MEMORY HIERARCHY DESIGN FOR CHIP MULTIPROCESSORS

This book presents memory hierarchy design and data
access management methodology. First, memory
hierarchy design and data allocation problem is
discussed, followed by a compiler-driven approach to
data compression for reducing memory space
occupancy. Third, compiler supported loop-data
optimization problem to improve locality of data
accesses for a given application code is discussed.
Since parallelization is a key concept in achieving
the best performance, a constraint network based
formulation has been discussed in the fourth
chapter. In the last part, book discusses advanced
techniques such as the memory optimization in a 3D
architecture to minimize data access costs under
temperature.

Autorentext
Özcan Öztürk is an assistant professor in the Computer
Engineering Department at Bilkent University. Prior to joining
Bilkent, he worked as a software engineer at Intel(Marvell). He
is the author of more than 60 papers. Dr. Öztürk received his
Ph.D. degree from the Pennsylvania State University and M.S.
degree from University Of Florida.

Klappentext
This book presents memory hierarchy design and data access management methodology. First, memory hierarchy design and data allocation problem is discussed, followed by a compiler-driven approach to data compression for reducing memory space occupancy. Third, compiler supported loop-data optimization problem to improve locality of data accesses for a given application code is discussed. Since parallelization is a key concept in achieving the best performance, a constraint network based formulation has been discussed in the fourth chapter. In the last part, book discusses advanced techniques such as the memory optimization in a 3D architecture to minimize data access costs under temperature.