| We run the parallel ray-tracing application on JESSICA2 using 64 Pentium 4 PCs from the Gideon cluster as shown in the Figure below. We created 64 worker threads in the master node and start up 63 worker nodes to join the parallel execution. The 63 worker threads are migrated to the other 63 different worker nodes automatically when the nodes register to the master node. The network connection of the worker threads are continued through our I/O redirection mechanism. For 64 nodes, we obtained the speedup of 40.75. This is about 63.7% efficiency. |
