Celoxica Adds Floating-Point Toolkit to IP Portfolio; Cores Optimized for FPGA-Based Digital Signal Processing & High-Performance Computing Applications
OXFORDSHIRE, England--July 19, 2006--Celoxica (AIM:CXA) today announced availability of a toolkit for Floating-Point arithmetic implementation in high-density programmable logic. The IEEE 754 compliant libraries support single, double and custom precision Floating-Point. They are fully parameterizable for area, latency and performance optimization across all leading programmable logic architectures and can be tuned to specific application requirements delivering industry-leading performance and unrivalled flexibility.
Developed in conjunction with the Institute of Information Theory and Automation (UTIA) in the Czech Republic, the underlying IP behind the Floating-Point ToolKit represents the first milestone between Celoxica and UTIA since the two organizations signed a multi-year research agreement. The Floating Point ToolKit is being deployed into high-performance computing (HPC) application areas such as the automated trading of equities and related financial derivatives, oil and gas exploration and embedded applications including audio and video DSP filtering.
Parallel designs using the Floating-Point core modules can deliver up to 50 Giga Floating-Point Operations per Second (GFLOPS) and support the processing requirements in real-time imaging and signal processing applications, and compute-intensive routines typically used in clustered server and supercomputer architectures. By exploiting the inherent parallelism of FPGA hardware, exceptionally high Floating-Point arithmetic can be achieved with significant performance-per-watt energy savings. The combination of the Floating-Point ToolKit with programmable logic means the number of instances of the Floating-Point operator is user configurable and scalable.
The Floating-Point ToolKit supports pipelined add, subtract, multiply, divide and square-root operations, and conversions to and from 32bit fixed point. The add module has a five stage pipeline for all precisions and the multiply module is provided with user configurable pipelines to support the embedded multiplier and DSP blocks in modern FPGA devices. 18m12, 24m16, 32m24, 40m32, 64m53 precisions are supported as is IEEE754 rounding to the nearest and even. The custom precisions provided in the ToolKit are optimized for designers who want to take maximum advantage of embedded FPGA resources such as DSP blocks and make optimum use of existing gate counts.
For simulation and verification the ToolKit supports processor-in-the-loop and hardware-in-the-loop simulation. Bit-exact Simulink blocks for all arithmetic operations supported by the library (precisions 18m12, 24m16, 32m24, 40m32 and 64m53) are included and test vectors can be created in Simulink models and debugged in Celoxica's DK Design Suite. Breakpoints help debug the implementation at the register level and DK's simulator reads data generated in the Simulink model and returns data back for verification of the bit exact results. Implementation to the FPGA is supported with API layers that allow application reuse across different programmable platforms without additional development or recoding.
"This important addition to our IP portfolio addresses customer need in both the DSP and HPC space for a very high-performance, fully parameterizable and platform-independent floating-point capability," said Chris Sullivan, director of strategic marketing for Celoxica. "We can readily tune and re-tune the libraries to meet changing architectures and specifications, and the current implementation not only provides industry-beating performance and flexibility, but the libraries themselves demonstrate the extreme circuit optimization that can be achieved using the DK Design Suite."
The ToolKit is available from Celoxica in a variety of different configurations. For further information contact info@celoxica.com or your local Celoxica sales office.
About Celoxica
A leader in electronic system level design (ESL), Celoxica is enabling the next generation of advanced electronic products by producing tools, boards, IP and services that turn software into silicon. Celoxica technology raises design abstraction to the algorithm level, accelerating productivity and lowering risk and costs by generating semiconductor hardware directly from C-based software descriptions. Adding to a growing installed base, Celoxica provides the world's most widely used C-based behavioral design and synthesis solutions to companies developing semiconductor products in markets such as consumer electronics, defense and aerospace, automotive, industrial and security. Celoxica is a publicly traded company on the Alternative Investment Market of the London Stock Exchange under the symbol CXA. For more information, visit: www.celoxica.com.
Related Semiconductor IP
- AES GCM IP Core
- High Speed Ethernet Quad 10G to 100G PCS
- High Speed Ethernet Gen-2 Quad 100G PCS IP
- High Speed Ethernet 4/2/1-Lane 100G PCS
- High Speed Ethernet 2/4/8-Lane 200G/400G PCS
Related News
- Cutting-edge 18-bit 100dB Stereo Audio ADC IP Core proven in 28nm Silicon, Offering Unmatched Audio Signal Processing Capabilities is available for immediate Licensing into Audio Chipsets, Digital Cameras, and Automotive Applications
- MIPI C-PHY / D-PHY Combo IP (4.5Gbps) and CSI Tx Controller IP Cores, to meet the highest standards of performance and reliability for a wide range of applications
- Cortus Announces FPS6 32 bit Floating Point Microcontroller IP Core for High Performance Control and Signal Processing Applications
- SiFive Rolls Out Powerful New RISC-V Portfolio to Address Unmet Performance and Feature Needs of Rapidly Evolving Next-Gen Digital Automobiles
Latest News
- HPC customer engages Sondrel for high end chip design
- PCI-SIG’s Al Yanes on PCIe 7.0, HPC, and the Future of Interconnects
- Ubitium Debuts First Universal RISC-V Processor to Enable AI at No Additional Cost, as It Raises $3.7M
- Cadence Unveils Arm-Based System Chiplet
- Frontgrade Gaisler Unveils GR716B, a New Standard in Space-Grade Microcontrollers