Message boards :
Number crunching :
The MT and non-MT app binaries are exactly the same?
Message board moderation
Author | Message |
---|---|
Send message Joined: 8 Jun 10 Posts: 4 Credit: 89,727,552 RAC: 49,480 ![]() ![]() ![]() |
In the applications page, we have 1.87 and 1.87 (mt) for Milkyway@home N-Body Simulation with Orbit Fitting. On the host, I have two binaries, milkyway_nbody_orbit_fitting_1.87_x86_64-pc-linux-gnu and milkyway_nbody_orbit_fitting_1.87_x86_64-pc-linux-gnu. A "diff" comparison shows me that they are exactly the same. Similar situation on Windows as well, except the different binary names. I noticed this because I have <plan_class>mt</plan_class> for my config, but some tasks still run single thread. The single thread tasks can become multi-thread just fine if I remove that plan_class line. I wonder what's the point of having two exactly same application but different plan class then? Would the project care if I run the non-mt plan class tasks with multiple threads too? |
![]() Send message Joined: 19 Jul 10 Posts: 650 Credit: 19,514,602 RAC: 1,042 ![]() ![]() |
Yes, they are the same and you simply choose the number of threads per task in your MilkyWay@home preferences, no need for any config files. Best performance you get with 1 thread per task. ![]() |
Send message Joined: 8 Jun 10 Posts: 4 Credit: 89,727,552 RAC: 49,480 ![]() ![]() ![]() |
Thanks. The website preference says "Max # of threads", so I was not sure if it will always set the number I selected. Let me give it a try. My main motivation to configure the threads is to keep the longest runtime in control, without too much efficiency loss. I mostly need only 2-4 threads to achieve my goal. |
©2025 Astroinformatics Group