Efficient Partitioning Based Hierarchical Agglomerative Clustering Using Graphics Accelerators With Cuda

We explore the capabilities of today’s high-end Graphics processing units (GPU) on desktop computers toefficientlyperform hierarchical agglomerative clustering (HAC) through partitioning of gene expressions.Our focus is to significantly reduce time and memory bottlenecks of the traditional HAC algorithm byparallelization and acceleration of computations without compromising the accuracy of clusters. We usepartially overlapping partitions (PoP) to parallelize the HAC algorithm using the hardware capabilities ofGPU with Compute Unified Device Architecture (CUDA). We compare the computational performance ofGPU overthe CPU and our experiments show that the computational performance of GPU is much fasterthan the CPU. The traditional HAC and partitioning based HAC are up to 66 times and 442 times faster onthe GPU respectively, than the time taken by a CPU for the traditional HAC computations. Moreover, thePoP HAC on GPU requires only a fraction of the memory required by the traditional algorithm on theCPU. The novelties in our research includes boosting computational speed while utilizing GPU globalmemory, identifying minimum distance pair in virtually a single-pass, avoiding the necessity to maintainhuge data in memories and complete the entire HAC computation within the GPU.