u only got 8 thread, why would you choose 16 for? that prolly why your results are all over the place.
Printable View
Updated to v2.53!
- Linpack binaries updated (12-18) + patched for AMD CPUs
- Added "All" threads option to allow full utilization of SMT threads (e.g. HyperThreading units)
- Default times to run is now set to 10
Note: for Core i7 CPU owners, since v2.53, use the "All" threads option to fully utilize HyperThreading units.
My two cents but I would NOT use this or anything like it for testing.
These apps are designed to kill chips not test them in any real world way.
Again, I know most here will disagree but this is how I feel and I'm not alone on this.
Test with the apps that you will use.
Movieman: i mentioned this I think it was 4 years ago (?), when the Linpack thread created all the rage here, and I got rediculously flamed for it. Gamers should game, folders/crunchers should use prime, etc. Linpack's target audience are chip relibility testing, to test and bin silicon. Definitely NOT to run overclocked, overvolted chips for hours on end with Linpack.
I mean, movieman, it's gotten a bit absurd now what people are doing here. When Linpack/IBT first came out, everyone agreed that 5 runs was enough. Then when people wanted to find any little itty ounce of instability, it became 10 runs. Now people are running these CPU killing tools *OVERNIGHT*.
Stress testing is all about making sure your hardware will work long term, reliabilty, at *STOCK* settings, and weeding out defects. That's the entire point of stress testing. Once you start throwing overvolting and overclocking into the mix, you're already running out of specification, so you can't be surprised if your CPU's wind up degrading from it (funny how so many people claimed it was a myth, but now, there are many reports from people that had to raise their once stable voltages, even in the 2600k batch thread, and the other thread on degradation, which seems to have died). As well as on other forums.
I have no problems using these Apps but some idiots started to spread Chinese whispers that you need to run 100loops or XX hours which is totally stupidity!
I run the default 5 loops and would run say 15 at most ?(takes longer the more RAM you have if you set to use all it all!
The only software I know of that could kill hardware was that certain build of FurMark until Nvidia limited it in drivers (you could override that though) and I am sure the newer builds are a bit gentler!
I have stopped using it and use Heaven instead!
I just run 10 loops max memory
Thanks for the update.
That's 10 too many!:D
My personal feeling is if you want to do stability tests then upclock your machine 200-300mhz above what you will run it at and then test your apps
at that speed. If issues you'll find out and if not your fairly sure to be fine at your normal speeds and you don't risk killing your chips.
I highly doubt the peep who kindly makes this App (freely) wants you to be scaremongering (more than once) in his thread.
I also doubt anyone here cares what you say and will do their own thing.
Your probably scared to run it as your OC (if you do OC) is not stable. ;)
1)The "peep" as you describe them are the ones who explained exactly what the purpose of this app was and they cringed when I told them how it was being used on XS and elsewhere. Yes, I got my info from the Intel engineers who designed this app.
2) Now that is up to the individual. I'm just expressing my feelings.
3)Now thats where your wrong. I do OC on everything I own and mine are stable and not just for a few seconds or minutes but for years at 100% load 24/7
The defense rests.:D
Stable in your opinion and running PI no doubt? LOL!
I know who made it but AFAIK AgentGOD has worked on it for us (I could be wrong) and I highly doubt you have inside info at Intel but many claim such things!
Do not even see the point in you posting in this thread if you dislike this App TBH!
1) No, not PI my friend..I do DC apps where the machines run 100% load 24/7/365.25
2) HMM, how to answer this...Ok, don't take my word for it.
Ask around. I know a few people and one more point: I don't lie,ever.
3) Posted to say to people that running IBT is a mistake that can have very expensive consequences.
I have to agree with Dave on this one. IBT can do some real damage to your CPU and if they are overclocked high running this for more then one pass of the long run it truely not needed.
Run the Crunching app like Dave has said and if passes after a day you can't ask for more really.
I run many different stress tests on my rigs before I send them out. Some check out memory issues very well, other stress the caches and others this and that. I can run some Mathlab scripts that will not pass but IBT will. IBT can and will kill CPU's if run long enough, if not degrade the heck out of them for no good reason. Crap run Prime95 for a little bit and if that passes what more do you want.
The program was created to save time on CPU stability testing while at the same time being more accurate in doing so than the traditional ways (e.g. Prime95).
It was not meant to be run for an absurd number of loops (e.g. 20+, 50+, 100+) as some people here (and elsewhere) are doing.
Correct and if idiots (2nd time I have said this) start telling peeps they need to run 100 loops etc then that is when issues can arise!
On the other side of the coin scaremongers are not really bringing anything to this thread and if not happy with this App should ignore thread!
I have not heard of anyone killing a modern CPU with it esp when the CPU will throttle and PC will no doubt BSOD/Restart!
Nothing worst than running Prime for many hours trying to reach a target of XX hours and you get up next morning and its crashed, at least if you run a few loops of this 1st it takes far less time and you know it stands a better chance of passing Prime and not wasting your time before you adjust settings to try Prime once again!
AG is there any truth that SB does not ideally like LinPack/LinX or has that been changed by updates as AFAIK this App is based opn LinPack/LinX correct?
Just Chinese Whispers I have read on various sites, no biggy!
I guess that it's become fashionable to call people names when you disagree with their perspecive but be that as it may I repeat that IBT was not designed
to be any sort of stability test but rather to root out any weak chips by putting such a heavy load on them that any potentiak weakness would cause failure.
You folks decide for yourself whether you want to risk expensive chips by running this app.
No I've had my 3 cents of imput and will leave this thread.
Best of luck to you all and a safe and Happy holiday.
Think this threads resident Troll needs to take his meds as I do not see anyone calling him names! :confused:
lol yeah really movieman, stop being such a troll LMAO
lol you guys are taking this way too seriously :p:
i seriously don't think i can be any more down under (unless i go to Tasmania -errrrrrrrr) :p:
meh anyone that calls movieman a troll is a bit "loco"