Frequency-response masking (FRM) technique produces a filter network which comprises several sub-filters with very sparse coefficient values. If the sub-filters are optimized ...
Abstract. Loop optimizations such as loop unrolling, unfolding and invariant code motion have long been used in a wide variety of compilers to improve the running time of applicati...
Nowadays, key characteristics of a processor's instruction set are only exploited in high-level languages by using inline assembly or compiler intrinsics. Inserting intrinsic...
Exponential increases in architectural design complexity threaten to make traditional processor design optimization techniques intractable. Genetically programmed response surface...
Recent superscalar processors issue four instructions per cycle. These processors are also powered by highly-parallel superscalar cores. The potential performance can only be expl...
Thomas M. Conte, Kishore N. Menezes, Patrick M. Mi...