Decomposition method for fast computation of gigapixel-sized Fresnel holograms on a graphics processing unit cluster

Abstract

A parallel computation method for large-size Fresnel computer-generated hologram (CGH) is reported. The method was introduced by us in an earlier report as a technique for calculating Fourier CGH from 2D object data. In this paper we extend the method to compute Fresnel CGH from 3D object data. The scale of the computation problem is also expanded to 2 gigapixels, making it closer to real application requirements. The significant feature of the reported method is its ability to avoid communication overhead and thereby fully utilize the computing power of parallel devices. The method exhibits three layers of parallelism that favor small to large scale parallel computing machines. Simulation and optical experiments were conducted to demonstrate the workability and to evaluate the efficiency of the proposed technique. A two-times improvement in computation speed has been achieved compared to the conventional method, on a 16-node cluster (one GPU per node) utilizing only one layer of parallelism. A 20-times improvement in computation speed has been estimated utilizing two layers of parallelism on a very large-scale parallel machine with 16 nodes, where each node has 16 GPUs.

Fast Computation with Efficient Object Data Distribution for Large-Scale Hologram Generation on a Multi-GPU Cluster

“…This enables us to reconstruct the original 3D object by time sequential reconstruction of sub-objects. We have proved that if the time sequential reconstruction is performed at an appropriate speed, the whole original 3D image is perceived [5], [6].…”

“…For the theoretical background and optical verification of the decomposition method as well as the preliminary results of its application to the Fourier and Fresnel hologram generation, see [5] and [6], respectively. Figure 4 shows the basic flow of the decomposition method that generates each sub-CGH from each sub-object.…”

“…To this problem, we have been working on a distributed algorithm, called a decomposition method, for generating 2D Fourier holograms [5] and 3D Fresnel holograms [6]. In order to estimate the performance of the method, we per-formed a simulation.…”

“…In parallel with the simulation, we have also developed a real GPU cluster using 8 nodes of multi-GPU machines, implemented the decomposition method for 3D Fresnel hologram generation on the cluster and applied application-oriented optimizations [19]- [21]. Thus, the major differences of this research from [6] are that this research utilizes a real multi-GPU cluster and applies various application-oriented, single-node and multinode optimizations to verify their effectiveness.…”

Review of fast methods for point-based computer-generated holography [Invited]

“…In any case, the time taken to generate a hologram of a 3D object for both approaches is still overwhelming. Although the computation time can be further decreased with special hardware, such as the graphic processing unit (GPU) [15][16][17], the speed improvement is insufficient to meet the video rate. In the following sections, we shall study some contemporary research works for fast generation of digital holograms.…”