phenom II X4 965) two degrees less (reference: EasyTune temp monitor for both P95 and OCCT Linx tests). It seems to heat up the CPU more if you go by the core temp sensors in OCCT (it shows the per core temp, not the CPU temp), but the CPU temp as measured by EasyTune is still consistently lower than Prime95. And another thing that bothered me was the few second breaks every now and then, as if OCCT Linpack was giving the CPU a break every few minutes.

Is Prime still the defacto? Or should I be stressing the CPU with something else?
Still good, but there are others out there like OCCT and Linx.
Prime 95 Small FFT = CPU stability
Prime 95 Large FFT = IMC / CPU-NB stability
Prime 95 Blend = RAM stability (not a very good one, Memtest HCI would be better)
 

flexy

Diamond Member
Sep 28, 2001
8,466
0
106
#4
I would also say OCCT as a quick "all-in-one" test.

HOWEVER - linpack is one crazy extreme stress test..be warned. I personally dont use linpack but the normal OCCT test for an hour or so. Linpack you can use if you want to use your CPU to cook some meals..honestly..it scares me. It's like the "laugh in the face" for every overclocker and CAN produce some temps +20c more than what you ever see with prime/standard OCCT.

I also agree with memtest HCI...it is very good. OCCT/memtest HCI is all you need if you overclock.
 
Last edited:
May 22, 2007
11,135
0
86
#5
that's because it's testing it using real world conditions. just stressing it for an hour or even 24 for that matter doesn't paint the whole picture. that's why you have people who claim 4ghz "prime" stable but can't play games or actually do anything unless it's at 3.8.

linx is probably best for pure heat testing, occt is probably best for actual real world usage testing. both are good, but in your case it sounds like occt is the one to get.
 

flexy

Diamond Member
Sep 28, 2001
8,466
0
106
#7
that's because it's testing it using real world conditions. just stressing it for an hour or even 24 for that matter doesn't paint the whole picture. that's why you have people who claim 4ghz "prime" stable but can't play games or actually do anything unless it's at 3.8.
The reason is probably that "prime stable" does not really take anything into account, like NB testing, memory etc..etc.. unless you do several runs with different values.

As far as i know..the latest OCCT use prime95 code anyway?
 

evolucion8

Platinum Member
Jun 17, 2005
2,872
0
61
#8
I like OCCT (classic), but not OCCT linpack. Some have pointed out that its Linpack implementation is flawed, or at least not quite as effective as those of LinX.

I can't say if it is really flawed, but In my experience, compared to Prime95 Small FFT, it heated up the CPU (Phenom II X4 965) two degrees less (reference: EasyTune temp monitor for both P95 and OCCT Linx tests). It seems to heat up the CPU more if you go by the core temp sensors in OCCT (it shows the per core temp, not the CPU temp), but the CPU temp as measured by EasyTune is still consistently lower than Prime95. And another thing that bothered me was the few second breaks every now and then, as if OCCT Linpack was giving the CPU a break every few minutes.
In my case, OCCT Linpack warmed my CPU more than Prime95, I wonder if your experience its only related to Phenom's II? I think that the little breaks every now and then is when is compiling the results for redundancy and error detection.
 
Mar 1, 2000
26,741
129
126
#9
heh. I thought this would be an easy answer ;)
 

jvroig

Platinum Member
Nov 4, 2009
2,399
0
76
#10
In my case, OCCT Linpack warmed my CPU more than Prime95, I wonder if your experience its only related to Phenom's II?
It could be, I've really only compared it on my Phenom II.

that's why you have people who claim 4ghz "prime" stable but can't play games or actually do anything unless it's at 3.8.
That most certainly is possible. Another common thing is the fact that Prime 95 has three different tests which test three different things, and sometimes people claim "Prime" stability when they only used one test which may not even be the right test for them. I fell into that mistake myself (which is how I learned to do better), using only Small FFT at first, and being stable for hours. After declaring victory and being "prime stable", I saw I wasn't in real life. I was stumped until by accident I chose Large FFT, and saw that it produced an error in less than 30 mins. My IMC (clocked it to 2.6) was not stable, and needed a little bump in CPU-NB Voltage.

Also (and again just anecdotal, and there's really little solid science to go about), there have been lots of members here who have mentioned being Prime or OCCT or LinX stable, only to find out they fail another different test, or maybe a game. Hence, the recommendation is generally to use more than just a single test (including a game or two maybe), especially if you are really pushing your CPU to the limit.

Personally, mine is:
OCCT
Prime 95 Small FFT
Prime 95 Large FFT
Memtest HCI
 

badboyeee

Senior member
Nov 12, 2001
664
0
0
#11
Hey i remember you from my thread, im in the same boat as you but dont plan to get my cpu/mobo until couple months.

But in the meantime, ive been trying to overclock my Opteron 165 to its limit. I'm doing this for fun & practice before I get my new system. I too had my Opteron set to 2.4ghz for these past 3-4 years. I'm up to 2.61ghz right now but my first core is having errors. Gonna try to increase the voltage now..

I've been using what everyone else has been using here.. Prime 95 is very simple (I use small FFT then large FFT) and then OCCT last. OCCT is good if your into graphs hehe. I also use Memtest when im pushing my memory speed to its limit, but i changed the divider to a lower speed already to get to 2.6ghz and beyond.

Don't forget to keep an eye on your CPU temp. But you didnt ask for software for that so Im guessing your good. Main one I use is CoreTemp. OCCT also keeps track of the temp too.

Let me know if you get to unlock your cores and how much you get to overclock, I'll be looking forward to your results. What motherboard did you get??
 
Jun 30, 2010
1,140
0
0
#12
usually I fire up one of my home studio programs and do some recording. If it is flaky it shows up mighty fast. Hence why my machines are very lightly overclocked.
 

badboyeee

Senior member
Nov 12, 2001
664
0
0
#13
That most certainly is possible. Another common thing is the fact that Prime 95 has three different tests which test three different things, and sometimes people claim "Prime" stability when they only used one test which may not even be the right test for them. I fell into that mistake myself (which is how I learned to do better), using only Small FFT at first, and being stable for hours. After declaring victory and being "prime stable", I saw I wasn't in real life. I was stumped until by accident I chose Large FFT, and saw that it produced an error in less than 30 mins. My IMC (clocked it to 2.6) was not stable, and needed a little bump in CPU-NB Voltage.

Also (and again just anecdotal, and there's really little solid science to go about), there have been lots of members here who have mentioned being Prime or OCCT or LinX stable, only to find out they fail another different test, or maybe a game. Hence, the recommendation is generally to use more than just a single test (including a game or two maybe), especially if you are really pushing your CPU to the limit.
That happened to me too, I was only using Small FFT and was stable. Later found out Large FFT uses some memory and I wasnt stable. It was because my RAM was being overclocked too much.
 
Mar 1, 2000
26,741
129
126
#14
Hey i remember you from my thread, im in the same boat as you but dont plan to get my cpu/mobo until couple months.

But in the meantime, ive been trying to overclock my Opteron 165 to its limit. I'm doing this for fun & practice before I get my new system. I too had my Opteron set to 2.4ghz for these past 3-4 years. I'm up to 2.61ghz right now but my first core is having errors. Gonna try to increase the voltage now..

I've been using what everyone else has been using here.. Prime 95 is very simple (I use small FFT then large FFT) and then OCCT last. OCCT is good if your into graphs hehe. I also use Memtest when im pushing my memory speed to its limit, but i changed the divider to a lower speed already to get to 2.6ghz and beyond.

Don't forget to keep an eye on your CPU temp. But you didnt ask for software for that so Im guessing your good. Main one I use is CoreTemp. OCCT also keeps track of the temp too.

Let me know if you get to unlock your cores and how much you get to overclock, I'll be looking forward to your results. What motherboard did you get??
ha! yeah that was/is me.
I'm buying my stuff used, so the cores have been unlocked/tested already with the combo I picked up. I'm assuming/hoping to get the same results.

I don't plan on trying to squeeze every drop out of the new CPU as the temps in my house vary so much (we have AC, but I'm not a huge fan so it only gets turned on when its sweltering). I actually had my Opteron up to 2.6 (IIRC) when I first got it on was more into tweaking, but again, come summer it would flake out, so it was just easier to leave at 2.4Ghz year round.

This is the MB in question:
http://www.newegg.com/Product/Produc...82E16813130237
 

evolucion8

Platinum Member
Jun 17, 2005
2,872
0
61
#15
That means that small FFT stresses more the CPU and large FFT stresses more the RAM making it prone to find memory related errors easier.
 

Kenmitch

Diamond Member
Oct 10, 1999
7,121
43
126
#16
OCCT works good for testing....But I like to start with Intel Burn Test to find voltages required alot faster as it will error out very quickly. Does get toasty hot tho so be warned :)
 

Juddog

Diamond Member
Dec 11, 2006
7,851
0
71
#17
I usually go the full gamut of testing when overclocking. Currently I'm stable at 4.2 Ghz and was goofing around with LinX last night, and found that LinX failed at the 9th run so I had to boost the core voltage some more. After boosting core to 1.34 from 1.325 I got it stable at 20 passes.

My overclocking steps:
1) Prime95 on all cores. Good starter to check stability.
2) OCCT. Typically I run it first for an hour, then do the LinX below, if it passes, then I run it 24 hours afterwards.
3) LinX. I usually make sure the setting is totally stable at 20 passes. I noticed that LinX stresses not only your CPU but your memory too; on my system it almost maxed out all 6 GB of RAM (left about .2 GB I think).

If you can do 20 passes of LinX plus 24 hours OCCT, you're golden - the CPU is stable. I also like to check task manager for hard faults in memory as well.
 

Similar threads



ASK THE COMMUNITY

TRENDING THREADS