本篇論文提出了High Efficiency Video Coding (HEVC) Inverse Discrete Cosine Transform (IDCT)的硬體架構,可支援4K(3840x2160)@30fps 的畫質,並且能夠針對不同尺寸的轉換單元(Transform Unit)32×32、16×16、8×8、4×4進行運算。使用一顆1-D IDCT Core以及一顆Transpose Memory來實現2-D的IDCT運算,在1-D IDCT Core裡採用16條運算路徑來實現高吞吐率,並且可以讓1-D和2-D的資料同時進行IDCT運算。在矩陣運算方面採用分散式演算法(Distributed Arithmetic)、加法器以及移位器,來取代傳統使用乘法器的矩陣運算。在製程技術上,使用台積電90 nm製程的製程技術來做合成,在2-D的轉換核心運算上有2G pels/s的吞吐率,整個架構的Gate counts為337K,硬體效率為5.93((〖10〗^3 pels)/s/gate)。
In this paper, a hardware design which can support for High Efficiency Video Coding (HEVC) Inverse Discrete Cosine Transform (IDCT) for 32×32 Transform Unit (TU) sizes is proposed and is implemented by a using single 1-D IDCT core with a transpose memory to achieve low cost design. The proposed 1-D IDCT core employs 16 calculating paths to achieve a high throughput rate and is implemented by a 1-D inverse transform core which can calculate 1st-D and 2nd-D data simultaneously in 32 parallel paths. The proposed 2-D transform core has a throughput rate of 2-Gpels/s with 337k gate counts when implemented into the TSMC 90nm CMOS technology.
為了持續優化網站功能與使用者體驗,本網站將Cookies分析技術用於網站營運、分析和個人化服務之目的。
若您繼續瀏覽本網站,即表示您同意本網站使用Cookies。