Time/iter

PurdueME · April 11, 2009, 12:02

When printing the residuals during iterations, one of the things printed out is "time/iter". What does this number mean? For example, one iteration printed out 43:20:10. The iteration takes about one second, so I don't understand what that number means and I can't seem to find it online.

PurdueME · April 13, 2009, 15:06

Anyone have any idea?

ngrube · April 13, 2009, 20:02

To increase your chance of getting useful replies, you should tell us what program you are using, and what you are trying to use it for.

time/iter might mean something like CPU time per step or it could mean physical time per step.

In your case, unless you are dealing with some really big scales, neither makes much sense. If your simulation is crashing and generating NaNs, sometimes that can lead to an erroneously large delta t (computed from a CFL condition) for the next step. Perhaps that would give you a ridiculous number like you are reporting. If your simulation is running fine, then my guess would be that the number you see is simply incorrect and not worth worrying about. Or maybe there is some interpretation that I am overlooking. Obviously a manual is the best source of answers, and without knowing what program you are using, people probably can't help much.

PurdueME · April 13, 2009, 20:41

Wow, I didn't realize I never mentioned the program -- total oversight on my behalf.

I'm using Fluent 6.3 to run a steady turbulent model.

I'm referring to the print-out on the screen during iteration of a solution.

The iter continuity x-velocity y-velocity k epsilon time/iter line.

I understand all the others, obviously, but not the time/iter one.

ngrube · April 13, 2009, 21:02

Quote:

Originally Posted by PurdueME

Wow, I didn't realize I never mentioned the program -- total oversight on my behalf.

I'm using Fluent 6.3 to run a steady turbulent model.

Okay. :-) Well, I should defer to people who use Fluent then. Even if you don't have a paper manual, for such a sophisticated program as that, I would be surprised if there isn't a manual available from a help menu or something.

PurdueME · April 13, 2009, 22:50

Quote:

Originally Posted by ngrube

Okay. :-) Well, I should defer to people who use Fluent then. Even if you don't have a paper manual, for such a sophisticated program as that, I would be surprised if there isn't a manual available from a help menu or something.

Oh, I've looked through the documentation for it. Tried every corner of it that I could think of. Tried a Google search for it. Tried talking to my professor who uses it daily (he said he didn't even notice that it was there). It isn't so much a matter of technical importance as it is curiosity, honestly.

ngrube · April 13, 2009, 23:07

I wonder whether people who run Fluent on totally different machines get numbers that make more sense? If you are curious, you might try to find someone who uses as different an architecture as possible and ask them. It could be that however your system returns the system time info differs from what the Fluent programmers were expecting. I have seen garbage timing results in similar situations but I always just got the timing info in a different way rather than getting to the bottom of the mystery.

A_Lyaskin · April 14, 2009, 04:09

I'm not a Fluent user (BTW, why don't you try posting your question in Fluent forum, not in the general one?), but my guess that it can be CPU time passed from the beginning of the simulation till the end of the iteration.

jpando · September 7, 2012, 15:01

I think the time/iter is the total time left for how many iterations are left. i.e. 400:10:05 9910 will mean that there will be 400 hrs, 10 min, 5 sec, left to calculate 9910 iterations at the current iteration rate.

twolf59 · July 1, 2015, 11:22

A little late, but just want to say that this makes the most sense since the number appears to countdown from when I start my simulation.

jthiakz · October 25, 2018, 08:47

Estimated run time for the no.of iterations that has been set.
After every iterations this is estimated w.r.t the remaining iterations and reported in the run window.

April 11, 2009, 12:02	Time/iter	#1
PurdueME New Member Andrew Join Date: Apr 2009 Posts: 5 Rep Power: 17	When printing the residuals during iterations, one of the things printed out is "time/iter". What does this number mean? For example, one iteration printed out 43:20:10. The iteration takes about one second, so I don't understand what that number means and I can't seem to find it online.

September 7, 2012, 15:01		#9
jpando New Member Join Date: May 2012 Posts: 26 Rep Power: 13	I think the time/iter is the total time left for how many iterations are left. i.e. 400:10:05 9910 will mean that there will be 400 hrs, 10 min, 5 sec, left to calculate 9910 iterations at the current iteration rate. sq_325 likes this.

October 25, 2018, 08:47	iter/time	#11
jthiakz Member Thiagu Join Date: Oct 2012 Location: India Posts: 60 Rep Power: 13	Estimated run time for the no.of iterations that has been set. After every iterations this is estimated w.r.t the remaining iterations and reported in the run window. T_Rus-UK14 likes this.

April 13, 2009, 15:06		#2
PurdueME New Member Andrew Join Date: Apr 2009 Posts: 5 Rep Power: 17	Anyone have any idea?

April 13, 2009, 20:02		#3
ngrube New Member Nathan Grube Join Date: Mar 2009 Posts: 14 Rep Power: 17	To increase your chance of getting useful replies, you should tell us what program you are using, and what you are trying to use it for. time/iter might mean something like CPU time per step or it could mean physical time per step. In your case, unless you are dealing with some really big scales, neither makes much sense. If your simulation is crashing and generating NaNs, sometimes that can lead to an erroneously large delta t (computed from a CFL condition) for the next step. Perhaps that would give you a ridiculous number like you are reporting. If your simulation is running fine, then my guess would be that the number you see is simply incorrect and not worth worrying about. Or maybe there is some interpretation that I am overlooking. Obviously a manual is the best source of answers, and without knowing what program you are using, people probably can't help much.

April 13, 2009, 20:41		#4
PurdueME New Member Andrew Join Date: Apr 2009 Posts: 5 Rep Power: 17	Wow, I didn't realize I never mentioned the program -- total oversight on my behalf. I'm using Fluent 6.3 to run a steady turbulent model. I'm referring to the print-out on the screen during iteration of a solution. The iter continuity x-velocity y-velocity k epsilon time/iter line. I understand all the others, obviously, but not the time/iter one.

April 13, 2009, 23:07		#7
ngrube New Member Nathan Grube Join Date: Mar 2009 Posts: 14 Rep Power: 17	I wonder whether people who run Fluent on totally different machines get numbers that make more sense? If you are curious, you might try to find someone who uses as different an architecture as possible and ask them. It could be that however your system returns the system time info differs from what the Fluent programmers were expecting. I have seen garbage timing results in similar situations but I always just got the timing info in a different way rather than getting to the bottom of the mystery.

April 14, 2009, 04:09		#8
A_Lyaskin Member Anton Lyaskin Join Date: Mar 2009 Location: Samara, Russia Posts: 66 Rep Power: 17	I'm not a Fluent user (BTW, why don't you try posting your question in Fluent forum, not in the general one?), but my guess that it can be CPU time passed from the beginning of the simulation till the end of the iteration.

July 1, 2015, 11:22		#10
twolf59 New Member daniel Join Date: Sep 2014 Posts: 17 Rep Power: 11	A little late, but just want to say that this makes the most sense since the number appears to countdown from when I start my simulation.