Prime95 versus Everest for stress testing

Nov 26, 2005
15,194
403
126
Prime 95 can run for hours and not find anything. I would then boot up Unreal Tournament III and have a crash. When I would be LinX stable, UT3 would not crash.
 

EarthwormJim

Diamond Member
Oct 15, 2003
3,239
0
76
Everest is not very demanding compared to Prime or linpack based tests (Linx).


LinX will usually find an unstable CPU within the first few passes or so. Prime95 can take hours or not even find an unstable CPU at all.
 

faxon

Platinum Member
May 23, 2008
2,109
1
81
Generally use OCCT. It's a modified prime algorithm that finds errors many times faster than p95. It also has realtemp and LinX built in lol
 
Last edited:
Nov 26, 2005
15,194
403
126
I think LinX is also a very good RAM tester. I'm not totally sure about how it works, I think Ruby or N7 know more about it but id go out on a limb and say LinX > all

Generally you want to run more than 25 passes at the max RAM you can use.
 

EarthwormJim

Diamond Member
Oct 15, 2003
3,239
0
76
I think LinX is also a very good RAM tester. I'm not totally sure about how it works, I think Ruby or N7 know more about it but id go out on a limb and say LinX > all

Generally you want to run more than 25 passes at the max RAM you can use.

It's a very good memory subsystem tester. It's filling up your RAM, so if your memory is incorrectly storing or retrieving data, LinX will throw an error.

It's not as good as HCI's memtest for testing RAM though.

I like 100 passes, which at 7000mb RAM usage can take almost all night. It's rather arbitrary which number of passes you choose to prefer.

Generally use OCCT. It's a modified prime algorithm that finds errors many times faster than p95. It also has realtemp and occt built in lol

While it does seem to find errors faster than Prime, it still does not compare to LinX.
 
Last edited:

Baasha

Golden Member
Jan 4, 2010
1,989
20
81
Does anyone have a link to the latest version of LinX? I downloaded Prime95 but after seeing what you guys are saying about that vs LinX, I think I'll go with LinX.
 

EarthwormJim

Diamond Member
Oct 15, 2003
3,239
0
76
Does anyone have a link to the latest version of LinX? I downloaded Prime95 but after seeing what you guys are saying about that vs LinX, I think I'll go with LinX.

Do both tests. LinX first though, since it's typically a better and faster indicator.

I have had an overclock pass 100 passes of LinX only to fail Prime 95 blend after a few hours.
 

BonzaiDuck

Lifer
Jun 30, 2004
16,743
2,097
126
I do it this way:

1) Preliminary testing with MEMTEST86+ -- to assure that a new set of modules are "OK" at stock settings.

2) Short, half-hour or hour-long tests with PRIME95 to locate the beginning of a stable range for settings.

3) At least 8 hours with PRIME95 small FFTs.

4) Finally, an hour or two with Linpack/Linx.
 

MagickMan

Diamond Member
Aug 11, 2008
7,460
3
76
I'm just the opposite.

1. 5 passes of Linpack to see if the OC bombs quickly
2. 2 hours of OCCT
3. 20 passes of Linpack
4. 12 hours of Prime95 "blend"

done.
 

faxon

Platinum Member
May 23, 2008
2,109
1
81
It's a very good memory subsystem tester. It's filling up your RAM, so if your memory is incorrectly storing or retrieving data, LinX will throw an error.

It's not as good as HCI's memtest for testing RAM though.

I like 100 passes, which at 7000mb RAM usage can take almost all night. It's rather arbitrary which number of passes you choose to prefer.



While it does seem to find errors faster than Prime, it still does not compare to LinX.
/bonk i didnt see i mistyped OCCT where i meant to put LinX. yea i would normally use LinX as well, but it's not safe to run on my system since my CPU goes past 80c and thermal reboots on me with current cooling. havent got around to installing my water loop yet, been to lazy LOL.
 

EarthwormJim

Diamond Member
Oct 15, 2003
3,239
0
76
So how does a LinX error look like?

windows-7-bsod.jpg


Or it just says there's an error and stops running the test.