Stress Test A Gaming System

[DasFox]

I hear people saying to stress test a system properly you should run Prime95 and memtest for 24 hours.

Now on an overclocked system I can see some benefit here, but as long as you're not doing an OC, is 24 hours really needed just to test out a stock system and make sure all the parts are working ok?

I read someone say at 20 hours, and 22 hours things can surface on the testing, that might not otherwise with less time.

Well I'm at 19 hours on Prime95 and no errors.

ALOHA

 

[Duvie]

If you are not ocing I wouldn't really run it at all...loot at temps and if they are in line with others just start playing....

If you oc'd the system whether it be the memory, FSB, or cpu clock speed from stock then I would likely run it for 12-24 hours for gaming system. If it was DC client stuff then I would run for 48hrs...

 

[DasFox]

loot at temps, huh?

DC Clients?

THANKS

 

[Leros]

20 hours is fine. I've had prime fail at 8 hours before. But thats the latest.

 

[Duvie]

I mean just LOOK at your temps...either from software provided from the motherboard manufacturer or 3rd part stuff like speedfan...If your temps are in line with others I dont see an issue...

DC client stuff is "distributed computing"...stuff like SETI, Folding@Home, etc....number crunching

 

[DasFox]

Originally posted by: Leros
20 hours is fine. I've had prime fail at 8 hours before. But thats the latest.

Oh boy goodie one more hour to go and ----> Game Time

ALOHA

 

[DasFox]

Ahh do you think I should run memtest86+ too, or Prime95 is enough?

ALOHA

 

[Duvie]

If you are running all stock I find it wasteful to even run it...just start playing....The only think that would screw this up is 1) defective part 2) too high of temps often resulting from user error in installing thermal paste and thermal heatsink device....

 

[DasFox]

I don't have any Temp apps installed, all I ever look at is the Nvidia panel from time to time to look at the vid card temps is all. I keep my side case off my box with a fan blowing in, and I'm kinda old school I'll just keep an eye on the fans and touch the HSF, LOL, that's my sensor.

ALOHA

 

[DasFox]

Ok thanks, but what did you mean by, "temps are in line with others"?

ALOHA

 

[FP]

I run Prime95 and memtest from a boot disk. Both for 24 hours. I have had memory start failing after the 3 complete test for some reason.

It can't hurt and neither can not running it. You just may save yourself some time down the road.

 

[DasFox]

What changes in the way of testing?

I see Prime95 saying;

Self-test 96k passed
Self-test 1280k passed

All these Ks are going up, what are those?

ALOHA

 

[Leros]

I'm not sure why Duvie is making a big deal about temps. Sure, you should check your temps to make sure they are reasonable, but it doesn't really pertain to stress testing.

What you are trying to do with your stress test is make sure that your CPU is making accurate calculations. 12+ hours of Prime 95 will do this.

You also want to test your RAM. Running Memtest for 12+ hours should do a good job with that. Run it while you are sleeping or something.

Oh, and go ahead and play your game. A game is an excellent stress test. If it crashes, you know something is wrong!

 

[Duvie]

Originally posted by: Leros
I'm not sure why Duvie is making a big deal about temps. Sure, you should check your temps to make sure they are reasonable, but it doesn't really pertain to stress testing.

What you are trying to do with your stress test is make sure that your CPU is making accurate calculations. 12+ hours of Prime 95 will do this.

You also want to test your RAM. Running Memtest for 12+ hours should do a good job with that. Run it while you are sleeping or something.

Oh, and go ahead and play your game. A game is an excellent stress test. If it crashes, you know something is wrong!

BECAUSE!!!!!

THERE IS NO FVCKIN POINT TO STRESS TEST A STOCK SYSTEM IN MY OPINION. Since all components are being ran stock the only real concern to the system is likely temperature. Since they wont be overclocked and overvolted the only enemy to the component will likely be overheating resulting from user error in installing HSF's or poor case cooling and ventilation.

 

[DasFox]

Ok

I did run memtest86 before, but I forgot for how long, maybe I'll let it run just for the heck while I'm sleeping.

THANKS

 

[Leros]

Originally posted by: Duvie

Originally posted by: Leros
I'm not sure why Duvie is making a big deal about temps. Sure, you should check your temps to make sure they are reasonable, but it doesn't really pertain to stress testing.

What you are trying to do with your stress test is make sure that your CPU is making accurate calculations. 12+ hours of Prime 95 will do this.

You also want to test your RAM. Running Memtest for 12+ hours should do a good job with that. Run it while you are sleeping or something.

Oh, and go ahead and play your game. A game is an excellent stress test. If it crashes, you know something is wrong!

BECAUSE!!!!!

THERE IS NO FVCKIN POINT TO STRESS TEST A STOCK SYSTEM IN MY OPINION. Since all components are being ran stock the only real concern to the system is likely temperature. Since they wont be overclocked and overvolted the only enemy to the component will likely be overheating resulting from user error in installing HSF's or poor case cooling and ventilation.

Good point, I didn't realize that he wasn't overclocking.

 

[DasFox]

Well Prime ran for 20 hours, no errors

Game Time

ALOHA

 

[FP]

Originally posted by: Duvie

Originally posted by: Leros
I'm not sure why Duvie is making a big deal about temps. Sure, you should check your temps to make sure they are reasonable, but it doesn't really pertain to stress testing.

What you are trying to do with your stress test is make sure that your CPU is making accurate calculations. 12+ hours of Prime 95 will do this.

You also want to test your RAM. Running Memtest for 12+ hours should do a good job with that. Run it while you are sleeping or something.

Oh, and go ahead and play your game. A game is an excellent stress test. If it crashes, you know something is wrong!

BECAUSE!!!!!

THERE IS NO FVCKIN POINT TO STRESS TEST A STOCK SYSTEM IN MY OPINION. Since all components are being ran stock the only real concern to the system is likely temperature. Since they wont be overclocked and overvolted the only enemy to the component will likely be overheating resulting from user error in installing HSF's or poor case cooling and ventilation.

I disagree with this. I never overclocked my system and had to go through 3 sticks of dual channel memory before finding a pair that could pass a memtest test. Would my system have died a horrible death had I used the first few sticks? Probably not, but I would have most likely encountered some funny errors every now and then. And no it wasn't my mobo as I checked the same sticks in another machine with the same results.

Components can be faulty without being completely DOA.

 

[Operandi]

Originally posted by: binister

Originally posted by: Duvie

Originally posted by: Leros
I'm not sure why Duvie is making a big deal about temps. Sure, you should check your temps to make sure they are reasonable, but it doesn't really pertain to stress testing.

What you are trying to do with your stress test is make sure that your CPU is making accurate calculations. 12+ hours of Prime 95 will do this.

You also want to test your RAM. Running Memtest for 12+ hours should do a good job with that. Run it while you are sleeping or something.

Oh, and go ahead and play your game. A game is an excellent stress test. If it crashes, you know something is wrong!

BECAUSE!!!!!

THERE IS NO FVCKIN POINT TO STRESS TEST A STOCK SYSTEM IN MY OPINION. Since all components are being ran stock the only real concern to the system is likely temperature. Since they wont be overclocked and overvolted the only enemy to the component will likely be overheating resulting from user error in installing HSF's or poor case cooling and ventilation.

I disagree with this. I never overclocked my system and had to go through 3 sticks of dual channel memory before finding a pair that could pass a memtest test. Would my system have died a horrible death had I used the first few sticks? Probably not, but I would have most likely encountered some funny errors every now and then. And no it was my mobo as I checked the same sticks in another machine with the same results.

Components can be faulty without being completely DOA.

I agree 100%. Every machine I build gets at least 12 hours of prime95. If it's a machine that will see gaming it will also get a UT2k3 (nearly) infinite bot match while simultaneously running Prime95.

 

[bob4432]

Originally posted by: Operandi

Originally posted by: binister

Originally posted by: Duvie

Originally posted by: Leros
I'm not sure why Duvie is making a big deal about temps. Sure, you should check your temps to make sure they are reasonable, but it doesn't really pertain to stress testing.

What you are trying to do with your stress test is make sure that your CPU is making accurate calculations. 12+ hours of Prime 95 will do this.

You also want to test your RAM. Running Memtest for 12+ hours should do a good job with that. Run it while you are sleeping or something.

Oh, and go ahead and play your game. A game is an excellent stress test. If it crashes, you know something is wrong!

BECAUSE!!!!!

THERE IS NO FVCKIN POINT TO STRESS TEST A STOCK SYSTEM IN MY OPINION. Since all components are being ran stock the only real concern to the system is likely temperature. Since they wont be overclocked and overvolted the only enemy to the component will likely be overheating resulting from user error in installing HSF's or poor case cooling and ventilation.

I disagree with this. I never overclocked my system and had to go through 3 sticks of dual channel memory before finding a pair that could pass a memtest test. Would my system have died a horrible death had I used the first few sticks? Probably not, but I would have most likely encountered some funny errors every now and then. And no it was my mobo as I checked the same sticks in another machine with the same results.

Components can be faulty without being completely DOA.

I agree 100%. Every machine I build gets at least 12 hours of prime95. If it's a machine that will see gaming it will also get a UT2k3 (nearly) infinite bot match while simultaneously running Prime95.

regardless of oc or not always stress test. it is much easier to get a rma on a part you have had for a week instead of 6mos in the terms of how much you have installed and tweaked your rig.

 

[DasFox]

Well I did 20 hours, and the more I think about it OC or not hardware does fail, so it's a good idea. I just wasn't sure how long I should run it for on stock hardware was all.

ALOHA