D2K Performance and Batch Mode

Subject:

Comments:
: I am not going to have time to look at this today, but I will try to look : into it tomorrow. A few thoughts: are you using ssh to log into the : machines in your lab? That slows things down considerably, as all data : that passes between the machines must be encrypted (or decrypted). Use : telnet instead. Is this a memory-intensive itinerary? If so, you should : set the initial and maximum heap sizes for the jvm to be large, because : incrementing the heap size at runtime is very, very slow. : To run d2k in "batch mode" try this (the -ms and -mx are optional): : java -ms50M -mx100M -classpath : LocalController.jar:jaxp.jar:parser.jar:swingall.jar : ncsa.d2k.infrastructure.ComLineApp <itinerary name> : This will execute your itinerary without showing the d2k window. : At 10:47 AM 9/27/00 -0500, Laura Kruse wrote: : >Mr. Clutter, : > : >I'm am working in Dr. Hsu's group. We have been running some : >performance test's on the K2 itinery that Haipeng made. We run into : >some interesting results. On my personal computer, it runs in about : >30-40 minutes, running both the Standalone Version and running it in : >D2k. Running it on the computers in our lab tends to take anywhere from : >3 hours to 12. (3 Topeka, 7-8 Salina, 12 Yavanna) This seems to give : >interesting results. Dr. Hsu asked me to have you at it to see if you : >could notice something. I have attached the Java Source and the : >iternary that was built. : > : >Also could you explain how d2k batch mode works. Thanks. : > : >-- : >Laura Kruse : >509 Goodnow Hall : >Manhattan, KS 66506-2000 : >(785) 395-2128

Optional Link URL:
Link Title:
Optional Image URL: