Research Article
A Protein Sequence Analysis Hardware Accelerator Based on Divergences
Table 8
Related work and comparison with our solution.
| Reference | Number of PEs | Max. number of HMM nodes | Max. sequence size | Complete Plan7-Viterbi algorithm | Clock (MHz) | Performance (GCUPS) | Gain | FPGA |
| [3] | 68 | 544 | 1024 | N | 180 | 10 | 190 | Xilinx Virtex II 6000 | [13] | 90 | 1440 | 1024 | N | 100 | 9 | 247 | Xilinx 2VP100 | [11] | 50 | — | — | N | 200 | 5 to 20 | — | Not Synthesized | [14] | 72 | 1440 | 8192 | N | 74 | 3.95 | 195 | XC2V6000 | [15] | 10 | 256 | — | Y | 70 | 7 | 300 | XC3S1500 | [16] | 50 | — | — | N | 66 | 1.3 | 50 | Xilinx Spartan 3 4000 | [20] | 25 | — | — | Y | 130 | 3.2 | 56.8 | Xilinx Virtex 5 110-T |
| Our work | 85 | 2295 | 8192 | N | 67 | 5.8 | 254 (182
*
) | Altera Stratix II EP2S180F1508C3 |
|
|
*Including significant sequences reprocessing times.
|