Research Article
Transparent Runtime Migration of Loop-Based Traces of Processor Instructions to Reconfigurable Processing Units
Table 3
Communication overhead.
|
Kernels
| Communication overhead | #Inst. of PLB CR | #Inst. of FSL CR | DDR-PLB (%) | LMB-PLB (%) | LMB-FSL (%) |
| count | 27 | 12 | 92.4 | 56.39 | 26.22 | even_ones | 34 | 18 | 92.1 | 62.25 | 30.49 | fibonacci | 27 | 14 | 63.2 | 15.58 | 4.22 | ham_dist | 35 | 17 | 91.8 | 61.18 | 29.98 | pop_cnt32 | 35 | 17 | 92.1 | 61.34 | 30.49 | reverse | 35 | 17 | 92.3 | 61.34 | 30.49 |
| compress | 35 | 19 | 95.0 | 71.82 | 39.97 | divlu | 25 | 10 | 92.2 | 53.79 | 26.83 | expand | 35 | 19 | 95.0 | 71.76 | 40.75 | gcd | 32 | 15 | 77.3 | 34.51 | 12.81 | isqrt | 34 | 16 | 96.2 | 74.77 | 45.52 | maxstr | 25 | 10 | 92.5 | 55.25 | 26.89 |
| popcount3 | 37 | 18 | 46.97 | 8.16 | 2.77 | mpegcrc | 36 | 20 | 85.5 | 47.00 | 21.69 | usqrt | 31 | 18 | 89.9 | 59.56 | 28.93 |
| merge1 | 56.3 | 22.5 | 87.0 | 58.14 | 17.07 | merge1 (n/s) | 32.17 | 15.8 | N/A | 41.94 | 16.13 | merge2 | 57 | 22.0 | 89.6 | 70.32 | 24.43 | merge2 (n/s) | 31 | 14.8 | N/A | 48.04 | 21.26 |
|
|