X-Ray Laser Imaging of Biomolecules Using Multiple GPUs
Extremely bright X-ray lasers are becoming a promising tool for 3D imaging of biomolecules. By hitting a beam of streaming particles with a very short burst of a high energy X-ray and collecting the resulting scattering pattern, the 3D structure of the particles can be deduced. The computational complexity associated with transforming the data thus collected into a 3D intensity map is very high and calls for efficient data-parallel implementations.
We present ongoing work in accelerating this application using multiple GPU nodes. In particular, we look at the scaling properties of the application and give predictions as to the computational viability of this imaging technique.
KeywordsGPU cluster CUDA/MPI Single molecule imaging X-ray laser
This work was financially supported by the Swedish Research Council, the Röntgen Ångström Cluster, the Knut och Alice Wallenbergs Stiftelse, the European Research Council (JL), and by the Swedish Research Council within the UPMARC Linnaeus center of Excellence (SE, JL).
- 3.Jacobsen, D.A., Thibault, J.C., Senocak, I.: An MPI-CUDA implementation for massively parallel incompressible flow computations on multi-GPU clusters. In: 48th AIAA Aerospace Sciences Meeting and Exhibit, vol. 16 (2010). doi:10.2514/6.2010-522Google Scholar
- 10.Wang, Y., Dou, Y., Guo, S., Lei, Y., Zou, D.: CPU-GPU hybrid parallel strategy for cosmological simulations. Concurr. Comput.: Pract. Exper. (2013). doi:10.1002/cpe.3046Google Scholar