Performance Analysis of Cache Memory in CPU

The architecture of a computer can be efficiently designed, by synchronizing the major sections of a computer namely the Central Processing Unit (CPU), memory, data path, pipeline structure, Arithmetic, and Logic Unit (ALU). The data storage and organization depend on the memory allotment and specific memory design carried out in accordance with the CPU specifications. A specific memory hierarchy is followed for the CPU, which starts with the units near the processor such as memory registers, cache memory, main memory, secondary memory, and finally the flash storage devices. This hierarchy is based on parameters like speed, response time, program complexity, and overall storage capacity for the computer. This work focuses on the cache memory which is a high-speed special memory, essential in the memory section of computer architecture. A modified version of the vector triad benchmark program is run on four computer systems with different specifications to check the performance and understand the cache levels. The problem size, i.e. the execution of the benchmark program is defined to get iterations and plotting comparison curves for time and throughput analysis for test cases of computers. As the level of cache changes from level 1 to level 3, the throughput curve decreases. Also, when more processes are active, while the benchmark code is running, then a significant rise in computation time, data access time, and total time values are observed which reflects on the computer performance.