Title :
Numerical applications and sub-word parallelism: the NAS benchmarks on a Pentium 4
Author :
Daniel, Etiemble
Author_Institution :
Dept. of Electr. & Comput. Eng., Toronto Univ., Ont., Canada
Abstract :
We examine the impact of Pentium 4 SIMD instructions on the Fortran and C versions of the NAS benchmarks, either by compiler vectorization or by assembly code in-lining. If few functions generally profit from the SIMD operations, the ones using complex numbers or random number generators can be efficiently accelerated.
Keywords :
microprocessor chips; parallel processing; performance evaluation; program assemblers; program compilers; NAS benchmarks; Pentium 4 SIMD instructions; assembly code in-lining; compiler vectorization; random number generators; sub-word parallelism; Acceleration; Assembly; Benchmark testing; Costs; Microprocessors; Parallel machines; Parallel processing; Personal communication networks; Random number generation; Registers;
Conference_Titel :
High Performance Computing Systems and Applications, 2002. Proceedings. 16th Annual International Symposium on
Print_ISBN :
0-7695-1626-2
DOI :
10.1109/HPCSA.2002.1019156
