xkcd

by **Govalant** » Fri Feb 08, 2008 2:21 am UTC

I made a program to measure the megaflops of my pc, and was surprised by the low results. Not once it went over 4.

I'm pretty sure I tested this before and got like 60. Is my memory failing or what?

thanks.

by **wing** » Fri Feb 08, 2008 2:28 am UTC

You're running a parallelized operating system. You're not going to get accurate results, no matter what.

by **Govalant** » Fri Feb 08, 2008 2:32 am UTC

Good point.

I'll try to do it without windows.

by **tinyterror** » Fri Feb 08, 2008 10:02 pm UTC

What do you mean you wrote a program to test megaflops? As it has been said before, doing this is not impossible, it is just hard. If you do it with an OS running on the system, you will not get correct results.

by **davean** » Fri Feb 08, 2008 11:09 pm UTC

tinyterror wrote:What do you mean you wrote a program to test megaflops? As it has been said before, doing this is not impossible, it is just hard. If you do it with an OS running on the system, you will not get correct results.

If you schedule it batch-realtime you will. It is possible to do, you just have to throw out any time sliced results or avoid time slicing. SCHED_FIFO maybe?

xkcd

usual megaflops?

usual megaflops?

Re: usual megaflops?

Re: usual megaflops?

Re: usual megaflops?

Re: usual megaflops?

Who is online