Rain, rain
May. 1st, 2010 09:56 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
altivoBut it isn't raining. They keep telling us that it is, or it will, but the sky is clear here.
Wasted part of the day geeking around with the computers. My brief foray into Microsoft's 8-bit FORTRAN from 1983 this week left me curious. I ended up with two test programs running on CP/M and written in Fortran-80 (Microsoft's name for what was really a FORTRAN 77 compiler.) One uses the Monte Carlo method for approximating the value of PI. The other calculates a table of square roots using two different methods and prints the results in two columns so the values can be compared easily.
I modified both programs slightly so they would run on g77, the FORTRAN compiler from the Gnu project. The main issue was just the file OPEN statement. Then I remembered that the DEC Alpha was once considered a premier scientific machine with enhanced math libraries. I do have FORTRAN installed on my Alphas, so why not? I ported the programs there as well. All three environments were fairly close in speed on the square root table. The 8-bit processor was a bit less consistent, showing some variations in the fifth digit after the decimal, but that's still a lot better than a slide rule and the speed was amazing. The Gnu compiler running on Linux seemed fastest, but that may be subjective, and especially so since the Alpha was communicating over an Ethernet link and hub to get to the Linux display.
The Monte Carlo program showed more variation. The 8-bit processor definitely became draggy when the last thousand or so iterations (of 10K) were reached, but it still delivered results to match those of the 32-bit Intel/Linux combo, and even the 64-bit Alpha with OpenVMS. Speedwise, the Alpha and the Intel seemed to be in a dead heat on this.
Does it matter that a 27 year old FORTRAN compiler for an 8-bit CPU environment can still stand up to today's equipment? Perhaps not to most, but I find it interesting or even encouraging.
Wasted part of the day geeking around with the computers. My brief foray into Microsoft's 8-bit FORTRAN from 1983 this week left me curious. I ended up with two test programs running on CP/M and written in Fortran-80 (Microsoft's name for what was really a FORTRAN 77 compiler.) One uses the Monte Carlo method for approximating the value of PI. The other calculates a table of square roots using two different methods and prints the results in two columns so the values can be compared easily.
I modified both programs slightly so they would run on g77, the FORTRAN compiler from the Gnu project. The main issue was just the file OPEN statement. Then I remembered that the DEC Alpha was once considered a premier scientific machine with enhanced math libraries. I do have FORTRAN installed on my Alphas, so why not? I ported the programs there as well. All three environments were fairly close in speed on the square root table. The 8-bit processor was a bit less consistent, showing some variations in the fifth digit after the decimal, but that's still a lot better than a slide rule and the speed was amazing. The Gnu compiler running on Linux seemed fastest, but that may be subjective, and especially so since the Alpha was communicating over an Ethernet link and hub to get to the Linux display.
The Monte Carlo program showed more variation. The 8-bit processor definitely became draggy when the last thousand or so iterations (of 10K) were reached, but it still delivered results to match those of the 32-bit Intel/Linux combo, and even the 64-bit Alpha with OpenVMS. Speedwise, the Alpha and the Intel seemed to be in a dead heat on this.
Does it matter that a 27 year old FORTRAN compiler for an 8-bit CPU environment can still stand up to today's equipment? Perhaps not to most, but I find it interesting or even encouraging.
