The hardware resource utilizations of reEngine (the one given in L2 test as an example on U200) is shown in the table below (performance optimized version at FMax = 200MHz).
Item | LUT | REG | BRAM | URAM | DSP |
reEngine (U200) |
499292 | 341196 | 792 | 576 | 36 |
54.94% | 17.27% | 70.53% | 60.00% | 0.53% |
Number of PUs on each SLR is listed in the table below:
SLR | Number of PUs |
0 | 5 |
1 | 2 |
2 | 5 |
Therefore, the kernel throughput should be:
Throughput = 12 * 387 MB/s = 4.64 GB/s