Timothy Farrar has some interesting ideas embedded in this recent post describing the design guidelines driving the development of his GPGPU OS. After thinking about it a bit, I've extracted these rules from his post:
- Code-to-test iteration should be instant
- Built-in memory, cycle, and bandwidth performance monitoring should be provided
- Save and restore of current data and execution state should be possible
- Inputs and outputs should be explicit and naturally limited to valid addresses
- High performance data structures should have natural representations
- Parallelism and dataflow should be explicitly supported
Clearly others are thinking on the same lines; functional languages are another approach these rules would work with naturally.