Large parallel systems for QCD
Large parallel systems for QCD
- Large parallel systems optimized for LQCD
- L. Tripiccione in the plenary, Columbia project (A. Gara), APEmille (P. Vicini, E. Panizzi), CP-PACS (A. Ukawa)
- Continuos effort since mid ‘80s
- Goal: top computing performance
- 3rd generation now:
- CP-PACS 614 GFLOPS completed Oct ‘96
- QCDSP 400-800 GFLOPS fall ‘97
- APE mille 504-1081 GFLOPS spring ‘98