| Re: Is IDL 8.1 Useable!? [message #77521] |
Fri, 09 September 2011 07:21  |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
David Fanning writes:
> OK, let's think about this for a minute. What did we
> learn yesterday?
>
> First, a LOT of people find IDL 8.1 either crashing itself
> or crashing their machines. In my case, we found that this
> is because a graphics driver was out of date.
I think I spoke too soon. :-(
Some things are MUCH better this morning. The program I
wrote yesterday that absolutely crashed my machine actually
works! But, when I went back to my original problem of
trying to make a filled contour plot that had a defined
number of colors, and I ran the program Mark Piper had
sent me yesterday morning, I crashed again (twice!) in
exactly the same way.
So, I would say IDL 8.1 and function graphics is pretty
much hopeless for me. I don't think there is anything
I can do to make these things work, except possibly
offer to purchase the machines and hardware ITTVIS
is using in their offices. (If they are selling!)
I still think the problem lies in the COLORBAR function,
because this is still the place where my machine goes
south.
Maybe I should get used to the "new" way of thinking
that doesn't use color bars. That really is so 1970s! ;-)
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
| Re: Is IDL 8.1 Useable!? [message #77522 is a reply to message #77521] |
Fri, 09 September 2011 06:46  |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
M. Katz writes:
>
> print, !version
> { x86_64 darwin unix Mac OS X 8.1 Mar 9 2011 64 64}
> This is a late 2010 MacBookPro w/ 8GB RAM running Axl's code.
>
> **IDL Command line speeds
> DG elaps: 1.0532050
> NG elaps: 10.646090 (worked fine)
>
> **IDL Workbench speeds
> DG elaps: 0.29327607 (WOW!)
> NG elaps: bought the farm. (Ugh!)
> (Note on NG case: IDL not responding. CPU usage 0.1%, Program is
> easily quit from Activity Monitor without interrupting other processes
> on the machine).
>
> I guess that's why I'm still a xterm command-line / BBEdit guy.
OK, let's think about this for a minute. What did we
learn yesterday?
First, a LOT of people find IDL 8.1 either crashing itself
or crashing their machines. In my case, we found that this
is because a graphics driver was out of date. IDL's function
graphics depends on hardware acceleration for its speed.
(I know, but let's just assume this is true.) The question
I am asking myself this morning is how many of these crashes
can be explained by out-of-date or inadequate (because not
everyone can or will pay big money for these) graphics cards?
I don't know the answer to that, but for sake of argument, let's
say the number is 50%. Wouldn't it make more sense, then, for
IDL to make *software* rendering the default mode and to let
people know that if they wanted to make graphics *faster* they
would have to turn hardware acceleration on. This way, when
things go south, you know EXACTLY who to blame: the NVIDIA
engineers!
I realize this can potentially make function graphics SLOWER
than it is now (who would have thunk it!), but I think most
of the slowness is due to iTool overhead and not graphics
rendering. This is especially true for most 2D plots, which I
imagine are the bulk of things people do.
The upside of doing this, is that your graphics system actually
works for people and you don't get an undeserved reputation for
producing shoddy products.
Anyway, just thinking out loud. :-)
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
|
|
|
|
| Re: Is IDL 8.1 Useable!? [message #77530 is a reply to message #77529] |
Thu, 08 September 2011 19:40 |
Paul[3]
Messages: 18
Registered: September 2010
|
Junior Member |
|
|
I keep trying to use the object graphics, and I keep hitting hangups.
Lack of control, slow output (even if I'm adding objects to a buffered
window), missing features. (For example, contour labels are stretched,
skewed, upside-down, etc.) Postscript file output is also larger with
the object graphics.
Even for applications where I don't hit the bugs in the object
graphics, the slowness and large file-size make it difficult enough to
work with that I usually don't bother. Dissappointing, really; they're
awful convenient.
|
|
|
|
| Re: Is IDL 8.1 Useable!? [message #77539 is a reply to message #77530] |
Thu, 08 September 2011 10:07 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
Paul van Delst writes:
> Let me now try it in the Workbench:
>
> IDL> p = randomn(rien, 1024L*1024)
> IDL> t=systime(1) & plot, p, PSYM=3 & print,'DG elaps:',systime(1)-t
> DG elaps: 0.95030618
> IDL> t=systime(1) & q=plot(p, LINESTYLE=6, SYMBOL='dot') & print,'NG elaps:',systime(1)-t
> % Loaded DLM: XML.
> % IDLITWINDOW::ONEXPOSE: Failure to acquire window rendering context.
> % Unable to acquire device context.
> % Execution halted at: $MAIN$
>
> Oh boy! (wipes tears from eyes.... :o)
>
> <shake attribute="fist" target="monitor">
> I'll give you a rendering context....
> </shake>
Please let me know if you hear of any big sale of
1970s-era computers and software!
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
| Re: Is IDL 8.1 Useable!? [message #77540 is a reply to message #77539] |
Thu, 08 September 2011 09:33 |
Paul Van Delst[1]
Messages: 1157
Registered: April 2002
|
Senior Member |
|
|
Hello,
alx wrote:
> On 8 sep, 17:41, David Fanning <n...@dfanning.com> wrote:
>> Paul van Delst writes:
>>> The biggest issue I have with NG is that it is so agonisingly slow. A big selling point for these sorts of products
>>> (IDL, matlab, etc) is that they make you more productive because visualising your data is easy and quick. NG kills the
>>> latter and, based on posts to this newsgroup, is doing a good job of nobbling the former (although I attribute some of
>>> that to resistance to shifting one's perception anchor from how one thinks things *should* work, to how they actually
>>> *do* work).
>> Well, if there was some instruction in how they *do* work
>> we wouldn't be having to make so many guesses as to how
>> they *should* work!
>>
>> My plan was to help provide a solution to the first
>> problem, but it is going awry. :-(
>>
>> Cheers,
>>
>> David
>> --
>> David Fanning, Ph.D.
>> Fanning Software Consulting, Inc.
>> Coyote's Guide to IDL Programming:http://www.dfanning.com/
>> Sepore ma de ni thui. ("Perhaps thou speakest truth.")
>
> I'am not so sure that DG is faster than NG !
> alx.
>
> IDL> p = randomn(rien, 1024L*1024)
> IDL> t=systime(1) & plot, p, PSYM=3 & print,'DG elaps:',systime(1)-t
> DG elaps: 16.110000
> IDL> t=systime(1) & q=plot(p, LINESTYLE=6, SYMBOL='dot') & print,'DG
> elaps:',systime(1)-t
> % Loaded DLM: XML.
> NG elaps: 11.740000
Interesting!
Here are my results (I cut-n-pasted your commands into my command line):
IDL> p = randomn(rien, 1024L*1024)
IDL> t=systime(1) & plot, p, PSYM=3 & print,'DG elaps:',systime(1)-t
DG elaps: 0.74409294
IDL> t=systime(1) & q=plot(p, LINESTYLE=6, SYMBOL='dot') & print,'NG elaps:',systime(1)-t
% Loaded DLM: XML.
NG elaps: 18.482085
IDL> print, !version
{ x86 linux unix linux 8.1 Mar 9 2011 32 64}
If my DG plots took as long as on your system, I would never have used IDL in the first place!
Let me now try it in the Workbench:
IDL> p = randomn(rien, 1024L*1024)
IDL> t=systime(1) & plot, p, PSYM=3 & print,'DG elaps:',systime(1)-t
DG elaps: 0.95030618
IDL> t=systime(1) & q=plot(p, LINESTYLE=6, SYMBOL='dot') & print,'NG elaps:',systime(1)-t
% Loaded DLM: XML.
% IDLITWINDOW::ONEXPOSE: Failure to acquire window rendering context.
% Unable to acquire device context.
% Execution halted at: $MAIN$
Oh boy! (wipes tears from eyes.... :o)
<shake attribute="fist" target="monitor">
I'll give you a rendering context....
</shake>
:oD
cheers,
paulv
|
|
|
|
| Re: Is IDL 8.1 Useable!? [message #77541 is a reply to message #77540] |
Thu, 08 September 2011 09:22 |
lecacheux.alain
Messages: 325
Registered: January 2008
|
Senior Member |
|
|
On 8 sep, 17:41, David Fanning <n...@dfanning.com> wrote:
> Paul van Delst writes:
>> The biggest issue I have with NG is that it is so agonisingly slow. A big selling point for these sorts of products
>> (IDL, matlab, etc) is that they make you more productive because visualising your data is easy and quick. NG kills the
>> latter and, based on posts to this newsgroup, is doing a good job of nobbling the former (although I attribute some of
>> that to resistance to shifting one's perception anchor from how one thinks things *should* work, to how they actually
>> *do* work).
>
> Well, if there was some instruction in how they *do* work
> we wouldn't be having to make so many guesses as to how
> they *should* work!
>
> My plan was to help provide a solution to the first
> problem, but it is going awry. :-(
>
> Cheers,
>
> David
> --
> David Fanning, Ph.D.
> Fanning Software Consulting, Inc.
> Coyote's Guide to IDL Programming:http://www.dfanning.com/
> Sepore ma de ni thui. ("Perhaps thou speakest truth.")
I'am not so sure that DG is faster than NG !
alx.
IDL> p = randomn(rien, 1024L*1024)
IDL> t=systime(1) & plot, p, PSYM=3 & print,'DG elaps:',systime(1)-t
DG elaps: 16.110000
IDL> t=systime(1) & q=plot(p, LINESTYLE=6, SYMBOL='dot') & print,'DG
elaps:',systime(1)-t
% Loaded DLM: XML.
NG elaps: 11.740000
|
|
|
|
| Re: Is IDL 8.1 Useable!? [message #77542 is a reply to message #77541] |
Thu, 08 September 2011 08:41 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
Paul van Delst writes:
> The biggest issue I have with NG is that it is so agonisingly slow. A big selling point for these sorts of products
> (IDL, matlab, etc) is that they make you more productive because visualising your data is easy and quick. NG kills the
> latter and, based on posts to this newsgroup, is doing a good job of nobbling the former (although I attribute some of
> that to resistance to shifting one's perception anchor from how one thinks things *should* work, to how they actually
> *do* work).
Well, if there was some instruction in how they *do* work
we wouldn't be having to make so many guesses as to how
they *should* work!
My plan was to help provide a solution to the first
problem, but it is going awry. :-(
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
|
|
| Re: Is IDL 8.1 Useable!? [message #77545 is a reply to message #77543] |
Thu, 08 September 2011 08:09 |
Paul Van Delst[1]
Messages: 1157
Registered: April 2002
|
Senior Member |
|
|
David Fanning wrote:
> Folks,
>
> I don't want to put a big damper on all the fun, but
> I have a serious question. Is anyone actually using
> IDL 8.1 and function graphics?
Yes.
But only for small, simple things. When I have to plot 2^18 points to double-check my Fourier-interpolation routines I
sure as heck don't use NG - I think it would take a week for the plot to actually appear! (just kidding... mostly :o).
The biggest issue I have with NG is that it is so agonisingly slow. A big selling point for these sorts of products
(IDL, matlab, etc) is that they make you more productive because visualising your data is easy and quick. NG kills the
latter and, based on posts to this newsgroup, is doing a good job of nobbling the former (although I attribute some of
that to resistance to shifting one's perception anchor from how one thinks things *should* work, to how they actually
*do* work).
But, when the planets align and the wind blows the incense smoke in just the right counterclockwise spiral, the NG plots
that result sure are purty. Look great in documents too.
> Not only are function graphics commands completely
> opaque to me when I just try to do the simplest things,
> but in the past several days I have COMPLETELY buggered
> my machine four times running function graphics
> programs! This is a nifty little machine, with
> a pretty darned expensive graphics card, the very
> latest Windows 7, JAVA updates, etc. And it has been
> completely and utterly buggered four times!
> This is four times more than this machine has been
> buggered since I got the damn thing.
>
> I can't even catch its attention long enough to
> summon the task manager to kill a process. I have
> to kill the whole machine by holding the start button
> in for 5 seconds. It takes about 20 minutes to re-boot
> and get going again. And as God is my witness, I don't
> dare run the function graphics program again, for fear
> I'll be twiddling my thumbs for another half hour!
>
> I don't know. This just seems utterly hopeless to me. :-(
I've never had these sorts of issues with IDL 8.0, 8.0.1, or 8.1. I'm a die-hard linux+command-line only bloke though so
my experience are probably not indicative of anything in particular.
cheers,
paulv
|
|
|
|
| Re: Is IDL 8.1 Useable!? [message #77549 is a reply to message #77545] |
Thu, 08 September 2011 06:46 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
David Fanning writes:
> If I want to continue learning about the new features
> in IDL 8 (questionable right now!) I guess I'll have
> to use the IDL command line to do it. :-(
If I go to the Start menu and chase down the IDL 8.1
folder, there are several options under the "Tools"
directory. Here, for example, is where I find the
link to the IDL command line.
One of the links is to "IDL (32-bit)". Thinking this
might solve some of my problems, I chose this. This
crashes, too, won't start the Workbench, and refers
me to a log file. The file is a long list of JAVA
exception errors, etc. I can't really make heads or
tails of it.
My question is, has anyone in a Windows 64-bit
environment successfully run this IDL 32-bit
application? Does it help at all? Is it worth
my time to pursue this?
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
| Re: Is IDL 8.1 Useable!? [message #77550 is a reply to message #77549] |
Thu, 08 September 2011 06:23 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
alx writes:
> I guess that IDL 8.1 would be usable, but actually, on my Windows
> machines, the Workbench is not ! As stated in a previous message, I am
> facing frequent Workbench crashes in the course of long sessions. The
> crashes may occur during graphic screen plotting as well as during the
> execution of a simple statement. I even got one crash while only
> moving the mouse ... In my opinion, the crashes might be due to the
> accumulation with time of "small" errors, like internal memory leaks,
> default in acquiring handles, etc... They certainly are related to the
> Workbench, since I never had one by using the IDL line command.
I tried running this offending program at the IDL command
line this morning. It still crashes, but with an error
message instead of buggering my machine. The error was this:
%Unable to allocate memory: Creating new procedure frame.
The line causing the error was one in which I was trying
to create a function colorbar. (I actually got the program I am
running from someone at ITTVIS.)
I have passed this information along to the support folks
at ITTVIS. Maybe it will help them track this down.
If I want to continue learning about the new features
in IDL 8 (questionable right now!) I guess I'll have
to use the IDL command line to do it. :-(
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
| Re: Is IDL 8.1 Useable!? [message #77551 is a reply to message #77550] |
Thu, 08 September 2011 03:53 |
greg.addr
Messages: 160
Registered: May 2007
|
Senior Member |
|
|
Hi David,
I gave up on 8.0 after about two weeks of the vanishing workbench (Windows 7, 64-bit). At that time, there seemed to be plenty of people having the same experience and complaining about it here. Just this last week, I decided to give 8.1 a go, since the grumbles seemed pretty much to have gone away. But I got the same as you - instead of just vanishing like before, the whole machine seizes up - no task manager, nothing. If I'm lucky, I can still shake the mouse pointer in frustration before switching off. Admittedly, it happens less often (say, twice a day) than before, but after two days, I was back on 7.1.
Utterly hopeless... agreed.
Greg
|
|
|
|
| Re: Is IDL 8.1 Useable!? [message #77552 is a reply to message #77551] |
Thu, 08 September 2011 01:59 |
lecacheux.alain
Messages: 325
Registered: January 2008
|
Senior Member |
|
|
On 8 sep, 05:50, David Fanning <n...@dfanning.com> wrote:
> Folks,
>
> I don't want to put a big damper on all the fun, but
> I have a serious question. Is anyone actually using
> IDL 8.1 and function graphics?
>
> Not only are function graphics commands completely
> opaque to me when I just try to do the simplest things,
> but in the past several days I have COMPLETELY buggered
> my machine four times running function graphics
> programs! This is a nifty little machine, with
> a pretty darned expensive graphics card, the very
> latest Windows 7, JAVA updates, etc. And it has been
> completely and utterly buggered four times!
> This is four times more than this machine has been
> buggered since I got the damn thing.
>
> I can't even catch its attention long enough to
> summon the task manager to kill a process. I have
> to kill the whole machine by holding the start button
> in for 5 seconds. It takes about 20 minutes to re-boot
> and get going again. And as God is my witness, I don't
> dare run the function graphics program again, for fear
> I'll be twiddling my thumbs for another half hour!
>
> I don't know. This just seems utterly hopeless to me. :-(
>
> Cheers,
>
> David
>
> P.S. Please, someone, give me some good news! Or at least
> some ideas for things I can try.
>
> --
> David Fanning, Ph.D.
> Fanning Software Consulting, Inc.
> Coyote's Guide to IDL Programming:http://www.dfanning.com/
> Sepore ma de ni thui. ("Perhaps thou speakest truth.")
I guess that IDL 8.1 would be usable, but actually, on my Windows
machines, the Workbench is not ! As stated in a previous message, I am
facing frequent Workbench crashes in the course of long sessions. The
crashes may occur during graphic screen plotting as well as during the
execution of a simple statement. I even got one crash while only
moving the mouse ... In my opinion, the crashes might be due to the
accumulation with time of "small" errors, like internal memory leaks,
default in acquiring handles, etc... They certainly are related to the
Workbench, since I never had one by using the IDL line command. The
appearence of the crashes is like David Fanning is describing: you
suddendly completely loose the control of your desktop (in fact the
windows refreshing mechanism becomes entirely corrupted, on the whole
desktop), and you have to rely on the Task Manager to stop IDL
(several successive attempts are usually needed) and to restart the
Explorer... All the machines I am working with are running various
Windows systems (win32 or 64, XP or W7), are regularly kept updated
(including JAVA) and are equipped with different graphic cards (but
all from NVIDIA). There is certainly something bad with Eclipse, with
Eclipse on Windows or with the implementation of the IDL GUI through
Eclipse by ITTVIS.
Hoping that this will be corrected in 8.2 ...
alx.
|
|
|
|
| Re: Is IDL 8.1 Useable!? [message #77599 is a reply to message #77543] |
Sun, 11 September 2011 12:52 |
Craig Markwardt
Messages: 1869
Registered: November 1996
|
Senior Member |
|
|
On Sep 8, 11:32 am, David Fanning <n...@dfanning.com> wrote:
> Paul van Delst writes:
>> The biggest issue I have with NG is that it is so agonisingly slow.
>
> I find I have a LOT easier time restarting my machine
> if I can catch the crash early and hit the re-boot button
> as soon as possible. But, I am having a difficult time
> telling if the machine has crashed, or the graphics are
> just taking a LONG time to show up! ;-)
>
> Cheers,
>
> David
>
> P.S. Let's just say if the delay is over 10 seconds and
> the fan kicks on, it is probably a crash.
I think ITT could use this as the new slogan for their IDL marketing
materials.
"IDL 8: finding new ways to reboot your computer faster!"
Craig
P.S. Meanwhile, I could be the testimonial for IDL version 5.
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
