IRAF and More - For Mac Using Astronomers and Astrophysicists
Well, the download page was rather confusing collection of links, and had the Intel installers rather hard to find sometimes, so I reorganized it a bit today. Hopefully, it’ll be less confusing now.
.Generally speaking, IRAF tasks are single-threaded, meaning they don’t take advantage of multiple cores or multiple processors. So, wanting to use all the power in my new Mac Pro - I discovered one can run tasks in the background in IRAF with an &, just like in UNIX.
So, for those of us with Intel processors that have 2 cores, or Mac Pros with 2 2-core chips, this is a good way to speed up tasks by running multiple processes simultaneously. For data reduction, this can speed things up, since each extraction of say a spectra is independent of the next one.
For example, I automate my extractions of echelle spectra. Usually I make doit files and just run them all sequentally. But now I can set up a “doit” file that contains lines like so
myspec file1 output1 &
myspec file2 output2 &
myspec file3 output3 &
myspec file4 output4
....
and then four more lines similarly. The first 3 are run in the background, the fourth is run normally… so the fifth line of this doit file doesn’t get run until the fourth (and presumably at about the same time the first 3) tasks finish. Feed this into the cl like so…
cl> cl < doit
And the end result? IRAF uses nearly 400% of my CPU… check out the screen cap below…
Update Ok this does work. If I want to log what’s happening in each “myspectask” I just need to do…
myspectask input1 output1 > log1 &
making sure the ampersand is after the redirect to a log file.
Update Number 2 Apparently, some times the next batch of 3 background tasks tries to fire up before the first three are done, and then for some reason I get a “no more background slots” error, even though I’m told that IRAF can have 20 background tasks. So, I stuck a sleep 3 after each non-background task to let the background tasks “catch up” and finish before the next four simultaneous tasks are started. That seems to work…
The headline says it all. I’ve actually tested this one. It’s on the download page and a direct download link is below.
Don’t forget that TABLES 3.6 is required to use STSDAS.
.MSCRED 4.8 is now available from the download page. I had to completely rebuild these, but thanks to IRAF.net I did not have to compile them myself. I haven’t had a chance to test them out, but they should work. Please post a comment if they work properly or not.
You can also find direct download links below.
The PPC link was broken for … well since this story was posted until Dec 8th. I guess that’s about four days - and yet no one posted a comment or emailed me. Perhaps everyone is using Intel these days…
.Happy Thanksgiving to those in the U.S. My gift to everyone, a TABLES 3.6 installer for Intel Macs (Download Link) which is also listed on the download page.
I have also completed some work on my new script-based system to help speed the process of creating and updating these installers. An STSDAS installer and a PPC version of TABLES should be up soon … the bottleneck with STSDAS is the sheer time to download and then to upload the installers.
Any problems with this installer, please note in the comments.
.Posting over at the IRAF.net Forums, Mike Fitzpatrick discusses the release of new Macintel binaries of IRAF and of external packages that fix a bus error. (Note the handy RSS based feed of IRAF.net forum posts in the left sidebar of this site.)
While work continues (slowly) on the final V2.13 release I’ve decided to update the macintel binaries in the download area that includes this bug fix. At the same time I’ve updated the external package directory with a more complete build set of binaries, and especially macintel binaries linked against the bug fix (e.g. even though you can get macintel STSDAS binaries in their latest release, the ones from this site will avoid the bus error problem).
There is no immediate need for every macintel system to update binaries, however if you experience a bus error using the V2.13b2 system I’d suggest you update the binaries before reporting an error. You do not need to update sources, simply reinstall the ib.macx.x86 and nb.macx.x86 distributions, and/or the ‘bin.macintel’ binaries for the package in question.
Of course, the few Intel .pkg installers I have made available use the old potential-bus-error prone binaries. So… I guess sometime I’ll update them to fix this, and subsequent Macintel binary .pkg installers will include the fix as well.
I have a plan to write some sort of script that automagically will create .pkg installers. Or least, partially automate the process. Certainly, some of the creation is rote and could be streamlined. I think there’s even a command line method of using package maker…
.TABLES and STSDAS 3.6 were released today, with binaries built for both PPC MacOS X and “Macintel” systems. You can check them out from the STScI STSDAS download page.
Also released was STSCI Python. I don’t really know what that is but it has to do with Pyraf.
I will endeavor to build .pkg installers for these before the end of the month. Until then, remember how to install them. First, STSDAS requires TABLES. So, install TABLES first.
/iraf/extern/tables//iraf/extern/tables/bin.macosx or /iraf/extern/tables/bin.macintel.bin to the appropriate bin.whatever/iraf/extern/stsdas and the binaries in the appropriate /iraf/extern/stsdas/bin.something directory, also fixing the bin symlink.Then you should be good to go. Questions? Feel free to leave a comment.
.