...
- Imprecise or incomprehensible estimation of requested computational resources (especially NPL/core-hours); for example missing arguments to justify a certain number of N runs instead of a smaller number. One page is our recommended minimum.
- No proof, that the software to be used is suitable and efficient for parallel execution (and parallel I/O) on our specific supercomputer. Recycling a scalability proof by a third party is meaningless, without showing that your planned production run is fully comparable to it (algorithm selection within the software, I/O pattern, machine architecture, problem size per core).
- The overall aim and/or motivation behind the project is unclear.
- The applicant lacks HPC/Unix skills and an experienced co-applicant is missing.
- Insufficiency of the applicant's local resources is not indicated.
- The HLRN was not mentioned in relevant publications.
- Cut & paste of previous proposals.
...