Abstract

Vector Processing is an efficient way to exploit data-parallel with wide data operations, which is widely employed by wireless baseband signal processing. However, wide data operations also cost much more power and lead to high power consumption especially for vector register file (VRF). To resolve this issue, firstly, we profiled the read/write characteristics of LTE baseband signal algorithms to VRF and found three facts exist: a lot of copy or movement operations are sourced from vector load/store, the occasions for concurrent access to all VRF read/write ports are little and a large number of short-lived values exists. Based on these observations, a buffer-aided VRF is presented to exploit these characteristics for low power. The experiment results show: The proposed buffer-aided VRF can effectively reduce the vector copy and movement cost comparing with general register file architectures and also achieve better power saving effects than similar buffering technologies. A reduction of 40.9% power dispassion over centralized VRF can be reached in register file level with negligible performance loss.

Full Text
Paper version not known

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call

Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.