TABLE 5.
Estimated optimal MFLUPSmaxfrom LBM roofline performance model for a V100 GPU using direct addressing and equation 25.
Pattern | Bytes/FLUP (B/F) | D2Q9 B/F | D2Q9 Roofline |
---|---|---|---|
AB | 2Q*double | 144 | 6250 |
MR | 2M*double | 96 | 9375 |
Estimated optimal MFLUPSmaxfrom LBM roofline performance model for a V100 GPU using direct addressing and equation 25.
Pattern | Bytes/FLUP (B/F) | D2Q9 B/F | D2Q9 Roofline |
---|---|---|---|
AB | 2Q*double | 144 | 6250 |
MR | 2M*double | 96 | 9375 |