My Overclocking Stress Test

AdamK47

Lifer
Oct 9, 1999
15,207
2,838
126
Here is what I'm running.

What do you think? So far everything has been stable. The smoke demo has always had the ability to heat up a GeForce card tremendously. I added a profile for nVidia smoke in the control panel and set it to use split frame rendering SLI. I also set it up to show the SLI Visual Indicator. Setting the affinity to CPU-0 stopped the skipping of frames and allowed it to run smoothly. MemTest uses 3GB and LinX uses 1GB. Perhaps I should reverse that. Prime95 is running 4 threads of in-place large FFTs.
 

LOUISSSSS

Diamond Member
Dec 5, 2005
8,771
54
91
why are you doing all those tests simultaneously? each cpu intenstive program is stealing cpu cycles from the other, which wastes time in testing. what you're doing is prob only useful if you're trying to test your PSU to see if it can handle your rig with all components 100% stressed.

if you're not testing your psu's load ability, u should run each test at a time..
 

AdamK47

Lifer
Oct 9, 1999
15,207
2,838
126
It creates maximum heat inside the case and stresses the entire system. It's a worst case scenario.
 

dguy6789

Diamond Member
Dec 9, 2002
8,558
3
76
Have your optical drives reading and your hard drives crunching too! Also transfer a file over your network.
 

LOUISSSSS

Diamond Member
Dec 5, 2005
8,771
54
91
you're aware that other than when "stress testing." your system will NEVER reach that amount of stress right?
you should test each individual component until you're satisfied then move onto the next when each is stable.

what u gunna do when one of those things freezes your computer or gives u a bsod or restart? how u gunna know which program or component caused it?

for example, when you run p95 along with linx, they're going to be stealing cpu cycles from each other, so each will only get 50% of the cpu, hence you will have to test for twice as long just to get the same results. but in you're case you're doing it with the nvidia shadow crap, memtest, linx, p95: all cpu intensive programs that will steal cycles from the cpu so you'll have to test 4x as long (assuming u have a quad) if you have a dual then even longer.

but then again its your computer do whatever u want.
 

AdamK47

Lifer
Oct 9, 1999
15,207
2,838
126
Originally posted by: LOUISSSSS
you're aware that other than when "stress testing." your system will NEVER reach that amount of stress right?
you should test each individual component until you're satisfied then move onto the next when each is stable.

what u gunna do when one of those things freezes your computer or gives u a bsod or restart? how u gunna know which program or component caused it?

for example, when you run p95 along with linx, they're going to be stealing cpu cycles from each other, so each will only get 50% of the cpu, hence you will have to test for twice as long just to get the same results. but in you're case you're doing it with the nvidia shadow crap, memtest, linx, p95: all cpu intensive programs that will steal cycles from the cpu so you'll have to test 4x as long (assuming u have a quad) if you have a dual then even longer.

but then again its your computer do whatever u want.

I'm already past that point chief. I've already tested each individual overclock independently. I appreciate your concern. It's a Core i7 system with 8 logical CPUs. A quad core with hyperthreading. Did you read the rig profile. It's in the sig ya know?
 

ajaidevsingh

Senior member
Mar 7, 2008
563
0
0
AdamK47 what cooling solution are you using also did you have any stability issues ??

Also whats your max temp for this baby?? We tried OC our 920 to 3.7Ghz but the thing went to 70-80C in benches and then a BSOD so waiting for a better native heat sink to come along.
 

ajaidevsingh

Senior member
Mar 7, 2008
563
0
0
Also i realized the best way for heating up the case was to run 8 instances of ATI tool with each instance prescribed its own thread from the Task manager manually. But i guess u do need a serious ass GPU like the 2x 4870x2 crossfire he has.

Hell i will try 8 instances of ATI tool on my Q6600 with 4850 CF..... Have not stressed it much now a days!
 

LOUISSSSS

Diamond Member
Dec 5, 2005
8,771
54
91
Originally posted by: AdamK47
Originally posted by: LOUISSSSS
you're aware that other than when "stress testing." your system will NEVER reach that amount of stress right?
you should test each individual component until you're satisfied then move onto the next when each is stable.

what u gunna do when one of those things freezes your computer or gives u a bsod or restart? how u gunna know which program or component caused it?

for example, when you run p95 along with linx, they're going to be stealing cpu cycles from each other, so each will only get 50% of the cpu, hence you will have to test for twice as long just to get the same results. but in you're case you're doing it with the nvidia shadow crap, memtest, linx, p95: all cpu intensive programs that will steal cycles from the cpu so you'll have to test 4x as long (assuming u have a quad) if you have a dual then even longer.

but then again its your computer do whatever u want.

I'm already past that point chief. I've already tested each individual overclock independently. I appreciate your concern. It's a Core i7 system with 8 logical CPUs. A quad core with hyperthreading. Did you read the rig profile. It's in the sig ya know?

i didn't ask you what you had in your system. what i'm saying applies to all computers. your PC will probably never be loaded 100% on each component simultaneously.
 

AdamK47

Lifer
Oct 9, 1999
15,207
2,838
126
Originally posted by: LOUISSSSS
Originally posted by: AdamK47
Originally posted by: LOUISSSSS
you're aware that other than when "stress testing." your system will NEVER reach that amount of stress right?
you should test each individual component until you're satisfied then move onto the next when each is stable.

what u gunna do when one of those things freezes your computer or gives u a bsod or restart? how u gunna know which program or component caused it?

for example, when you run p95 along with linx, they're going to be stealing cpu cycles from each other, so each will only get 50% of the cpu, hence you will have to test for twice as long just to get the same results. but in you're case you're doing it with the nvidia shadow crap, memtest, linx, p95: all cpu intensive programs that will steal cycles from the cpu so you'll have to test 4x as long (assuming u have a quad) if you have a dual then even longer.

but then again its your computer do whatever u want.

I'm already past that point chief. I've already tested each individual overclock independently. I appreciate your concern. It's a Core i7 system with 8 logical CPUs. A quad core with hyperthreading. Did you read the rig profile. It's in the sig ya know?

i didn't ask you what you had in your system. what i'm saying applies to all computers. your PC will probably never be loaded 100% on each component simultaneously.

I know your the expert, but I also know what works for me. Have you tried running LinX on a Core i7. It won't use 100% of the CPU.

Edit - Lou, we've been down this road many times before on this forum even well before you registered. I refuse to use a PC that may or may not fail depending on the load on the system. What makes you think I don't already know that there is probably no application that will load a system up like this?
 

AdamK47

Lifer
Oct 9, 1999
15,207
2,838
126
Originally posted by: ajaidevsingh
AdamK47 what cooling solution are you using also did you have any stability issues ??

Also whats your max temp for this baby?? We tried OC our 920 to 3.7Ghz but the thing went to 70-80C in benches and then a BSOD so waiting for a better native heat sink to come along.

With 8 large FFT threads of Prime95 alone I saw the temps get as high as 72 degrees C according to SpeedFan which is probably close to 80 with CoreTemp. I'm using a Cooler Master V8. There was no way I could run at 3.8GHz with 1.35V using the stock cooler I had before. There aren't any stability issues at 3.8GHz. This was the last test I was going to throw at it. It passed flawlessly.
 

LOUISSSSS

Diamond Member
Dec 5, 2005
8,771
54
91
I refuse to use a PC that may or may not fail depending on the load on the system.

i'm not sure what kind of tricky language you're trying to use.. but why wont u use a system that "may not fail" OR "may fail." to my understanding this is inclusive of all statements. hmm.. clarify?

unless you're talking about a system that simply "will never fail."
 

cusideabelincoln

Diamond Member
Aug 3, 2008
3,268
11
81
LOUISSSSS is right. If you're running these tests simulataneously then you're not actually stressing the CPU as much as it can be, since there will be some downtime as the scheduler switches from one task to the next. But you did say you've already tested each component individually (and I'm assuming this means you only ran one program, like Prime 95, at a time), then I don't see any harm in this test. If you want to get maximum heat, then you'll just have to test for a longer period of time.