red930 Posted July 10, 2005 Posted July 10, 2005 BTW I run OCCT for two successful passes before ever attempting a run of prime95. OCCT fails gracefully and has never locked up my rig or caused an OS problem. Share this post Link to post Share on other sites More sharing options...
groovetek Posted July 10, 2005 Posted July 10, 2005 i think the point he wanted to make was that, there should be a benchmark utility that can stress out the system as much as an OS+StressTest together can do. Unfortunately, MemTest86+ is still pretty much the only one, and hardly stresses out the memory, let alone the system enough to gauge a point of perfect stability. I DO find however, that it is generally quite good for finding a point where at least the system will be DESKTOP STABLE, that is, can boot up, and at least do basic tasks, and usually the max memory overclock is no more than 5-10mhz lower than the speed there MemTest first begins to report errors. Share this post Link to post Share on other sites More sharing options...
Rezag1000 Posted July 10, 2005 Posted July 10, 2005 ah hell, I just might create another parition and install windows on that one too. make that my "testing OS" As far as registering, 30 days is plenty to test just the CPU. Share this post Link to post Share on other sites More sharing options...
Kamel Posted July 10, 2005 Posted July 10, 2005 dos is a 1970's single threaded os. The early ones shipped on a single floppy. The complex pre windows ones shipped on 3 floppies. In the early days it was written for and ran on 8080, z80, 80186, 80286 80386 etc. procs 64k of memory in the day, was a lot. but yes, it was written for and ran on a two register 8080 proc that is like a model A to a formula 1 car of today. If you want to call running the old model a around the block once a stable test ... go for it. However, you have a lot to learn about the difference between a unit level test and a system level test how about linux? (which is in many ways a much more advanced OS than windows) Share this post Link to post Share on other sites More sharing options...
Angry_Games Posted July 10, 2005 Posted July 10, 2005 linux is not even close to being comparable with DOS...not at all... Share this post Link to post Share on other sites More sharing options...
Kamel Posted July 10, 2005 Posted July 10, 2005 yea, that was my point... it could be done in linux, in which case it could be done on a single floppy. Share this post Link to post Share on other sites More sharing options...
Angry_Games Posted July 10, 2005 Posted July 10, 2005 as tim stated so very clearly... dos is a 1970's single threaded os. The early ones shipped on a single floppy. The complex pre windows ones shipped on 3 floppies. In the early days it was written for and ran on 8080, z80, 80186, 80286 80386 etc. procs 64k of memory in the day, was a lot. but yes, it was written for and ran on a two register 8080 proc that is like a model A to a formula 1 car of today. If you want to call running the old model a around the block once a stable test ... go for it. However, you have a lot to learn about the difference between a unit level test and a system level test you are very much missing the point if you think that a DOS-based application which is limited to 640k of system RAM by default could even be comparable with a complex, system stressing operating system designed for multiple threads and processes running at once. the bootable linux cd is not DOS like in any way. Just because it is (usually) command-line driven has absolutely nothing to do with it being like DOS, again a simple single threaded program. you could probably write a simple DOS-based cpu stresser...in fact i believe there are probably some out there...but what good does that do if you don't have a complex OS like Linux or Windows installed? That's the real test...who cares if it is DOS stable...DOS and Windows/Linux are like comparing apples to blue whales...no comparison can be made at all except that both are living organisms (ie both DOS and Windows/Linux are computer programs that run on x86 cpu's...thats the only real comparison that can be made) take this example as this should explain to you: you have a car...in a lab setting on a dyno it can crank out 480hp @ 440ft/lbs of torque with a maximum speed of 170mph hey...thats great right? Sure it is. (imagine this as testing in DOS) but take it out on a REAL track/highway. whoa...wait...now you have...wind resistance...tire wear...road conditions...all sorts of complexities that are NOT in that lab testing. (imagine this as testing in Windows/Linux). now...you may still get 480hp and still the same amount of torque...but your maximum speed might not be the same...but now you encounter all sorts of OTHER things that you could never encounter on that dyno in a lab... air gets up under the frame and maybe makes the car unstable...or worse, lifts the front end up off the ground. maybe the tires/suspension, while great while only one axle turns on a dyno, is not really that good in real situations and the tire blows, the alignment is so whacked that the tires wear quickly, the car steers to the left or right too bady, it shakes like a washing machine trying to escape the laundry room... then the engine itself...great on a dyno when it is not actually having any wind resistance...and it is NOT carrying any real weight and inertia...but on a real track, it not only has to contend with wind, but how much the car actually weighs as it pushes or pulls (depending on FWD or RWD cars), which also leads to making the engine work a lot harder and can cause all kinds of other problems.... i could go on, but if you don't understand the differences then there's really not much more to discuss. Installing the OS is like taking that badass lab car out on a real highway and finding out what it will really do. if you burn out an engine or blow a tire in the lab on the dyno (fail memtest in DOS) then yeah, chances are the car is gonna do that and possibly worse on the real highway...but if it passes on the dyno, that doesn't prove anything other than it can function in a simple situation...taking it out on a real highway might show you all the other problems that the car has....or the car might perform excellent just like it did on the lab's dyno... Share this post Link to post Share on other sites More sharing options...
Rezag1000 Posted July 10, 2005 Posted July 10, 2005 then I guess memtest86 is a waste of time as well, huh? Share this post Link to post Share on other sites More sharing options...
red930 Posted July 10, 2005 Posted July 10, 2005 Once more with feeling... Memtest serves its purpose to qualify your RAM before you ever install the OS. Prime95 serves its purpose to qualify your rig once the OS is installed. That is all... then I guess memtest86 is a waste of time as well, huh? Share this post Link to post Share on other sites More sharing options...
timpanogos Posted July 10, 2005 Posted July 10, 2005 In testing environments (I have a fair amount of experience in software testing) .. UNIT testing always has it's place and all unit testing must be thoroughly and succesfully completed before you bother going into the much more difficult task of system level testing ... of course memtest is important ... as a unit test fwiw, I wrote Novell's original "Super Lab" automation system used to do "System Level" testing of the interactions of complex os/utilitlies/hardware Share this post Link to post Share on other sites More sharing options...
Recommended Posts
Please sign in to comment
You will be able to leave a comment after signing in
Sign In Now