General observations on all KAW-applications

General observations

1. All applications are well-written and request resources for competitive research projects.
2. Several of the applications propose a resource that uses close to the maximum budget for investments. It is not always clear from those applications why a facility of exactly this size is needed.
3. Several of the applications request resources for consortia with a rather large number of groups in a span of different research fields. It is not always clear that the system proposed would be optimal for covering the needs within all the fields described in the application. It would presumably be more cost-efficient to provide resources on a variety of systems.
4. Some of the applications describe resources that may end up being shared resources similar to those provided via SNAC. In general, these applications do not describe how the allocations on the proposed resources should be determined. An alternative to establishing dedicated resources for these consortia would be that SNAC imposes clearer priorities (provides larger allocations) for specific user groups, possibly also including dedicated resources for user support. The new SNAC policy with “medium-size” and “large” allocations is a step in this direction. Also, the information from the SNIC/KAW application round will be forwarded to the SNAC committee.
5. A few applications mention that the current SNIC facilities are insufficient. It would have been useful if these projects would have provided more evidence for this. For example, one of the applications state that today it is difficult for researchers to run applications on the SNIC resources that require many CPUs or long execution times. However, details about the types of jobs that that do not get the right priorities on the SNIC resources are not provided.
6. Often, the needs for resources that the applicants described in recent SNAC allocation rounds do not conform to the needs described in the corresponding SNIC/KAW application. It is quite important for SNIC that a consistent view of user requirement is available, e.g. for planning requests for funding, investments and user supports.
7. The running costs presented in the applications are in many cases unrealistic. The costs are often apparently based on peak-power numbers (and in some cases even higher), and the power consumption will be significantly lower in practice. Also, aspects of recycling of energy, corresponding to reduced costs, are not discussed.