Divisible Job Vortal

Animations

This part contains animations which show communication algorithms in chain (with and without communication coprocessors), star (with and without communication coprocessors), mesh and hypercube architectures (with communication coprocessors).

You can choose proper architecture by clicking on one of the hyperlinks. Chain and star parts contain two animations (with and without communication coprocessors). Mesh and hypercube parts contain one animation each.

The processors shown in the animations can be in the following states:

idle or sending data (yellow),
processing data (red).

Blue lines indicate communications.

Here is short description of every communication algorithm:

Chain without communication coprocessors: the originator sends data to the second processor and waits until all the load will be received. Then it starts to process data. This procedure is repeated until the last processor is activated. All processors stop computing at the same time.

Chain with communication coprocessors: the originator starts to process data and sends the load to the second processor in the same time. When all data is received the second processor starts to send and to process the load. This procedure is repeated until the last processor is activated. All processors stop computing at the same time.

Star without communication coprocessors: the originator sends data to all processors consecutively and waits until all the load will be received. Then it starts to process the local data. The other processors are activated when they receive all the load. All processors stop computing at the same time.

Star with communication coprocessors: the originator starts to process local data and sends the rest of the load to all processors consecutively. The other processors are activated when they receive all data. All processors stop computing in the same time.

Mesh with communication coprocessors: the originator starts to process local data and sends the load in four directions in the same time. When all data is received the activated processors start to process their share of the load and send the rest of the load to four more processors each. When that data is received all processors are active. The processors stop computing simultaneously.

Hypercube with communication coprocessors: the originator is the processor with number 000. It starts to process local data and sends the rest of the load to processors 001, 010, 100 (one 1 in the binary notation). After all data is received they simultaneously send data to processors 011,101; 011,110; 101,110 (two 1`s in binary notation). Then all three processors 011,101,110 send data to processor 111. All the processors stop computing at the same time.