Originally Posted by blakmumba;13120663
I seriously question the need for all this 24hr testing going on your brand new CPU, I see numerous cases here of guys bragging about their 24hr stress testing, or saying someones overclock is not stable because they haven't benched for 24 hrs.
How much experience have you had with overclocking? This attitude is generally reserved for amateur overclockers who have never had corrupted data.
A stress test is running all of your CPU cores flat out for however long you run the test, unless of course a core drops out because your overclock is unstable, I see the need for short term testing say an hour or maybe 2, because it won't take that long for an unstable overclock to show itself, but 24hrs in my opinion is not necessary.
I think you are confused. The error is not found because the CPU is at 100%. A major error, or more likely, many tiny irrecoverable errors, happen at any % of CPU usage. The programs are designed to find inconsistencies as fast as possible, utilizing all resources available to determine the stability. Its not lets see how hard we can squeeze your CPU till it breaks. Is it stressful? Sure, is the stress that causes error? No, that would be instability or heat. Another good way to test your cooling solution too I might add.
There is not a thing on this planet designed to last forever, your precious CPU included when companies like Intel and AMD, guarantee their CPUs for 3 yr warranty time period, they're confident their CPUs will actually go past that time period, however that calculation is based on 3 yrs and beyond of normal use.
They know their CPUs are capable of lasting under normal spec'd use longer than the 3 yrs or they would not warranty them that long.
Not sure why this is even here. This has nothing to do with stability testing, unless you are assuming that all CPUs that are brand new should not be tested @ stock settings? Which a CPU should be tested to ensure its not defective. Rare, but it can happen.
Now the cold hard fact is that overclocking your CPU will shorten its lifespan, but we accept that for the additional speed we gain, most of us will replace that cpu with an upgrade before is life expectancy arrives anyway, so taking it out of specifications is acceptable.
Has nothing to do with stability testing. Its about error prevention.
However that's with normal day to day use, surfing the web, audio and video editing and converting, gaming, office documentation Etc, and though serious bench stress testing is necessary to achieve a stable overclock, what damage does stress testing over a 24 hr time period and beyond, actually do to shorten the lifespan of your new CPU.
Why would you even overclock then?
For example if you stress test your new quad core CPU for 24hrs, and all 4 cores are running at 100% for 24 hrs, how much normal daily processing is lost, because no daily application uses all 4 cores at 100% for 24hrs, so at least think about that when you overclock your CPU, and use at least some kind of wisdom in your stress testing.
Again its not the stress that creates the error, its stress that finds the error. Errors can occur with as little as 1% core usage. The alternative is to use your computer until one day, your boot sector is corrupted, and Windows 7 repair, can't fix the files as the damage has occurred widespread through your file-system. At this point you are even lucky to recover data off the hard drive. Assuming you are still not using an unstable overclock, recovering data, will most likely keep it corrupt/re-corrupt it.
Because if you don't think you're adding to shortening the life of your CPU, you're wrong!
Stress testing your CPU, if its adequately cooled, is not shortening your hardwares life, its the extra voltage/clock cycles from the overclock if anything. It sounds like you have a ton still to learn / grasp and thats okay. Overclocking has many useful applications, such as a technical server upgrade, gaming, scientific calculations, protein folding, etc. Many of those demand stability for personal and professional reasons. Folding requires 100% accurate data, or its wasteful time for both the user and server, which anyday can provide a cure to cancer (hopefully.) A non tested overclock in that field is extremely frowned upon by many people who dedicate entire machines to just folding.
I am not saying you have to stress test your CPU, or it will die and kill everything in your computer, but you are not going to convince anyone who knows what they are doing that its a waste of time.
I don't get why people are ever against having a fully stable system? It takes about a day if you actually have good settings. Maybe apathy or general laziness, but still, it will create so many headaches in the long run trying to figure out why the computer just doesn't want to work today.Edited by RagingCain - 4/14/11 at 2:19pm