Thermal aware energy efficient Gurumukhi Unicode reader for natural language processing
Amanpreet Kaur, Gurpreet Singh, Bishwajeet Pandey, Furqan Fazili · International Conference on Computing for Sustainable Global Development · 2015
In this work, we are making thermal aware design. That is able to work in range of ambient temperature from 20 °C to 80 °C with consideration of different airflow (250 LFM-500LFM) and Heat sink (medium profile, high profile and standard). LFM is an acronym for Linear Feet per Minute. Gurumukhi is the widest use script in Punjabi language. But, there is no energy efficient design available in order to read and process Gurumukhi Unicode. Our work is filling the research gap available in natural language processing domain especially for Punjabi language. There are 48.96%, 36.77% and 20.91% reductions in leakage power for 1GHz device operating frequency, when we scale down ambient temperature from 80 °C to 20 °C, 40 °C and 60 °C respectively. There are 73.83%, 49.34% and 24.85% reductions in junction temperature for 1GHz operating frequency & 250LFM airflow & Medium profile Heat Sink, when we scale down ambient temperature from 80 °C to 20 °C, 40 °C and 60 °C respectively. With high profile Heat sink and 500LFM airflow, we are making this design more energy efficient and thermal aware.