6. Conclusion
GPUs are increasingly being used in supercomputers, desktops and, increasingly, embedded systems. This requires a great deal of programming effort to get the most out of them, particularly in terms of performance per watt. The efficiency of GPUs relies essentially on their high memory bandwidth, thanks to large memory buses, enabling large quantities of data to be processed in parallel. To achieve this, GPU-based algorithms must enable independent processing of large numbers of data to be executed in parallel. This scalability is essential if it is to be portable from one generation to the next. Memory bandwidth is set to increase significantly with the advent of 3D silicon chip technology and stacked memories, enabling matrix connections with buses currently of 4096 bits, which are set to grow considerably, enabling much stronger coupling between cores and memory. This parallel computing...
Exclusive to subscribers. 97% yet to be discovered!
Already subscribed? Log in!
Conclusion
Glossary
Article included in this offer
"Signal processing and its applications"
(
151 articles
)
Updated and enriched with articles validated by our scientific committees
A set of exclusive tools to complement the resources
Bibliography
Exclusive to subscribers. 97% yet to be discovered!
Already subscribed? Log in!
