bob@imspw6.UUCP (Bob Burch) (09/02/88)
I hate to say it, but the best shot available at porting mainframe Fortran stuff to micros would probably involve one of Jack Tramiel's 1040 STs and a copy of the good Absoft Fortran which was available for Macs and Amigas last I saw and should be there for the STs by now. The amount of pain and grief saved as compared to trying to get any kind of halfway serious system up under DOS will probably pay for the Atari several times over. The Absoft 68000 Fortran has been around on 680x0 based super micros since about 1983 and is very highly rated. Ted Holden HTE
silvert@dalcs.UUCP (Bill Silvert) (09/02/88)
In article <141@imspw6.UUCP> bob@imspw6.UUCP (Bob Burch) writes: >I hate to say it, but the best shot available at porting mainframe >Fortran stuff to micros would probably involve one of Jack Tramiel's >1040 STs and a copy of the good Absoft Fortran which was available >for Macs and Amigas last I saw and should be there for the STs by >now. The amount of pain and grief saved as compared to trying to >get any kind of halfway serious system up under DOS will probably >pay for the Atari several times over. The Absoft 68000 Fortran has >been around on 680x0 based super micros since about 1983 and is very >highly rated. I wish that I could agree, but it doesn't jibe with my experience. I've been doing numerous ports of ecological simulation software to micros and have found the AbSoft ST compiler pretty flaky. We reported several bugs with version 2.2 and had to pay about $60 US for an upgrade to 2.3 which was if anything worse. We are now trying a port to the Mac, so we should soon know how well that version of the AbSoft compiler works. For MS-DOS I've used both the Lahey and Microsoft compilers (both 4.0 and 4.1 of MS). Lahey is fast to compile, but we needed the Microsoft feature of emulating an 8087 (Lahey requires one), so we switched to Microsoft. It takes a long time to compile, but it is pretty reliable. Execution times are good. Microsoft is very sensitive to programming errors and generates lots of error messages, but this is actually kind of useful. One cute bug in Microsoft is that if you write a program in which the same unit number is used for both direct access and sequential files it won't compile, even if you are careful about closing and opening. A minor annoyance, but I've compiled lots of stuff and it seems OK. In all fairness I should point out that the AbSoft Fortran for the ST is much cheaper than the others and has good GEM support. Bill Silvert -- Bill Silvert, Habitat Ecology Division. Bedford Institute of Oceanography, Dartmouth, NS, Canada B2Y 4A2 UUCP: ...!{uunet,utai,watmath}!dalcs!biomel!bill CDN: biomel@cs.dal.CDN BITNET: bs%dalcs@dalac.BITNET
mcdonald@uxe.cso.uiuc.edu (09/02/88)
>The amount of pain and grief saved as compared to trying to >get any kind of halfway serious system up under DOS will probably >pay for the Atari several times over. The Absoft 68000 Fortran has >been around on 680x0 based super micros since about 1983 and is very >highly rated. I have used both the Absoft Fortran (on the Mac) and Microsoft Fortran on the PC. The results are far superior on the PC. In the first place, do Atari's have coprocessor chips? Ordinary Mac's don't (can't) and so Fortran programs run like molasses. (Mac II's excepted.) The code generation by Absoft is excruciatingly bad. Microsoft's current product is very good indeed, in fact its optimizer is very nearly up to that of VAX Fortran (say 80% as good). Every legal Fortran 77 program I have downloaded from our Vax has run the first time. DougMcDonald