| Project/Area Number |
17K00088
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Research Field |
Computer system
|
|---|
| Research Institution | Fukuoka University |
|---|
Principal Investigator |
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
請園 智玲 福岡大学, 工学部, 助教 (50610060)
|
|---|
| Project Period (FY) |
2017-04-01 – 2021-03-31
|
| Project Status |
Completed (Fiscal Year 2020)
|
| Budget Amount *help |
¥4,550,000 (Direct Cost: ¥3,500,000、Indirect Cost: ¥1,050,000)
Fiscal Year 2019: ¥910,000 (Direct Cost: ¥700,000、Indirect Cost: ¥210,000)
Fiscal Year 2018: ¥1,300,000 (Direct Cost: ¥1,000,000、Indirect Cost: ¥300,000)
Fiscal Year 2017: ¥2,340,000 (Direct Cost: ¥1,800,000、Indirect Cost: ¥540,000)
|
|---|
| Keywords | 低消費電力 / システムオンチップ / 計算機システム / 情報システム / ディペンダブル・コンピューティング / ディペンダブルコンピューティング |
|---|
| Outline of Final Research Achievements |
We have studied approximate arithmetic circuits. First, we proposed an approximate adder, which can dynamically reconfigure its accuracy. Power consumption can be reduced according to the selected accuracy, and at the same time, the calculation speed can be improved. Furthermore, we improved this design so that the accuracy was improved with keeping energy efficiency. Regarding the improvement of calculation accuracy, it should be noted that the handling of negative numbers, which is a weak point of previously studied approximate adders, is taken into consideration. Next, we proposed an approximate multiplier, which can dynamically reconfigure its accuracy. Power consumption can be reduced according to the selected accuracy, and at the same time, the calculation speed can be improved. Furthermore, we proposed a fixed-accuracy approximate multiply-accumulate unit, which is an extension of the approximate multiplier.
|
| Academic Significance and Societal Importance of the Research Achievements |
IoT時代が到来し，様々なセンサから時々刻々とビッグデータが生産されている．クラウド上のサーバに集約されたセンサビッグデータをディープラーニング等のAI技術で解析することで，人類の生活をより豊かにする試みがなされている．しかしセンサデータをそのままネットワークに流すと莫大なエネルギーを要する．出来るだけエッヂ（末端）で処理し，データ量を少なくした上で通信を行うべきである．しかしエッヂでは電力供給に制約があるうえ，それが故に高い演算性能を期待できない．本研究成果の近似演算器を用いることで性能を犠牲にしないで消費エネルギーを削減でき，IoTとAIを応用するエッヂコンピューティングを実現できる．
|
