|
From: | Peter Teeson |
Subject: | Re: [Bug-apl] miscellāneum |
Date: | Tue, 5 Mar 2019 14:54:50 -0500 |
In the days when they had a commercial time-sharing APL system, I was fortunate to work at I. P. Sharp in the Zoo (the R & D group that was responsible for the system SW) We made a number of ad hoc special case hand crafted optimizations. In my case that was for inner and outer product. Luckily I shared an office with someone who was a senior coder who had a particular interest in algorithms in general. He gently guided me and I will never forget his kindness. If we were not able to get at a minimum a 10% increase we didn’t bother. As you can imagine those two operators were good candidates and we did achieve a significant improvement in a number of frequently occurring special cases. That was on various mainframes - a 360 Model 50 at first (later twinned), an Amdahl, twin Model 158’s and even a Burroughs 6700 which never made it into our production. But these optimizations were hand picked by the analyses that were made on the fertile areas - not generic algorithms. (Long live Knuth! Still waiting for the next volume.) Good memories of the early days and what a blessing to have been able to participate. respect…. Peter
|
[Prev in Thread] | Current Thread | [Next in Thread] |