Compute Express Link (CXL) 2.0 Controller with AMBA AXI interface
Data Compression Accelerators from CAST Now Available on Xilinx Alveo Boards
Reduce bandwidth and storage requirements with standard GZIP/ZLIB/Deflate compression at over 90Gbps on Xilinx Alveo Data Center Accelerator Cards
Woodcliff Lake, NJ — September 27, 2019 — Semiconductor intellectual property (IP) provider CAST, Inc. today announced that its GZIP/ZLIB/Deflate Compression and Decompression reference designs are now available on Xilinx® Alveo™ Data Center Accelerator Cards.
Already successfully deployed by multiple customers on Xilinx Kintex® and Virtex® Ultrascale FPGA boards, the GZIP-RD-XIL GZIP & GUNZIP Accelerator Reference Design now running on Xilinx Alveo PCIe cards delivers an unmatched combination of good compression ratio, low latency, and high throughout. As shown in Table 1, data compression at over 90 Gigabits per second (Gbps) is possible with the compression IP running on the mid-range Alveo U200 Card.
The company believes this industry-leading hardware compression combined with the complete Xilinx Alveo ecosystem makes the GZIP-RD-XIL one of the best-available options for reducing bandwidth and storage requirements in data centers and other data-heavy applications.
Table 1. Representative configurations of the GZIP-RD-XIL reference design running on various Xilinx FPGA boards, with the key statistics of compression ratio (C/R) and performance (Gbps) highlighted.
About the GZIP Accelerator Reference Design
Sourced from partner Sandgate Technologies (www.sandgate.com), the lossless data compression and decompression engines in the reference design comply with the Deflate, GZIP, and ZLIB compression standards.
The ZipAccel-C Compression IP core offers a flexible architecture capable of extremely high throughput and latency as low as a few tens of clock cycles. The ZipAccel-D Decompression IP Core on average outputs three bytes of decompressed data per clock cycle with a latency of a few tens of clock cycles for blocks coded with static Huffman tables, or under 2,000 cycles for those with dynamic Huffman tables. Instances of the cores can be combined for easy scalability, and they are available for multiple ASIC and FPGA technologies.
Figure 1. GZIP-RD-XIL compression or decompression reference design IP block diagram
The ZipAccel cores are part of CAST’s broad IP portfolio, which includes 32- and 8-bit processors; hardware compression/decompression engines for data, images, and video; automotive and other interfaces and peripherals, and a comprehensive SoC security solution.
Learn more about the GZIP-RD-XIL GZIP & GUNZIP Accelerator Reference Design on Xilinx Alveo Cards and CAST’s complete line of IP by visiting www.cast-inc.com, emailing info@cast-inc.com, or calling +1 202.891.8300.
|
CAST, Inc. Hot IP
Related News
- CAST and Accelize Make GZIP Compression Instantly Available via Cloud-Based FPGA Accelerators
- CAST Introduces Secure GZIP/Deflate Data Compression IP Cores from Sandgate Technologies
- QOI Image Compression IP Cores available from CAST and Ocean Logic
- Xilinx Launches Alveo U55C, Its Most Powerful Accelerator Card Ever, Purpose-Built for HPC and Big Data Workloads
- Xilinx Revolutionizes the Modern Data Center with Software-Defined, Hardware Accelerated Alveo SmartNICs
Breaking News
- Credo at TSMC 2024 North America Technology Symposium
- Cadence Reports First Quarter 2024 Financial Results
- Rambus Advances AI 2.0 with GDDR7 Memory Controller IP
- Faraday Reports First Quarter 2024 Results
- RAAAM Memory Technologies Closes $4M Seed Round to Commercialize Super Cost Effective On-Chip Memory Solutions
Most Popular
- GUC provides 3DIC ASIC total service package to AI/HPC/Networking customers
- Omni Design Technologies Joins Intel Foundry Accelerator IP Alliance
- Faraday Partners with Arm to Innovate AI-driven Vehicle ASICs
- Semiconductor Capacity Is Up, But Mind the Talent Gap
- Efabless Announces the Release of the OpenLane 2 Development Platform, Transforming Custom Silicon Design Flows
E-mail This Article | Printer-Friendly Page |