Page 1 of 1

What means Mnbf/s, ns/day?

Posted: Fri Aug 20, 2010 8:44 pm
by tati
Hi all,

Just curious what means Mnbf/s and ns/day?
Other question: how long must to stay tuned for description of project 6012? It is nearing his end (just 102 WU's to go)...

Re: What means Mnbf/s, ns/day?

Posted: Fri Aug 20, 2010 10:08 pm
by Zagen30
Not sure abuot Mnbf/s. ns/day is nanoseconds/day. Are you getting that 102 WUs left figure from the server stats page regarding server 130.237.232.140? A) that server handles multiple projects, so there'd be X number of WUs remaining across all the projects it carries, and B) the next timestep of each particular trajectory is generated from the previous completed timestep, so there are almost always more WUs being generated as finished ones are turned in.

Re: What means Mnbf/s, ns/day?

Posted: Sat Aug 21, 2010 7:32 pm
by Galahad
Mnbf/s = Mega non-bonded forces per second

Re: What means Mnbf/s, ns/day?

Posted: Sun Aug 22, 2010 8:42 am
by Napoleon
Umm, nanoseconds per day? When I previously checked, there were a fixed amount of seconds per day: 24(h/day) *60(min/h) *60(s/min) = 86400(s/day). That would amount to 86400*10^9(ns/day), still a fixed value. Am I missing something here? What would be the purpose of such a non-unit?

EDIT: Describes the performance of simulating a particular protein folding, how many ns / day get simulated, maybe? Apparently I managed to misfold a brain cell or two on yesterday's pub crawl. :twisted:

Re: What means Mnbf/s, ns/day?

Posted: Sun Aug 22, 2010 8:54 pm
by bruce
Yes, it's the amount of simulated time that is being calculated.

Some atomic processes happen at the femto-second time frame (10^-15) so it takes a lot of work to simulate a single nano-second (10>-9) of motion. Many of the folding processes being simulated actually happen at the micro- (10^-6) to milli-second (10^-3) range. So it takes a lot of WUs to create a single trajectory, and statistically speaking, it takes a lot of trajectories to characterize the folding process of a single protein.

Proteins with relatively few atoms fold faster than proteins with a larger number of atoms, so the relationship between simulated time and clock time varies, but the unit of ns/day is a useful way to describe what's happening scientifically with that particular protein.

In case you wondered, FAH is a HUGE undertaking.