| profiler [message #13251] |
Thu, 29 October 1998 00:00  |
J.D. Smith
Messages: 214
Registered: August 1996
|
Senior Member |
|
|
Anyone else noticed that:
profiler,'module'
doesn't work but
profiler, 'MODULE'
does?
I thought IDL was supposed to be case insensitive!
--
J.D. Smith |*| WORK: (607) 255-5842
Cornell University Dept. of Astronomy |*| (607) 255-6263
304 Space Sciences Bldg. |*| FAX: (607) 255-5875
Ithaca, NY 14853 |*|
|
|
|
|
|
|
| Re: Profiler [message #27386 is a reply to message #13251] |
Mon, 22 October 2001 11:37  |
K. Bowman
Messages: 330
Registered: May 2000
|
Senior Member |
|
|
In article <3BD099C3.6557B059@noaa.gov>, Pavel A. Romashkin
<pavel.romashkin@noaa.gov> wrote:
> Ken,
>
> I just tried the simpliest thing that came to my mind. I set a
> breakpoint to the first line of a program, then set profiler to profile
> all. When you do one-stepping through the code, profile log updates for
> every line. I just had to make its window active (click on it) for it to
> update. You could use Step over if you didn't want to profile user procedures.
>
> Hope this helps. Surely beats chopping code into dozens of separate
> routines :-)
While not entirely click-free, that sounds like it will do the trick.
Ken
|
|
|
|
| Re: Profiler [message #27408 is a reply to message #13251] |
Sat, 20 October 2001 20:01 |
Paul Woodford
Messages: 43
Registered: June 2000
|
Member |
|
|
I recently used trace to figure out where I was slowing down. Set it to
run with no delay, and then watch to see which line it pauses on.
Paul
|
|
|
|
| Re: Profiler [message #27418 is a reply to message #13251] |
Fri, 19 October 2001 14:23 |
Pavel A. Romashkin
Messages: 531
Registered: November 2000
|
Senior Member |
|
|
Ken,
I just tried the simpliest thing that came to my mind. I set a
breakpoint to the first line of a program, then set profiler to profile
all. When you do one-stepping through the code, profile log updates for
every line. I just had to make its window active (click on it) for it to
update. You could use Step over if you didn't want to profile user procedures.
Hope this helps. Surely beats chopping code into dozens of separate
routines :-)
Cheers,
Pavel
"K. Bowman" wrote:
>
> In article <9qoeu3$hm8$1@mwrns.noaa.gov>, Pavel Romashkin
> <pavel.romashkin@noaa.gov> wrote:
>
>> Its been a while, nobody answered. Have you tried it, Ken?
>> I only profiled perfectly debugged code (yeah, right) in order to optimize
>> it. I never felt the need to profile line by line, it was obvious from the
>> profile which part of the code was casuing the drag.
>
> We have not figured out how to profile line-by-line. Doesn't seem to
> be possible Turning on all the system routines, etc. didn't help
> either for this problem.
>
> We have resorted to the simple expedient of commenting out blocks of
> code (where it won't affect the computation) or moving blocks of code
> into temporary subroutines. In our case, at least, it has turned out
> to be relatively simple to isolate the computationally-intensive block.
>
> We have discovered a couple of minor algorithmic optimizations that we
> are testing.
>
> Ken
|
|
|
|
| Re: Profiler [message #27420 is a reply to message #27333] |
Fri, 19 October 2001 13:11 |
K. Bowman
Messages: 330
Registered: May 2000
|
Senior Member |
|
|
In article <9qoeu3$hm8$1@mwrns.noaa.gov>, Pavel Romashkin
<pavel.romashkin@noaa.gov> wrote:
> Its been a while, nobody answered. Have you tried it, Ken?
> I only profiled perfectly debugged code (yeah, right) in order to optimize
> it. I never felt the need to profile line by line, it was obvious from the
> profile which part of the code was casuing the drag.
We have not figured out how to profile line-by-line. Doesn't seem to
be possible Turning on all the system routines, etc. didn't help
either for this problem.
We have resorted to the simple expedient of commenting out blocks of
code (where it won't affect the computation) or moving blocks of code
into temporary subroutines. In our case, at least, it has turned out
to be relatively simple to isolate the computationally-intensive block.
We have discovered a couple of minor algorithmic optimizations that we
are testing.
Ken
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
