=========================================== Xilinx Snappy Compression and Decompression =========================================== Snappy example resides in ``L2/demos/snappy`` directory. Xilinx Snappy compression/decompression is FPGA based implementation of standard Snappy. Xilinx implementation of Snappy application is aimed at achieving high throughput for both compression and decompression. This Xilinx Snappy application is developed and tested on Xilinx Alveo U200. To know more about standard Snappy application please refer https://github.com/snappy/snappy This application is accelerated using generic hardware architecture for LZ based data compression algorithms. Results ------- Resource Utilization ~~~~~~~~~~~~~~~~~~~~~ Table below presents resource utilization of Xilinx Snappy compress/decompress kernels with 8 engines for single compute unit. The final Fmax achieved for this design is 299MHz ========== ===== ====== ===== ===== ===== Flow LUT LUTMem REG BRAM URAM ========== ===== ====== ===== ===== ===== Compress 48K 10.8K 50.6K 48 48 ---------- ----- ------ ----- ----- ----- DeCompress 12.4K 2.9K 16.5K 16 4 ========== ===== ====== ===== ===== ===== Performance Data ~~~~~~~~~~~~~~~~ Table below presents the kernel throughput achieved with single compute unit during execution of this application. ============================= ========================= Topic Results ============================= ========================= Compression Throughput 1.8 GB/s Decompression Throughput 1 GB/s Average Compression Ratio 2.14x (Silesia Benchmark) ============================= ========================= Note: Overall throughput can still be increased with multiple compute units. Software & Hardware ------------------- :: Software: Xilinx Vitis 2020.2 Hardware: xilinx_u200_xdma_201830_2 (Xilinx Alveo U200) Usage ----- Build Steps ~~~~~~~~~~~ Emulation flows ^^^^^^^^^^^^^^^ :: make run TARGET= DEVICE=xilinx_u200_xdma_201830_2 Note: This command compiles for targeted emulation mode and executes the application. Hardware ^^^^^^^^ :: make all TARGET=hw DEVICE=xilinx_u200_xdma_201830_2 Note: This command compiles for hardware execution. It generates kernel binary ".xclbin" file. This file is placed in ./build/xclbin*/ directory under Snappy folder. Executable Usage ---------------- 1. To execute single file for compression : ``./build/xil_snappy -xbin ./build/xclbin__/ -c `` 2. To execute single file for decompression : ``./build/xil_snappy -xbin ./build/xclbin__/ -d `` 3. To validate single file (compress & decompress) : ``./build/xil_snappy -xbin ./build/xclbin__/ -t `` 4. To execute multiple files for compression : ``./build/xil_snappy -xbin ./build/xclbin__/ -cfl `` 5. To execute multiple files for decompression : ``./build/xil_snappy -xbin ./build/xclbin__/ -dfl `` 6. To validate multiple files (compress and decompress) : ``./build/xil_snappy -xbin ./build/xclbin__/ -l `` - ````: Contains various file names with current path - Note: Default arguments are set in Makefile The usage of the generated executable is as follows: .. code-block:: bash --help, -h Print Help Options --compress, -c Compress --decompress, -d Decompress --test, -t Xilinx compress & Decompress --compress_list, -cfl Compress List of Input Files --decompress_list, -dfl Decompress List of compressed Input Files --test_list, -l Xilinx Compress & Decompress on Input Files --max_cr, -mcr Maximum CR Default: [10] --xclbin, -xbin XCLBIN --device_id, -id Device ID Default: [0] --block_size, -B Compress Block Size [0-64: 1-256: 2-1024: 3-4096] Default: [0]