Abstract:
Method and system for more efficient checkpointing strategy in cross correlating (316) a forward (328) and backward (308) propagated wave such as in migrating (326) or inverting seismic data. The checkpointing strategy includes storing in memory forward simulation data at a checkpointed time step, wherein the stored data are sufficient to do a cross correlation at that time step but not to restart the forward simulation. At other checkpoints, a greater amount of data sufficient to restart the simulation may be stored in memory (314). Methods are disclosed for finding an optimal combination, i.e. one that minimizes computation time (1132), of the two types of checkpoints for a given amount of computer memory (1004), and for locating a checkpoint at an optimal time step (306, 1214, 1310). The optimal checkpointing strategy (1002) also may optimize (1408) on use of fast (1402) vs. slow (1404) storage.
Abstract:
A method for seismic modeling on a GPU includes creating a discretized model containing a volume enclosed by boundaries with arbitrary boundary conditions, dividing the discretized model into subsets, transferring a data to the GPU, initializing the data stored in a selected number of the subsets on the GPU, performing computation related with a number of time steps for the selected number of the subsets on the GPU, and substantially simultaneously computing, off-loading a finished subsets data from the GPU, and up-loading a new subsets data to the GPU. The provided method and corresponding system can reduce the amount of data stored in GPU memory by an order of magnitude, minimize the GPU device-to-host transfer bandwidth requirements, and reduce the GPU setup time.
Abstract:
Method and system for more efficient checkpointing strategy in cross correlating (316) a forward (328) and backward (308) propagated wave such as in migrating (326) or inverting seismic data. The checkpointing strategy includes storing in memory forward simulation data at a checkpointed time step, wherein the stored data are sufficient to do a cross correlation at that time step but not to restart the forward simulation. At other checkpoints, a greater amount of data sufficient to restart the simulation may be stored in memory (314). Methods are disclosed for finding an optimal combination, i.e. one that minimizes computation time (1132), of the two types of checkpoints for a given amount of computer memory (1004), and for locating a checkpoint at an optimal time step (306, 1214, 1310). The optimal checkpointing strategy (1002) also may optimize (1408) on use of fast (1402) vs. slow (1404) storage.
Abstract:
A method for seismic modeling on a GPU includes creating a discretized model containing a volume enclosed by boundaries with arbitrary boundary conditions, dividing the discretized model into subsets, transferring a data to the GPU, initializing the data stored in a selected number of the subsets on the GPU, performing computation related with a number of time steps for the selected number of the subsets on the GPU, and substantially simultaneously computing, off-loading a finished subsets data from the GPU, and up-loading a new subsets data to the GPU. The provided method and corresponding system can reduce the amount of data stored in GPU memory by an order of magnitude, minimize the GPU device-to-host transfer bandwidth requirements, and reduce the GPU setup time.