Prime 95 Yes or No

[runzwithsizorz]

In the process of a new build. i5 3570k. Is it important to run Prime 95 to test for stability or not test at all. If it's important, why??? Please reply with reason. Planning on mild overclock. Thanks very much.

(Runz and I are disagreeing on this point)

Wife of Runz (I think it should be done)!

 

[TuxDave]

If you're overclocking, I would do it. Checking it to prevent crashing is one thing. Checking it to make sure silent data corruption doesn't happen is another (not sure if Prime 95 checks this, I havne't overclocked in over a decade! lol)

 

[nOOky]

I have never used it to test stability. I overclock, it either works or crashes. I think it is more useful for trying to troubleshoot a problem or achieve the maximum overclock you can than just mildly overclocking and letting 'er rip.
Just my 2 cents.

Edit: I have used it in the past, but I was able to get roughly the same results just using trial and error.

 

[felang]

I recommend OCCT or IBT for finding a stable overclock

 

[zootedd]

I know a lot of people say that it is good to "burn in" a new build using p95 or similar software. I don't know the validity of this idea.

 

[felang]

IMO burn in is not needed, you do need to check for stability, especially if overclocking.

 

[VirtualLarry]

Checking for stability of a non-overclocked system, shouldn't be needed, in theory, but if you are going to be performing some important financial calculations on it, then yes, you should always do some sort of stress test.

If you are overclocking, then yes, it is essential to stress-test. Prime95 is one, but to be thorough, you need to run it overnight. Another good one is OCCT:Linpack 64-bit. I find that to be more stressful that Prime95. Linx is another good one. Linx is linpack, just like OCCT:linpack. I'm not sure if they differ much.

 

[Idontcare]

When I built my 2600k system it turned out I had a set of bad sticks of ram. I did not realize they were bad because they passed memtest86+. But I realized something was amiss when LinX would not pass even though all the hardware was at stock.

Turned out LinX was catching the ram error, which I later isolated and verified with HCI Memtest.

So I would argue that even if you are not OC'ing it is a good idea to use stress testers just to confirm your hardware is in good working order.

 

[MadScientist]

Run LinX. I have ram speced at 1600 mhz @1.65V from an old build and was trying to run it at 1.50V. It passed Memtest 86+ and HCl Memtest at 1.50V but LinX failed after 3 runs. Raised the vdimm to 1.575V and it passes 50 LinX runs at max ram.

 

[SiliconWars]

How long would you run LinX for? I've been having a few issues with Prime95 not reporting erroneous memory and have been using the windows 7 memory diagnostic (which actually appears to be quite good in most cases).

 

[Idontcare]

How long would you run LinX for? I've been having a few issues with Prime95 not reporting erroneous memory and have been using the windows 7 memory diagnostic (which actually appears to be quite good in most cases).

Minimum of 5 cycles, but be sure and select to use "all available memory".

 

[SiliconWars]

Yeah I checked the all memory part. Seems to take about 20 mins or so for two cycles on my own PC.

Edit - i must have a really bad idea of the passage of time, it was actually 6 mins per cycle

 

[SiliconWars]

I guess I can run this for an hour on customer PC's, looks pretty good. I have some bad memory that I can check with it for sure (stuff that passed Prime95) so I'll try that out as well.

Do you reckon Prime 95 is still ok for cpu testing? CPU's are almost always fine though, I've never had a problem with a cpu in all the PC's I've sold.

 

[Idontcare]

I guess I can run this for an hour on customer PC's, looks pretty good. I have some bad memory that I can check with it for sure (stuff that passed Prime95) so I'll try that out as well.

Do you reckon Prime 95 is still ok for cpu testing? CPU's are almost always fine though, I've never had a problem with a cpu in all the PC's I've sold.

Prime is fine for CPU. In fact, for AMD it appears to be superior over LinX when it comes to stability checking.

 

[grimpr]

I use Prime95 and Y-Cruncher in AMD rigs, Prime95 for core and temp checking and y-crunchers stress test mode for rigorous core and nb/memory controller testing. Y-Cruncher finds faults much faster, tells you about them and doesnt bsod or crash the system like prime95 does due to its coding.

 

[TJCS]

The answer is absolutely yes if you care about stability. I have had LinX pass and Prime95 fail, or Prime95 pass and HandBrake fail. The point is prime95 shouldn't be treated as the ultimate answer to stability, but a value tool in your stability testing toolbox.

 

[VirtualLarry]

Y-Cruncher finds faults much faster, tells you about them and doesnt bsod or crash the system like prime95 does due to its coding.

Prime95 does not "crash the system due to its coding". On a stable rig, it will run just fine forever. If your rig is crashing due to Prime95, that's due to the overclock.

 

[Vectronic]

I think he meant Y-Cruncher catches the errors before it BSODs, rather than P95 being poorly coded.

P95 seems to catch errors when you are "close" to stable... if you're pretty far off... BSOD.

I wouldn't expect Y-Cruncher to be any different, if it is... it doesn't seem very useful.