An empirical study of the CRAY Y-MP processor using the PERFECT club benchmarks
Journal
Conference Proceedings - Annual Symposium on Computer Architecture
Pages
170-179
Date Issued
1991
Author(s)
Abstract
An empirical study of a single processor of the CRAY Y-MP, using as benchmarks long-running scientific applications from the PERFECT Club benchmark set, is reported. Since the compiler plays a major role in determining machine utilization and program execution speed, the authors compile their benchmarks using the state-of-the-art Cray Research production Fortran compiler. They investigate instruction set usage, operation execution counts, sizes of basic blocks in the programs, and instruction issue rate. The vectorized fraction of the dynamic program operation count ranges from 4% to 96% for the benchmarks. Instructions that move values between the scalar registers and corresponding backup registers form a significant fraction of the dynamic instruction count. Basic blocks which are more than 100 instructions in size are significant in number; both small and large basic blocks are important from the point of view of program performance. The instruction issue rate is less than 0.5 instructions per clock cycle for each of the benchmark programs, suggesting that the instruction issue stage of the CRAY Y-MP is, on the average, not a bottleneck.
Other Subjects
Computers, Supercomputer--Performance; Benchmarking; CRAY Y-MP; Performance Evaluation; Computer systems, Digital
Type
conference paper