Table 2 Average execution time (ms) per sequence and speedup (Sp) for four sequence groups when different performance tuning methods are adopted gradually on GPU
From: CPU-GPU hybrid accelerating the Zuker algorithm for RNA secondary structure prediction applications
Opt. methods | A (L = 68) | B (L = 120) | C (L = 154) | D (L = 221) | ||||
|---|---|---|---|---|---|---|---|---|
Time | Sp. | Time | Sp. | Time | Sp. | Time | Sp. | |
Hierarchy parallelism | 0.452 | 1 | 2.402 | 1 | 3.530 | 1 | 20.067 | 1 |
Memory optimization | 0.199 | 2.27 | 0.855 | 2.81 | 1.561 | 2.26 | 8.876 | 2.26 |
Thread scheduling and tiling | 0.068 | 6.65 | 0.293 | 8.20 | 0.754 | 4.68 | 3.270 | 6.14 |