gl8f@astsun.astro.Virginia.EDU (Greg Lindahl) (06/07/90)
The TT is not yet shipping. Until it does: -- we won't have good benchmarks. -- we won't know what the street price will be. -- we won't know what is bundled with the machine. All of the current speculation is useless until it actually ships. So can we *please* stop talking about it until we have something hard to talk about? Thanks. -- "Perhaps I'm commenting a bit cynically, but I think I'm qualified to." - Dan Bernstein
andyc@hplsla.HP.COM (Andy Cassino) (06/08/90)
gl8f@astsun.astro.Virginia.EDU (Greg Lindahl) writes: | The TT is not yet shipping. Until it does: | | -- we won't have good benchmarks. Well, there are some good benchmarks... Darek Mihocka just uploaded to GEnie his report on benchmarks he ran on the TT at it's Canadian unveiling. (Sorry, the report carries a copyright notice so I cannot post it!) However, I think it tends to support the comment regarding a 1/3 speed increase for non-recompiled code. Apparently, this performance is projected (by Atari) to improve as the TOS on the TT is optimized. Disclaimer: I deny everything! %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Andy Cassino % % uucp: hplabs!hplsla!andyc internet: andyc@lsid.hp.com % % Hewlett-Packard Lake Stevens Instrument Division % % 8600 Soper Hill Road MS60 Everett, WA 98205-1298 % % Land-line: (206) 335-2211 % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
gl8f@astsun.astro.Virginia.EDU (Greg Lindahl) (06/08/90)
In article <5440122@hplsla.HP.COM> andyc@hplsla.HP.COM (Andy Cassino) writes: >gl8f@astsun.astro.Virginia.EDU (Greg Lindahl) writes: > >| The TT is not yet shipping. Until it does: >| >| -- we won't have good benchmarks. > >Well, there are some good benchmarks... Darek Mihocka just uploaded to >GEnie his report on benchmarks he ran on the TT at it's Canadian unveiling. This is *exactly* why I say we won't get good benchmarks until it ships. We don't have any/enough details about the benchmarks he ran... and I won't be satisfied until I can run my benchmarks myself. >(Sorry, the report carries a copyright notice so I cannot post it!) However, >I think it tends to support the comment regarding a 1/3 speed increase for >non-recompiled code. Apparently, this performance is projected (by Atari) to >improve as the TOS on the TT is optimized. That comment isn't useful until we get an idea of what the benchmark does. Disk I/O? Screen I/O? Floating point? Compute-intensive integer operations? Just throwing a number around with no details is not useful, and inspires a lot of moaning and groaning. I'm not trying to insult here, but just make the point that the information isn't complete until we can all sit an ST next to a TT and run our favorite program. -- "Perhaps I'm commenting a bit cynically, but I think I'm qualified to." - Dan Bernstein
csbrod@medusa.informatik.uni-erlangen.de (Claus Brod ) (06/11/90)
gl8f@astsun.astro.Virginia.EDU (Greg Lindahl) writes: >All of the current speculation is useless until it actually ships. So >can we *please* stop talking about it until we have something hard to >talk about? Talking about the TT is necessary and not useless. The sooner ATARI gets to know that we don't like it much, they will start thinking about improvements for the TT/X. (Well, I can't tell if _you_ like the TT, so I should not say "we don't like it", so: I don't like it.) Moreover, the TT is "in the shops" in Germany which means that selected dealers get one for a week or two to play around with it, just like some developpers. ---------------------------------------------------------------------- Claus Brod, Am Felsenkeller 2, Things. Take. Time. D-8772 Marktheidenfeld, West Germany (Piet Hein) csbrod@medusa.informatik.uni-erlangen.de ----------------------------------------------------------------------
andyc@hplsla.HP.COM (Andy Cassino) (06/12/90)
| >Well, there are some good benchmarks... Darek Mihocka just uploaded to | >GEnie his report on benchmarks he ran on the TT at it's Canadian unveiling. | | This is *exactly* why I say we won't get good benchmarks until it | ships. We don't have any/enough details about the benchmarks he ran... | and I won't be satisfied until I can run my benchmarks myself. | ... | That comment isn't useful until we get an idea of what the benchmark | does. Disk I/O? Screen I/O? Floating point? Compute-intensive integer | operations? I erroneously assumed people would recognize Darek Mihocka as co-author of the Quick Index benchmark utility for the ST, and therefore surmise he used that as the basis of his report. Darek's report contains some very specific numbers for the specific areas his program tests. People interested in relatively hard information on the TT will find the report interesting. Since it is copyrighted, I must refer those people to GEnie. Disclaimer: The opinions above are solely those of the author, who has no pecuniary interest in any of the companies mentioned. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Andy Cassino % % uucp: hplabs!hplsla!andyc internet: andyc@lsid.hp.com % % Hewlett-Packard Lake Stevens Instrument Division % % 8600 Soper Hill Road MS60 Everett, WA 98205-1298 % % Land-line: (206) 335-2211 % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
gl8f@astsun9.astro.Virginia.EDU (Greg Lindahl) (06/12/90)
In article <5440123@hplsla.HP.COM> andyc@hplsla.HP.COM (Andy Cassino) writes: > >I erroneously assumed people would recognize Darek Mihocka as co-author >of the Quick Index benchmark utility for the ST, and therefore surmise >he used that as the basis of his report. This report was published in this week's Z*Net. He stated that the TT ran 350% to 500% of a straight ST on the Quick Index benchmark with the cache on, and 30% slower with the cache off. I would interpret this to mean that it ran 250% to 350% faster with the cache off. He also said that it appears that the TT is only 30% faster than an ST with a T16 accellerator board, on screen redraws. If that's where the "1/3" figure is coming from, then 1/3 is not a good number to describe the speedup you get on a typical application. No benchmarks for programs which do floating point, and it would seem that the TT he was using did not have any "fast ram" in it, either. It would be nice to see some more benchmarks. Z*Net # 523 may be ftped from terminator.cc.umich.edu. -- "Perhaps I'm commenting a bit cynically, but I think I'm qualified to." - Dan Bernstein