Benchmark Results

Compression Performance

The following table presents compression ratio (CR), compression kernel throughput achieved with single and 8 engines, kernel clock frequency met and resource utilization when executed on Alveo U200.

Reported compression ratio is measured on Silesia Corpus compression benchmark.

Architecture Compression Ratio Best Throughput FMax LUT BRAM URAM
LZ4 Streaming (Single Engine) 2.13 287 MB/s 300MHz 4.1K 4 6
Snappy Streaming (Single Engine) 2.13 260 MB/s 300MHz 2.9K 4 6
LZ4 Memory Mapped (8 Engines with Data Movers) 2.13 2.2 GB/s 295MHz 47K 56 48
Snappy Memory Mapped (8 Engines with Data Movers) 2.13 2.2 GB/s 300MHz 47K 48 48
Zlib Memory Mapped (Dynamic Huffman, 8 Engines with Data Movers) 2.74 2 GB/s 284MHz 82.7K 105 48
GZip Memory Mapped (Dynamic Huffman, 8 Engines with Data Movers) 2.74 2 GB/s 284MHz 82.7K 105 48
[*]The amount of resources used indicate that we still have room on Alveo U200 to go for more compute units which can further improve the throughput.

De-Compression Performance

The following table presents decompression kernel throughput achieved with single and 8 engines, kernel clock frequency met and resource utilization when executed on Alveo U200.

Architecture Best Throughput FMax LUT BRAM URAM
LZ4 Streaming (Single Engine and Datawidth: 64bit) 1.42 GB/s 253MHz 6.2K 0 4
Snappy Streaming (Single Engine and Datawidth: 64bit) 1.58 GB/s 300MHz 7.8K 0 4
Zlib Streaming (High Throughput, Datawidth: 64bit) 1.18 GB/s 227MHz 11.6K 3 2
GZip Streaming (High Throughput, Datawidth: 64bit) 1.18 GB/s 227MHz 13.9K 4 2
LZ4 Memory Mapped (8 Engines with Data Movers) 1.8 GB/s 300MHz 30.6K 146 0
Snappy Memory Mapped (8 Engines with Data Movers) 1.8 GB/s 300MHz 31.1K 146 0
[†]The amount of resources used indicate that we still have room on Alveo U200 to go for more compute units which can further improve the throughput.
[‡]Zlib Streaming: Dynamic Huffman and Single Engine performance is provided
[§]GZip Streaming: Full standard support (Dynamic Huffman, Fixed Huffman and Stored Blocks supported)