Skip to main content
Figure 8 | BMC Bioinformatics

Figure 8

From: Fine-grained parallel RNAalifold algorithm for RNA secondary structure prediction on FPGA

Figure 8

The master-slave PE array algorithm. The first part of Figure 8 shows the parameter and variable definitions. At the initial phase, each PE assigns its PE identifier, PID, to the column index indicating the initial column assignment. Before the energy calculation phase several data items are loaded into local memory as shown in S21. Only the master PE loads data from SDRAM and the V cache, the other slave PEs only load data from Trans Regs of the previous adjoining PE. When the energy data are ready in local memory, the free energy calculation is performed as shown in S22. From S23 to S25, all PEs transfer data received from the previous adjoining PEs to their own Trans Regs, store results to local memory, and move the computation point upward in the same column. When the computation point moves to the top of the current column, the PE enters the column synchronization phase, during which it writes its local results to SDRAM. When all PEs arrive at the synchronization points, indicating the end of the current section, the section advances by each PE adding p to its column index. The process repeats until the column index is greater than the RNA sequence length.

Back to article page