project

scgra v2.0:

1. ALU has regular two stage pipelines for all the computation.

2. MM has supersmall, small, medium and large data sets. FIR, Sobel and Kmean have small, medium and large data sets.

3. Given the loop unrolling information, the script quick-compile could produce bitstream for Zedboard using either SCGRA2x2 or SCGRA5x5.

4. The data used in QuickDough paper comes from this program.

Source code, benchmarks, etc.