Blatantly commercial post here from a newbie who just joined the forum for this purpose. I work for Perforce, and IMSL has been part of the business for a while now. I have IMSL in my portfolio of products I deal with. Absoft ended, who embedded IMSL in their product their business without a migration path in Sept 2022, and we’ve talked occasionally to Mac users who are casting about for a next step.
We’ve offered IMSL for Fortran for Mac in the past, and may again. I’d like to hear from folks who’ve relied on IMSL for Mac in the past through Absoft and have continued running Fortran on the Mac. It would be good to know what compilers you’re using, which Mac HW platforms (Intel, M1, M2, M3).
I hope you won’t mind my broadening the thread beyond IMSL-with-Absoft. I once bought with personal funds the Digital Visual Fortran compiler for Windows, professional version, which included IMSL. The extra cost of the compiler with IMSL was a few hundred dollars. When IMSL was acquired, the extra cost of IMSL rose greatly, I believe to more than $1000, and I never used IMSL again. A commercial numerical library such IMSL or NAG can save users much time, but a user will consider not just the current cost but the risk that the cost could spike or that the compiler for the library is purchased becomes obsolete (such as Absoft).
It would be nice if IMSL were available for gfortran and if a user could buy IMSL for multiple compilers (say gfortran, ifx, nvfortran) at a discount to the sum of the individual prices.
Before the Perforce acquisition, IMSL was part of Visualnumerics and then Roguewave. In those times, there was a company hosted user forum for users of IMSL. The late Dr. Richard Hanson used to respond regularly in that forum.
I used IMSL on a CDC 6600 in the early 1970s, then IMSL4 with CVF, later with Intel and Absoft Fortran.
Now, however, if you are not the holder of a current license, it appears that Perforce does not want to be bothered by users of older versions, and I am afraid to ask what an IMSL license costs these days.
I see what you mean about the cost of the bundled IMSL rising so much; inflection points such as acquisitions do represent risks to one’s budget. Otherwise, I don’t think we’ve increased the price of the libraries themselves so rapidly.
The way to mitigate the risk of the libraries falling out of support is (surprise!) to buy them directly from Perforce.
Yes, it would be good to have IMSL support gfortran. I’m recommending that to the product manager. A license covers all supported compilers on that platform, and I doubt that would change, so you’d get your wish there.
IMSL does go back a ways, doesn’t it? We have employees who came from Visual Numerics, some have been with us for 35 years, and I think we have some in the range of 45 years. Suffice to say, they know the product and area pretty well. It doesn’t look like you’re wanting for experience in the field either
In general, yes, Perforce supports users with support contracts, and is less likely to place a priority on helping someone without one. The version of the software, I don’t think it would make much difference; if you’ve paid for support, you’ll get it.
I’m not burdened with the responsibility of quoting prices for our products - I leave that to the sales guys. Happy to talk to you about the product of course, and pull in resources as necessary.
@mecej4 It saddened me to learn about Dr. Richard Hanson passing from this post. I have his book Numerical Computing with Modern Fortran (along with Tim Hopkins) handy here on my bookshelf. I have a signed copy from him. I remember working with him on several examples (fixing bugs in ifort to get the examples to work). This siam book sits in a place of honor in my bookshelf with my short stack of useful Fortran books.
Also sorry to hear about Richard. My interaction with him goes back to his participation in the old DoD HPCMP PET program in the early 2Ks. He was working with Henry Gabb and Clay Breshears on the Fortran pthreads library they developed. I believe he was at Rice University at the time. The PET program required periodic stays down in Vicksburg for our academic partners. I lent Richard a dual processor Linux box I built so he could have something to work on until he got all the approvals he needed to access the DoD systems.
Very nice guy.
Dr. Hanson passed away a little over seven years ago. We find his name in the introductory comment lines of many of the BLAS and Lapack source files at Netlib. Here is an obituary of R.J.Hanson, written by another illustrious numerical analyst, Dr. Fred Krogh (retd. from JPL).
