Energy-efficient block coding method for GPU data transmission
Abstract In order to reduce the energy consumption of data transmission from DRAM to GPU and overcome the shortcomings of the existing technology, the utility model discloses an energy efficient block coding method for GPU data transmission. According to the binary number of logical value of 1 in 00...
Saved in:
Main Author | |
---|---|
Format | Patent |
Language | English |
Published |
22.07.2021
|
Subjects | |
Online Access | Get full text |
Cover
Loading…
Summary: | Abstract In order to reduce the energy consumption of data transmission from DRAM to GPU and overcome the shortcomings of the existing technology, the utility model discloses an energy efficient block coding method for GPU data transmission. According to the binary number of logical value of 1 in 0000-1111, each 4 bit of the original data is viewed as a basic unit and then falls into different groups, calculate the number of basic units in each group, according to the number of different groups and relations, by applying value mapping method, further reducing the number of logical value of 1 in data transmission, to achieve the purpose of reduce data transfer energy consumption. The utility model has the advantages of simple to-utilize, less computation involved in coding and low cost. Compared with the existing technology, the utility model is more conducive to explore the energy consumption optimization space in the original data, further reducing the number of logical value 1 in the original data, and minimizing the energy consumption of data transmission. |
---|---|
Bibliography: | Application Number: AU20210102878 |