Re: [tlug] issues with format of double - SOLVED

[Michal Hajek <hajek1@example.com>]

Hello, 


just a short note for those who remember my previous struggle with speed
and format of double in relation with measurement which I do...

Short summary: I have a system with several instruments connected to the
pc by means of GPIB bus. I need the data to be collected as quickly as
possible. First I thought the speed bottleneck was the atof() function,
but it was definitly _not_ the case (but I keep the Subj. line for
reference). Recently I have come back to the issue and I finaly have the
solution.  

The problem was in the "-pg" compilation option. After communication
with the GPIB bus linux driver author [1] I found out, that this is the
couse of my problem. Without this option, everything works perfectly,
although there are still some speed limitations. I will work on them
later. 

Another possibly useful information, which appeared during my
investigation of the problem is, that perhaps Keithley gpib chips (most
probubly those build in their older devices) are not capable of very
quick responses and thus usleep(10) or even usleep(100) might be
neccessary between gpib calls. In my case, this is not needed.  

Thanks everyone for your kind help. 

[1] see the archives of linux-gpib mailing list here:
http://sourceforge.net/mailarchive/forum.php?forum_id=2853

Best regards
Michal