Hello, A new project on the LCRC cluster has been requested. Please forward the information on to the LCRC Allocation sub-committee. Applicant's name: Hemant Sharma Applicant's institution: ANL Applicant's division: XSD-MPE Project Name: far_field_hedm Project title: Realtime analysis of far-field high energy diffraction microscopy (HEDM) data Associated funding: Other Systems: orthros cluster at APS using 384 cores Science: Far-field High Energy Diffraction Microscopy (ff-HEDM) is a novel technique for non-destructive characterization of structure of polycrystalline materials using X-Rays. The brilliant x-rays produced at 3rd generation synchrotron sources, including Advanced Photon Source (APS), enable scientists to look at the constituent structure of materials at the micrometer level. Furthermore, the data can be used to refine the crystallographic orientation and strain state of the sub-structure (so-called 'grains') within the materials. Such knowledge can enable material scientists and modellers to understand, predict and enhance the macroscopic properties of materials for use in a wide variety of applications ranging from jet engines to nuclear materials. This project aims to enable realtime and post-mortem analysis of such diffraction data generate at beamline 1-ID of APS. Project description: We have developed a very efficient data-reduction algorithm for realtime analysis. The first step of computation involves the pre-processing of raw-diffraction images to reduce the size of the data and to find the diffraction peaks. The second step involves 'indexing' (grouping) the peaks belonging to the same 'grain' and then optimization of the centre-of-mass position, crystallographic orientation and strain state of each grain in the investigated volume of the sample. At the 1-ID beamline, we can investigate a sample volume every 3 minutes consisting of between 360 to 1440 diffraction images of 8MB each in size. The package is implemented in C99 and uses only the NLOPT package as a dependency. Parallelization at each stage is achieved using the SWIFT programming language. This allows for ease of implementation and scalability. The software is being used routinely on Orthros (the local HPC at APS) using 384 cores and has achieved upto 90% efficiency. Once we have finished testing (500 hours), we expect a different user every week to process data. The core team would be of 4 people involved in data analysis. Industry partnership: We have GE-Global Research, Air Force Research Laboratory as our main industrial partners and many more of our users from academia and industry will make use of the computation power. Project URL: Requested allocation: 980400 Q1: 230400 Q2: 250000 Q3: 250000 Q4: 250000 Justification: The software used to take about a month on a single CPU to run once and this time has been brought down to 3 minutes using the following improvements: 1. We have used the SWIFT parallel scripting language for parallelization. Even on the level of a single diffraction image, we parallelize the computation by 4x to achieve speedup. Furthermore, each grain is processed in parallel with no inter-process communication. 2. All the programming is done in C99. 3. The data volume is reduced from about 4-10 GB to about 100 MB. This has significantly eased data I/O problems. Storage requirements: 2 TB of data generated over the course of a day should be stored temporarily. We will accordingly clean up as new data is generated. The requester has used undetermined amount hours of their initial startup project. In addition to approving an initial amount, please specify a Category and Subcategory for this project. For a list of the current selection of approved categories, please see: https://wiki.lcrc.anl.gov/wiki/Processes/Categories Once the Allocation committee has approved the project, please go to the Project Management page to create it: https://accounts.lcrc.anl.gov/projects.php Thank You, The LCRC Accounts System