Treffer: Comparative analysis of CPU and GPU performances on matrix operations using OpenCL Language

The proposed research goal is to introduce a new architecture for systems to increase performance and throughput for data intensive applications while reducing power consumption and execution time. Taking advantage of the proposed architecture, GPU that adapts to the dynamic environment and requests will be developed. The proposed architecture is scalable and modular to maximize performance and lower power consumption. GPU has drastically progressed in parallelism and executes multiple requests much faster than a general purpose CPU. It is feasible to use the GPU instead of introducing a multi-core CPU. The architecture of the system is important for data-intensive applications and large file size. The proposed research goal is to introduce a new architectural concept for systems on large scale to increase bandwidth, performance, and throughput for data intensive applications while reducing power consumption.