Marketing...

[rodrigo@xxxxxxxxxxxxxxxx (Rodrigo de Almeida Sodré)]

Some threads ago the link below was mentioned...

"TURBO-CHARGING CLIENT-SIDE PERFORMANCE"
http://www.javasoft.com/features/1999/11/j2se.html

In the article, among many other things, there is a comment about the JDK
1.3 client hotspot speed:

"The numbers are impressive. Compared to Sun's internal benchmarks for JDKTM
v 1.2.1 software, the new performance improvements deliver a 40-percent
reduction in start-up time, a 25-percent reduction in memory footprint
(about a 3-to-4 MB reduction for a typical JFC/Swing technology-based
application), and a 60-percent faster raw Java Virtual Machine speed."

So I executed the SwingSet example to catch some personal benchmarks on
start-up time and memory footprint, and the Metalworks example to benchmark
the speed. I used the same swing 1.1.1 version. There are the results:

JDK 1.1.6:
	Start-up:       ~10 seconds
	RAM:            ~17.3 MB
       Virtual Memory: ~16.5 MB

	Dragging JinternalFrame: 30-40% of the processor

JDK 1.2.2 (using hotspot 1.0.1):
	Start-up:       ~14 seconds
	RAM:            ~26.1 MB
       Virtual Memory: ~26.7 MB

	Dragging JinternalFrame: 40-50% of the processor

JDK 1.3:
	Start-up:       ~9 seconds
	RAM:            ~22.8 MB
       Virtual Memory: ~23.5 MB

	Dragging JinternalFrame: no JInternalFrame appeared

I think we could reconsider the bug with JInternalFrame since jdk 1.3 is
still on beta.

Conclusion: well, we can't deny they told the truth but they *forgot* to
compare jdk1.3 with 1.1.x. Its startup time is only 10% faster and it uses
12.5 MB more memory. Anyway, it was a good try.

Regards


Rodrigo A. Sodre
Analista de Sistemas
Foton Informatica e Servicos
(061) 328-5060 R:243
rodrigo@xxxxxxxxxxxxxxxx

---
To unsubscribe, mail advanced-java-unsubscribe@xxxxxxxxxxxxxxxx
To get help, mail advanced-java-help@xxxxxxxxxxxxxxxx