System Scalability Driven Requirements
Overall System Scalability - Complex scientific applications such as radiation transport should achieve scaled parallel efficiencies greater than 70% on the full system (~20,000 processors).
- This implies the need for excellent interconnect performance, hardware and software.
- Overlap of communication and computation is difficult to achieve for most scientific codes.
Overall System Reliability - The usefulness of the system is strongly dependent on the time between interrupts.
- Ratio of calculation time to time spent checkpointing should be ~ 20 to 1 to make good progress.
- 100 hour MTBI is desirable