Jikes had a couple of problems with NetRexx files: first it didn't
handle a line of the form /* .. *//* .. */ (as generated by NetRexx if -format is not specified) and second it didn't handle names with a '$' in them. Both of these have been fixed; the first was a simple bug; the latter trickier -- I believe that you need to specify a '-$' option to make this work. If you don't like that, I'm sure they'd appreciate your feedback. On HPJ -- I'm told that it gives a very useful speedup when running NetRexxC. - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Mike Cowlishaw, IBM Fellow, IBM UK Laboratories mailto:[hidden email] [http://www2.hursley.ibm.com] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ To unsubscribe from this mailing list ( ibm-netrexx ), please send a note to [hidden email] with the following message in the body of the note unsubscribe ibm-netrexx <e-mail address> |
Let's see, without JIT compiler Java runs 20+ times slower than
native. With JIT it runs 3-5 times slower, or so. With Jikes are we there, or nearly so? And someday could we get a JIT implemented in hardware, for example something to plug into one of those USB ports? - Tony Dahlman BTW please accept apologies for my previous multiple post to this group. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ To unsubscribe from this mailing list ( ibm-netrexx ), please send a note to [hidden email] with the following message in the body of the note unsubscribe ibm-netrexx <e-mail address> |
In reply to this post by Mike Cowlishaw-2
>Jikes had a couple of problems with NetRexx files: first it didn't
>handle a line of the form /* .. *//* .. */ (as generated by NetRexx >if -format is not specified) and second it didn't handle names with >a '$' in them. > >Both of these have been fixed; the first was a simple bug; the latter >trickier -- I believe that you need to specify a '-$' option to make >this work. If you don't like that, I'm sure they'd appreciate your >feedback. > I'll go track down a copy of Jikes. Thanks for the info, Mike. >On HPJ -- I'm told that it gives a very useful speedup when running >NetRexxC. > The speed enhancements are quite noticable. However, if it a fast OS/2 native app that I want, I'd have written it in a different language. Java via NetRexx has the BIG advantage of running on EVERYTHING that has a JVM... -- /-------------------------------------\ | Jerry McBride | | | | ([hidden email] | \-------------------------------------/ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ To unsubscribe from this mailing list ( ibm-netrexx ), please send a note to [hidden email] with the following message in the body of the note unsubscribe ibm-netrexx <e-mail address> |
In reply to this post by Dr Tony Dahlman
>Let's see, without JIT compiler Java runs 20+ times slower than
>native. > >With JIT it runs 3-5 times slower, or so. > Some of the apps that I work with run almost on par with native code... The March 4th update really speeds things along... >With Jikes are we there, or nearly so? > Well, Jikes is a Java Code to Java Class compiler, it has nothing to do with running performance... >And someday could we get a JIT implemented in hardware, for >example something to plug into one of those USB ports? > Now that IBM has licensed the JavaChip technology from Sun, I'd say it's very possible that you'll see it turn up somewhere. On a PC card ( ala' JAVABLASTER) I don't know, but I'd buy a dozen... no questions asked... -- /-------------------------------------\ | Jerry McBride | | | | ([hidden email] | \-------------------------------------/ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ To unsubscribe from this mailing list ( ibm-netrexx ), please send a note to [hidden email] with the following message in the body of the note unsubscribe ibm-netrexx <e-mail address> |
Free forum by Nabble | Edit this page |