chas@gtss.UUCP (05/26/87)
In article <8705240941.AA17640@cogsci.berkeley.edu> bryce@COGSCI.BERKELEY.EDU (Bryce Nesbitt) writes: ... >The CLI TYPE and COPY commands have a lot of overhead due to DOS and BCPL, >the language they are written in. It is poor measure of blitz. Use DME.. ... Its not a poor measure of Blitz since you can't go through life using nothing but DME. Well, I can't. We gotta see what Blitz does with a variety of applications. I would still like to know if Type's speed was deliberately reduced. I haven't been running down Blitz, I just believe in quantifying claims and making clear how measurements were made, so that they can be replicated and understood. A non-BCPL Copy (cp) has just been posted. Anyone want to compare its speed with AmigaDos's Copy to address the point raised by Bryce? Not me, I refuse to touch cp until it stops messing with my file timestamps. I find people have been mentioning DME for a reason, however. We are fortunate that DME lets us issue commands like, paraphrasing, page down 50 times. Many editors would not let us do enough automatic pagings to time them. So now for the results. Using my 'standard' 73350 byte file (without clear screens, which are superfluous in a paging environment like DME). I was forced to use my watch's stopwatch mode. I got these times -- averaging over 4 trials -- for paging through the file (strictly Topaz, of course): Without Blitzfonts 19.56 s (Comparable with Copy with Blitzfonts) With Blitzfonts 5.21 s A speed increase of 275% (the overall factor was 3.75). That's about a tenth of a second a page -- 14078 characters per second, guys! Around 100,000 baud if that had meaning here. Not bad for a cheap little box ;-). But as I say we can't spend the rest of our lives using DME (Though it is tempting, what?). Now I guess I'll have to cough up the moola for the Blitz that handles other fonts to see how its speeds compare. I mean nobody else is going to time these things as far as I can tell :-). -- Charles Cleveland EDU: chas@ss.physics.gatech.edu Georgia Tech School of Physics UUCP: ...!{akgua,allegra,amd,hplabs,ihnp4, Atlanta, GA 30332 masscomp,ut-ngp,rlgvax,sb1,uf-cgrl, unmvax,ut-sally}!gatech!gtss!chas
chas@gtss.UUCP (05/27/87)
In article <122@gtss.UUCP> chas@gtss.UUCP (Charles Cleveland) [that's me] writes: >A non-BCPL Copy (cp) has just been posted. Anyone want to compare its speed >with AmigaDos's Copy to address the point raised by Bryce? Not me, I refuse OK, I volunteer. Using my by-now-standard scrolling and non-scrolling text files I get no measurable difference between Cp and Copy. I suspect that Copy is slower that DME because DME can take advantage of the knowledge that it *never* has to scroll. It must take advantage of something. But I've never seen DME's source: would the author care to summarize how he sped up it's output? Kim has pointed out to me that I have used Blitz where I meant BlitzFont. Let me clarify. I have never mentioned Blitz intentionally except to say that that wasn't what I was talking about. Blitz is a file viewer which goes like blue blazes. BlitzFont accelerates text output for other applications (which also go like blue blazes if they give BlitzFont a chance). -- Charles Cleveland EDU: chas@ss.physics.gatech.edu Georgia Tech School of Physics UUCP: ...!{akgua,allegra,amd,hplabs,ihnp4, Atlanta, GA 30332 masscomp,ut-ngp,rlgvax,sb1,uf-cgrl, unmvax,ut-sally}!gatech!gtss!chas