InverseUniverse
Newbie level 6
- Joined
- Oct 12, 2012
- Messages
- 12
- Helped
- 3
- Reputation
- 6
- Reaction score
- 3
- Trophy points
- 1,283
- Activity points
- 1,353
Hey guys
I am working on implementing the algorithm of 2D DCT on Nexy2 from Digilent, which is based on Xilinx Spartan 3E. My main concern is data precision and its implications on the memory.
The board has 16MB SDRAM (which I am planning to use for this calculation) and 16MB Flash Memory (which i'm only planning to use to fetch constant data).
I am planning to use MATLAB to establish a serial connection and send the pixel data of greyscale pictures. Normally, MATLAB would provide them with an 8-Bit precision (0-255), however, I am not convinced of using the same resolution for my calculations on the FPGA. I am looking at allowing room for more calculation precision by representing everything at 12-Bits as it is the good balance between memory usage and resolution.
I really need your ideas, what do you think?
I am working on implementing the algorithm of 2D DCT on Nexy2 from Digilent, which is based on Xilinx Spartan 3E. My main concern is data precision and its implications on the memory.
The board has 16MB SDRAM (which I am planning to use for this calculation) and 16MB Flash Memory (which i'm only planning to use to fetch constant data).
I am planning to use MATLAB to establish a serial connection and send the pixel data of greyscale pictures. Normally, MATLAB would provide them with an 8-Bit precision (0-255), however, I am not convinced of using the same resolution for my calculations on the FPGA. I am looking at allowing room for more calculation precision by representing everything at 12-Bits as it is the good balance between memory usage and resolution.
I really need your ideas, what do you think?
