|
|
Optimization of the Distributed I/O Subsystem of the k-Wave Project
Vysocký, Ondřej ; Klepárník, Petr (referee) ; Jaroš, Jiří (advisor)
This thesis deals with an effective solution of the parallel I/O of the k-Wave tool, which is designed for time domain acoustic and ultrasound simulations. k-Wave is a supercomputer application, it runs on a Lustre file system and it requires to be implemented with MPI and stores the data in suitable data format (HDF5). I designed three methods of optimization which fits k-Wave's needs. It uses accumulation and redistribution techniques. In comparison with the native write, every optimization method led to better write speed, up to 13.6GB/s. It is possible to use these methods to optimize every data distributed application with the write speed issue.
|
|
|
Optimization of the Distributed I/O Subsystem of the k-Wave Project
Vysocký, Ondřej ; Klepárník, Petr (referee) ; Jaroš, Jiří (advisor)
This thesis deals with an effective solution of the parallel I/O of the k-Wave tool, which is designed for time domain acoustic and ultrasound simulations. k-Wave is a supercomputer application, it runs on a Lustre file system and it requires to be implemented with MPI and stores the data in suitable data format (HDF5). I designed three methods of optimization which fits k-Wave's needs. It uses accumulation and redistribution techniques. In comparison with the native write, every optimization method led to better write speed, up to 13.6GB/s. It is possible to use these methods to optimize every data distributed application with the write speed issue.
|
|
|
Optimization of the Distributed I/O Subsystem of the k-Wave Project
Vysocký, Ondřej ; Hrbáček, Radek (referee) ; Jaroš, Jiří (advisor)
This thesis deals with an effective solution of parallel writing of variable amounts of data on the Lustre file system. The work will be used by the k-Wave project designed for time domain acoustic and ultrasound simulations. Since the simulation is computationally and data intensive, the project requires to be implemented with libraries for parallel computig (Open MPI) and large data processing (HDF5) and it must run on a supercomputer. The application is implemented in C and uses previously mentioned libraries. The proper settings of the Lustre file system leads to the peak write bandwith of 2.5 GB/s that corresponds to a speedup factor of 5 compared to the reference settings. The data aggregation improved the write bandwidth by a factor of 3 compared to a naive version. Here, the achieved I/O bandwidth for certain block sizes hits the limits of the Anselm I/O subsytem (3GB/s).
|
guest ::
