Abstract / Description of output
Difference between speed of processor and memory is increasing with advent of every new technology. Chip Multi Processors (CMP) have further increased the load on the memory hierarchy. So it has become important to manage on-chip memory judiciously to reduce average memory access time. The previous research has shown that it is better to have a shared cache at the last level of on-chip memory hierarchy. Sharing last level of cache gives rise to a new
category of cache misses; those were not present in uniprocessor, called “inter-processor misses”. This paper proposes a technique to eliminate interprocessor misses by giving replacement ownership of a block to a processor who brought
it into the cache. This reduction in interprocessor misses, which constitutes 40% of over all misses, will result in performance improvement. Also two different ways of relinquishing the ownership of a block are being proposed, so that if some other processor, other than owner, can make use of the block in a more
efficient way, ownership will be transferred to the new processor.
category of cache misses; those were not present in uniprocessor, called “inter-processor misses”. This paper proposes a technique to eliminate interprocessor misses by giving replacement ownership of a block to a processor who brought
it into the cache. This reduction in interprocessor misses, which constitutes 40% of over all misses, will result in performance improvement. Also two different ways of relinquishing the ownership of a block are being proposed, so that if some other processor, other than owner, can make use of the block in a more
efficient way, ownership will be transferred to the new processor.
Original language | English |
---|---|
Title of host publication | web proceedings of 15th International Conference on High Performance Computing, student symposium HiPC-SS08 |
Number of pages | 5 |
Publication status | Published - 2008 |