Q-A No 3 NOAA 2011-1 dtd 12-7-2010.pdf

NOAA 2011-1 Small Business Innovative Research Questions and Answers #3

December 7, 2010

Q: Research Topic 8.1.4F: Is this topic strictly for marine fish or can it be extended to freshwater fish?

A: NOAA’s research is more in the marine field while the USDA conducts research in freshwater aquaculture.

Q: Research Topic 8.1.4F: We have a process which has a patent and can produce a certain product in a laboratory setting. We would like to write the SBIR to do product development for this, such as testing production on a larger scale, testing production feasibility, economics, etc.

Would this be appropriate for the Phase I SBIR or does our development put us beyond that state already?

A: Based on the information provided, your proposed project sounds like feasibility has been determined and you are in the production phase which would not make it eligible for Phase I. In accordance with the solicitation under paragraph 1.2.1, Phase I’s purpose is “to determine the technical feasibility of the proposed research and the quality of performance of the small business concerning receiving an award…”

Q: Research Topic 8.3.5D: Does NOAA seek to have a particular “fast radiative-transfer model” that is currently being used for assimilation of hyperspectral radiance data in a NWP to run on a GPU? If so which one?

A: Yes, we utilize the community radiative transfer model (CRTM) which uses OPTRAN and OPDEP. Information on CRTM and OPTRAN is in the public literature.

Q: Research Topic 8.3.5D: Will NOAA be providing any government furnished information (GFI) for the SBIR project? Specifically, will source code for the “fast radiative transfer model” or sample hyperspectral radiance data being assimilated into a NOAA NWP be provided as GFI?

A: Yes, the government can provide both if requested in the contractor’s proposal and accepted by the government.

NOAA 2011-1 Small Business Innovative Research Questions and Answers #3

Q: Research Topic 8.3.5D: The successful speed up of any source code on GPU is heavily dependent on how the CPU algorithm is written. Given the severe resource constraints of a Phase I project, would dramatic speed up of a piece of the algorithm on a GPU provide a sufficient metric for proof of concept for Phase II funding?

A: Speed would need to be demonstrated for the channel set used by NWP (which is about 300 channels mainly between 650 to 1100 cm-1 and 1200 to 1500 cm1. This subset can be provided if requested in the contractor’s proposal and accepted by the government.

Q: Research Topic 8.3.5D: Does NOAA have specific run time and/or resource restrictions that are driving the move to GPUs? If so, what are these restrictions?

A: The reason why only 300 channels or so are being used is because we do not have computational power to use all 2400 channels (AIRS), 8431 channels (IASI). So we are interested in other approaches such as GPU.

Q: Research Topic 8.3.5D: Have you already explored using GPUs to speed up a fast radiative-transfer model?

A: No.

Q: Research Topic 8.3.5D: Have you funded similar SBIR projects in this topic area in the past?

A: No.