Volano Benchmark Report It is kind of silly when the applications we wrote responded *significantly* faster on our development P3 desktops than on the actual production Solaris servers. Why the significant difference in performance? Sloppy implementations? [ April 02, 2003: Message edited by: Chu Tan ]
I don't know what apps your running. My dev box is 1Gb ram 1.8Ghz P4 and it is a WHOLE lot slower than our prod Sun boxes. Those tests are loopback tests also, not 'over the wire'.
Joined: Apr 19, 2002
I didn't test out respond time over maximum load, just the time it takes for tomcat to initialize stuffs on start up. The live tomcat servers need 30 sec after issueing command to start serving. On the local boxes, the same framework starts up in less than 10 sec. The framework is a combination of struts, and some in house framework. I also tried running the MapTest code from the other thread, our Sun hardware is about 50% of my 1 ghz P3. Which is quite reasonable, as the Sun's CPU is 550MHz. [ April 03, 2003: Message edited by: Chu Tan ]
Hi Everyone, My experience has been that Intel CPU's are about 50% slower than Sun CPU's in general. But, since intel is so much cheaper but you do get better OS with SUN and in the past I have seen Sun boxes stay working at high CPU utilization where WinTel dies which is very important in a 24x7 environment. Also, tuning the jvm and app server can lead to very significant performance improvements and well worth the time.