Igor on Linux ???
| February 17, 2010 - 00:34 | |||
|---|---|---|---|
|
We are often asked if we might port Igor to Linux. The answer is that it is a possibility for the future but not the near future. Some Igor users have reported success running Igor on Linux under Wine or under Crossover Office which, as I understand it, is a commercialized version of Wine. If you have comments about Igor on Linux or want to encourage (or discourage) us from porting to Linux in the future, feel free to add your comments here. |
|||
Joined: 2009-09-22
Location: Italy
If you have comments about Igor on Linux or want to encourage (or discourage) us from porting to Linux in the future, feel free to add your comments here.
I for one strongly encourage you to do it! (moreover, I am more than willing to beta-test it :-) )
There has been some talk about better integrating Gizmo into Igor, to bring it on par with 2D graphs (in terms of usability and output). I think this would be an excellent opportunity to consider linux as well, as OpenGL is already quite mature on the platform. Naturally, porting the 2D graphics is a different matter completely, but then again, hardware-accelerated 2D routines have been available on all platforms for quite a while...
The non-graphics part of Igor should be comparatively less demanding to port. At any rate, if you also plan 64bit support, adding linux as a platform for this part should be a relatively minor effort during the necessary code review...
Thank you for Igor, and keep up the good work!
Joined: 2007-08-16
Location: Switzerland
Both my group and WM would benefit from a Linux version of Igor Pro. In my group with 5 scientists we are 2 using Windoze, 2 Linux (Ubuntu 9.x) and one Mac OS X. I use my own copy of Igor regularly, my colleagues have been learning it . My penguin colleagues (I am a windozer myself) have experimented with Igor under wine, which works, but it is slow, they have to use tweaks like the UseOldGraphics, and it is far more prone to freeze than in a native environment. I think they are currently using the windows version of Igor via Windows in a virtual machine, if they have to. Part of our data is happy with specialized free software like Gwyddion, part needs a more thorough analysis. Right now it is mainly me, who does the latter.
What's in it for WM? I successfully Igor-vangelized our general physics labs and the students are a mixture of OS users just the same way as we "grown-ups". Reaching out to the Linux users among them is bound to generate some more Igor users every year from those, who now mix GNUplot, Matlab and wossnames... :-)
Cheers!
Daniel
========
Time to find out, whether you are dead right, or just dead.
Joined: 2007-08-16
Location: United States
Having Igor working on Linux would be good for a series of reasons.
- It would enlarge the community. In absolute numbers Linux users might be a minority respect WinOS and OSX ones. However they represent a significant and growing number among the PC users involved in number crunching and data plotting/analysis.
- Developing countries as China and India are most prone to use Linux rather than WinOS. A Linux version of Igor would open up new markets for WMs.
- It would allow Igor to run on large clusters of PCs using an OS that makes better use of the hardware resources. The trend is clear in science. The amount of data to process, and not only in science, is growing following at least the Moore's Law. How would Igor keep up with that?
- Linux-embedded hardware systems would have an easy way to be programmed and interfaced. And Linux runs on practically any type of CPUs.
- Linux code is open then it should be much easier to write XOPs to interface Igor with any available hardware acquisition devices.
Just my two cents.
Joined: 2007-12-15
Location: United States
I am running the latest beta release of wine (1.1.37 if memory serves) built as a universal (32- and 64-bit) binary via Macports on my Snow Leopard iMac at work. (I'm home sick today.) I managed to get MS Office 2007 to run on it, and the speed is blazing fast. Unbelievably fast, actually.
So, I'm wondering if upgrading to the latest wine version will help. Or, could the problem be the way Ubuntu distributes the load onto the CPUs?
I'll try to install the IGOR demo on my work computer and report back, but my impression is that the latest wine builds are much, much more functional and a lot faster, at least on Snow Leopard.
Joined: 2009-02-26
Location: United Kingdom
I would just like to second the previous responses (and have my "vote" counted :). As has been pointed out in previous posts: especially for number crunching Linux is where a lot of action is at and being able to use Igor on that platform would be practical (not to mention comfortable).
Joined: 2010-02-17
Location: United Kingdom
i think it would be great to be able to remotely log into a linux box through a terminal (ssh?) and set igor running various (number crunching) tasks through the command line with no graphical interface. ive tried igor on wine and have been largely successful. However, im too scared to run any of my multi-threaded processing functions as im expecting something to go wrong instantly. Has anyone tried this?
Joined: 2007-06-29
Location: United States
It's unlikely that anything could get damaged, unless your multithreaded code deletes files and data. If it does, keep backups, and just try it. About the worst would be that you would have to re-boot.
John Weeks
WaveMetrics, Inc.
support@wavemetrics.com
Joined: 2008-02-22
Location: United States
I have had success using Igor in WINE, but I have a list of problems documented here:
http://www.igorexchange.com/node/1098
which people are encouraged to add to. Much easier to implement than a version of Igor native to Linux would be fixing the issues listed in that thread, and being open to fixing more as they are reported. However, I don't know how many of these issues can be fixed by changing Igor, and how many can only be fixed by changing WINE. It is an open source project, though, so one could always contribute... and to the extent that WINE does not faithfully replicate the Igor experience, it is probably a failure to faithfully replicate the Win32 API, which means it would be generally useful to the WINE community.
Is this a worthwhile intermediate step until a version of Igor native to Linux is released? I think it probably would be.
Joined: 2007-12-15
Location: United States
Per RGerkin's post above, I will post some benchmarks that I just ran on the demo version of IGOR Pro for Windows which I just installed on my iMac within wine 1.1.39 (which was built this morning via Macports). HTH.
FWIW, here are the benchmark 2.02 numbers. I hope they are relevant to claims of sluggishness reported above under wine.
**** test on Windows XP Professional Service Pack 3 (Build 2600)5.1.2600 using 6.12 and 21 passes; 2.66 GHz iMac
Create new graph time: 533.26ms, relative speed= 0.56
big data update time: 152.04ms, relative speed= 0.94
curve fit time: 4.21ms, relative speed= 0.45
user curve fit time: 16.51ms, relative speed= 2.15
double complex fft time: 812.66µs, relative speed= 1.81
single complex fft time: 744.13µs, relative speed= 1.60
double real fft time: 391.29µs, relative speed= 1.44
single real fft time: 348.60µs, relative speed= 1.41
5 pass smooth time: 317.84µs, relative speed= 1.71
Sort 8192 points time: 12.47ms, relative speed= 3.07
WaveStats time: 594.70µs, relative speed= 0.45
simple eqn time: 290.00µs, relative speed= 4.81
exp eqn time: 552.53µs, relative speed= 3.64
sqrt eqn time: 388.65µs, relative speed= 4.55
sin eqn time: 550.42µs, relative speed= 2.11
User fit fctn time: 492.83µs, relative speed= 2.62
MatrixOp eqn time: 67.97µs, relative speed= 0.35
**** done ****
total test time= 21.3781
Joined: 2010-04-26
Location: United States
In my experience it is the GUI and the data acquisition that will take the most effort if anyone is to convert Igor to run on Linux.
Joined: 2008-06-30
Location: United States
Not being a programmer, unfortunately I cannot contribute to the discussion of
how difficult it would be to make IGOR run on Linux.
I am running it occasionally on Linux using wine and most things just work fine.
But if there was a Linux version of IGOR, I definitely would be a user.
Joined: 2010-08-03
Location: Japan
does anyone have a code to convert Igor commands (usually saved in the procedure window) to Gnuplot commands? I am not so familiar with the Gnuplot commands. Therefore,
I am not sure whether the Gnu also is composed of commands like Igor Pro. If there would be a converter, it would be very helpful for me because I can skip several processes of manual
conversion for the graphs which were created by Gnuplot.
Joined: 2008-02-22
Location: United States
I think that it would be much, much easier to make contributions to the WINE project that fixed the remaining Igor-under-WINE bugs than to create an Igor-for-Linux from scratch. Plus, you'll probably get support from the WINE dev community -- and unlike the rest of the Linux world they probably won't care that Igor is closed-source, since that's pretty much true of everything people want to run under WINE.
It's also possible that in fixing the remaining bugs, you can actually maintain the Windows build more-or-less as is, without even needing a separate Linux build. Some of the bug fixes will be in WINE itself, and some others won't actually change the behavior of the Windows build in Windows. It seems like a very within-reach goal.
That said, Igor under WINE is no way to run hardware devices; but in my world, at least, most of my devices don't have Linux drivers anyway.
Rick
Joined: 2012-01-31
Location: Switzerland
My job would be greatly facilitated by a Linux version of Igor.
My acquiring computer runs on Win XP, but all our servers are based on Linux.
I guess we are not the only ones who do the hard core number crunching on Linux machines.
Vassilis
Joined: 2010-04-13
Location: United States
Here is Igor's wine appDB entry.