DO NOT REPLY TO THIS EMAIL, BUT PLEASE POST YOUR BUG RELATED COMMENTS THROUGH THE WEB INTERFACE AVAILABLE AT <http://nagoya.apache.org/bugzilla/show_bug.cgi?id=16906>. ANY REPLY MADE TO THIS MESSAGE WILL NOT BE COLLECTED AND INSERTED IN THE BUG DATABASE.
http://nagoya.apache.org/bugzilla/show_bug.cgi?id=16906 Add threadCount to Parallel task Summary: Add threadCount to Parallel task Product: Ant Version: 1.6Alpha (nightly) Platform: Other OS/Version: Other Status: NEW Severity: Enhancement Priority: Other Component: Core tasks AssignedTo: [EMAIL PROTECTED] ReportedBy: [EMAIL PROTECTED] As part of our build process sometimes we need to run large data processing tasks that can take hours, so we put it on a beefy multi-cpu box and let it run overnight. The problem is that unless we are content to use one CPU or make one seperate sequenctial portion for each CPU it's no more efficint that running it on a desktop. Running them all under the same parallel causes huge memory consumption beyond physical memory and leads to massive disk thrashing. The attached patch allows the user to specify the maximum number of tasks to execute at once, and an optional multiplier per processor (including jdk 1.4 introspection tricks and a fallback). A smoke and (a fragile) ordering test is also included. This will by no means create for perfect execution time, since the duration of the execution of the tasks is an unknown and dorting those for maxamum parallel usage is a NP complete knapcask variant. But it does make for easier tweaking, expecially when the target build machine is moved.
