BittWare FPGA partner program speeds AI into HPC network, sensor processing designs
BittWare (Concord, N.H.), a Molex company and supplier of enterprise-class accelerators for OEM edge and cloud-computing applications, recently introduced a new partner program designed to simplify and streamline customer deployments of high-performance, data-intensive applications.
As characterized by BittWare's April 20 press release announcing the program:
As Field Programmable Gate Arrays (FPGAs) increase in size, complexity and performance, organizations grapple with time-consuming, costly and resource-intensive processes to develop the customized Intellectual Property (IP) and board-level capabilities needed for specific applications. The new BittWare Partner Program removes major time-to-market and technology hurdles by facilitating an ecosystem of FPGA-based enablement IP and full solutions that utilize BittWare’s proven FPGA accelerator technology....Members of the new BittWare Partner Program have chosen the company’s enterprise-class accelerators as one of their preferred platforms for the deployment of advanced compute, network and storage solutions. These significant industry collaborations yield a multitude of benefits, including the ability to decrease development time from up to a year or more to three months or less, as reported by our partner organizations. Additionally, customers gain the peace of mind that performance, quality, reliability and interoperability have been validated.
“BittWare is taking a leadership role in developing a partner program that empowers FPGA designers to access a robust ecosystem of proven IP cores, tools, frameworks and solutions from a centralized source,” commented Craig Petrie, VP, Sales and Marketing for BittWare. “In doing so, we are uniquely positioned to close critical gaps in the FPGA design process while reducing risk and accelerating commercialization of innovative, high-performance applications.”
Value proposition
Fostering an ecosystem of collaborators and innovators, the company says the new partner program aligns technology solutions from both industry-leading and emerging IP providers with BittWare’s compute, network, storage and sensor processing accelerator technologies.
In addition to achieving faster out-of-the-box functionality, BittWare contends that the ability to combine critical components of the FPGA design process reduces engineering and programming requirements, enabling customers to focus in-demand resources on developing unique capabilities.
Moreover, the company maintains that customers "benefit from the opportunity to collaborate much earlier in the design process to meet the varying demands of powerful next-generation applications, such as artificial intelligence, machine learning inference, database acceleration, computational storage, 5G, test and measurement and security."
Partner list
As listed in the press release, BittWare says it inaugural list of ten partners "includes well-established leaders and innovative newcomers focused on solving tough challenges" in the following categories:
Compute:
- Intel: BittWare leverages the Intel Agilex FPGA technology and oneAPI toolkits to simplify development of High-Performance Computing (HPC) applications.
- EdgeCortix: An edge AI-focused fabless semiconductor company with a software-first approach, focused on delivering class-leading efficiency and low-latency for AI Inference.
- Megh Computing: A provider of real-time, AI-based video analytics solutions.
Network:
Atomic Rules: A provider of mission-critical, enterprise-grade IP cores and solutions from the data center to the edge.
Enyx: A developer of ultra-low-latency, FPGA-enabled technologies and solutions for the financial, telecom and HPC industries.
Grovf: A developer of application acceleration and network offload solutions using FPGA chips.
Siama Systems: A provider of Ethernet/IP network infrastructure test solutions for the 5G RAN, MEC and data centers.
Xiphera: A developer of secure and efficient cryptographic IP cores, designed directly for FPGAs.
Storage:
Eideticom: A developer of computational storage solutions for cloud, HPC and enterprise data centers.
IntelliProp: A provider of IP cores, ASIC design and verification services for the storage industry.
Supporting partner testimonials
“We are excited to bring the power of EdgeCortix’s Dynamic Neural Accelerator IP to Intel Agilex FPGAs through the new BittWare Partner Program,” said Sakyasingha Dasgupta, founder and CEO of EdgeCortix. “It’s a game changer on several levels in the AI acceleration market. Together with BittWare, we are delivering a complete solution, from servers to edge boxes, which can be used as ‘drop-in’ replacements for CPUs or GPUs, while improving the performance. From an end-customer perspective, the solutions are easily accessible to software engineers, using standard frameworks like PyTorch, TensorFlow and ONNX. While, from a performance perspective, our solutions deliver significantly lower inference latency on high-resolution streaming data with up to a 7X performance advantage compared to competing FPGA-based offerings.”
"Megh is proud to be part of BittWare's FPGA partner program," said Prabhat K. Gupta, CEO of Megh Computing. "Megh provides an AI-based, fully customizable, cross-platform Video Analytics Solution (VAS). The VAS Suite offering supports various use cases to reduce security risks and improve operational efficiencies for smart buildings and smart factories. This offering benefits from the acceleration provided by BittWare's cards with Intel FPGAs to deliver the highest performance for the lowest TCO."
"IntelliProp is excited to be a part of the BittWare Partner Program to bring value to the storage and memory market by enabling quick-to-market, production-ready IP and reference designs targeting robust BittWare hardware platforms,” said Hiren Patel, CEO of IntelliProp. “We feel the IntelliProp NVMe bridge platforms will bring great benefit to our customers who are looking to implement compression, encryption, or other inline computational functions to solid state drive (SSD)."