| Project/Area Number |
16K12407
|
|---|
| Research Category |
Grant-in-Aid for Challenging Exploratory Research
|
| Allocation Type | Multi-year Fund |
|---|
| Research Field |
Computer system
|
|---|
| Research Institution | Nara Institute of Science and Technology |
|---|
Principal Investigator |
Nakashima Yasuhiko 奈良先端科学技術大学院大学, 情報科学研究科, 教授 (00314170)
|
|---|
| Project Period (FY) |
2016-04-01 – 2018-03-31
|
| Project Status |
Completed (Fiscal Year 2017)
|
| Budget Amount *help |
¥3,250,000 (Direct Cost: ¥2,500,000、Indirect Cost: ¥750,000)
Fiscal Year 2017: ¥780,000 (Direct Cost: ¥600,000、Indirect Cost: ¥180,000)
Fiscal Year 2016: ¥2,470,000 (Direct Cost: ¥1,900,000、Indirect Cost: ¥570,000)
|
|---|
| Keywords | ライトフィールド / アクセラレータ / CGRA / GPU / DCNN / 深層学習 / HEVC圧縮 / 動画認識 / ライトフィールド動画 / アクセラレーション |
|---|
| Outline of Final Research Achievements |
For the purpose of speeding up transmission of light field movie and speeding up processing, we first studied dedicated video compression hardware. At the time of compression priority, we confirmed hardware increases in 50% and a compression ratio is improved by 34% compared with the preceding 8K motion picture compression hardware. At the time of miniaturization priority, we confirmed hardware is reduced by 24% and a compression ratio is improved by 6% compared with the previous research. For high-speed processing, CGRA was introduced, achieving 89% of the performance in rendering and 4 times faster in distance extraction with 1/3 of computing resources in embedded GPU. We also examined the employment of CGRA to motion picture recognition, confirmed the performance of 40x of ARM Coretex-A9, 11x of Vivante GC 2000+, 6x of Xilinx Zynq (Z-7020) high-level synthesis.
|
