SOLICITATION NOTICE
A -- Parallelization of a Two-Dimensional Hydraulic Model to Run on Multi-Core PCs
- Notice Date
- 6/15/2009
- Notice Type
- Presolicitation
- Contracting Office
- BUREAU OF RECLAMATION PO BOX 25007, 84-27810 DENVER CO 80225
- ZIP Code
- 80225
- Solicitation Number
- 09SQ810248
- Response Due
- 6/26/2009
- Archive Date
- 6/15/2010
- Point of Contact
- BILL HOLBERT CONTRACT SPECIALIST 3034452448 bholbert@do.usbr.gov;<br />
- E-Mail Address
-
Point of Contact above, or if none listed, contact the IDEAS EC HELP DESK for assistance
(EC_helpdesk@NBC.GOV)
- Small Business Set-Aside
- N/A
- Description
- THIS IS A SOURCES SAUGHT ANNOUNCEMENT: The Bureau of Reclamation is seeking experienced researchers or scientists to implement the existing Reclamation numerical model, SRH-2D, to run efficiently on a multi-core desktop PC and be applicable to practical river flows. An overview and statement of work follow, along with the mandatory requirements for this solicitation. Please contact Bill Holbert, Contract Specialist for the Bureau of Reclamation, at 303-445-2448 or wholbert@usbr.gov if you can meet these mandatory requirements 100%. Parallelization of a Two-Dimensional Hydraulic Model to Run on Multi-Core PCsA. OVERVIEW The goal of this effort is to develop a parallelization methodology and then implement an existing two-dimensional (2D) depth-averaged hydraulic flow model (SRH-2D) to run efficiently on multi-core desktop personal computers with Linux and Window based operating systems. At present, a desktop computer is equipped with quad-core processors with reasonable costs. More processors are being packaged into a single desktop system. However, most existing 2D hydraulic models do not take advantage of the multi-processor feature of the PCs though they have been routinely applied to river projects at Reclamation. A single processor is used to run the model at present and application often is limited to a reach less than 10 miles. There is a strong need at Reclamation to apply the 2D model to longer river reach (e.g. 100 miles). Reclamation would like to resort to parallel computing, on the multi-core PCs, to achieve the goal. B. OBJECTIVE OF REQUEST FOR PROPOSALS The objective of this RFP is listed below: Reclamation seeks experienced researchers or scientists to implement the existing Reclamation numerical model, SRH-2D, to run efficiently on a multi-core desktop PC and be applicable to practical river flows. SRH-2D is a two-dimensional (2D) depth-averaged hydraulic flow model which is specifically designed to simulation open channel flows. A few unique features which may impact the parallelization effort are listed below: "It is a finite-volume based model with sequential flow solver;"It is an implicit solver which resorts to a liner equation matrix solver (conjugate gradient, CG, type of solvers are used); therefore, an efficient parallel solver of the linear equation using the CG method is critical; and "It uses an unstructured mesh with mixed triangular and quadrilateral cells. Therefore, a domain decomposition strategy is needed which is applicable to complex flow domains; and"It is written with FORTRAN 90. Mandatory qualifications of the principal investigator (PI) for achieving the project objective are listed below: "The PI must have an extensive experience in the numerical model development either in the computational fluid dynamics (CFD) area or in the computational hydraulics (CH) area so that SRH-2D source code is familiar to him or her;"The PI must have prior experience in implementing an implicit fluid flow solver to run on multi-core desktop PCs. Experience in the development of the parallel version of CG (Conjugate Gradient) type linear equation solver, i.e. parallelization of an implicit model, is a must qualification; and"The PI should be fluent in FORTRAN 90. Those who do not meet the above qualifications will not be considered. C. STATEMENT OF WORK (SOW) The first year of this project has a committed fund of $50,000; and the first year work scope includes the following four tasks: 1. Task 1: Development of the Parallelization StrategyA general strategy is to be laid out first with regard to the parallelization of SRH-2D, such as what methods are selected and why. The strategy includes the domain decomposition method suitable for the unstructured mesh, the selection of the parallel language (MPI or OPEN-MP), operating system (Linux or Window), required software and hardwire needed, etc. 2. Task 2: Parallelization of SRH-2D ModelThis task focuses on the development of a parallel version of SRH-2D in FORTRAN 90. The source code of SRH-2D written in FORTRAN 90 will be available. A number of parallel issues are considered in this task including: a profile analysis to list the most time consuming routines (procedures), parallel linear equation solver with CG type solver or equivalent, domain decomposition issue, etc. 3. Task 3: Test and Demonstration of Parallelization Performance A sample case related to a real river flow will be provided to the PI by Reclamation to test, measure, and demonstrate the success of the parallelization effort. Results on 2-, 4-, and 8-nodes are anticipated. A successful parallelization of SRH-2D will be measured by two factors: (a) the model gives the correct results in comparison with the serial version of the model; and (b) the speed-up obtained on the multi-core PC. 4. Task 4: Documentation, Deliverables, and AssistanceA final report and the parallel version of SRH-2D will be the deliverables at the end of the first year effort. This task also includes an installation assistance by the PI so that the parallel SRH-2D may be run on Reclamation multi-core PCs. The proposer has to decide whether the first find limit ($50,000) is enough to accomplish the above four tasks. Higher funding request will not be considered. The second and third year work is contingent on the success of the first year results and is not guaranteed. The estimated available fund is about $25,000 per year and the work scope is listed below: 5. Task 5: Model Improvement and Refinement (Second Year Option)The second year task is to focus on rewriting SRH-2D model to make it more efficient for parallel computing. It may be achieved by a number of possible ways, but it envisioned change is the development of new solution algorithm. A final report will be delivered documenting the changes made to SRH-2D and the improvements in computing speed achieved. 6. Task 6: Parallelization of Sediment Transport ModuleThe third year task focuses on the parallelization of the sediment transport module, in addition to the flow model completed in the first year. A final report and the parallel version of the sediment transport module integrated into SRH-2D will be the deliverables.
- Web Link
-
FBO.gov Permalink
(https://www.fbo.gov/spg/DOI/BR/BR/09SQ810248/listing.html)
- Place of Performance
- Address: Denver Federal Center, Denver, CO 80225<br />
- Zip Code: 802250007<br />
- Zip Code: 802250007<br />
- Record
- SN01845091-W 20090617/090615235053-87949d861759b5ced01efeca96774c44 (fbodaily.com)
- Source
-
FedBizOpps Link to This Notice
(may not be valid after Archive Date)
| FSG Index | This Issue's Index | Today's FBO Daily Index Page |