Efinix has released the latest FPGA in its Titanium family that offers three times the logic density of its existing products for embedded computing applications, delivering 180,000 logic elements in a 10mm x 10mm footprint.
The new Titanium Ti180 FPGA is said to address the need for higher resolution, higher bit depth sensors designed to maximize accuracy in AI and ML applications. Higher MIPI data rates through its 2.5G MIPI interfaces enable sensor aggregation to unlock a cost-effective and flexible solution for applications in the automotive and VR markets.
Efinix Co-Founder, CEO, and President, Sammy Cheung, said, “Leveraging our radical quantum computing fabric, the Ti180 delivers 15 times the power efficiency of previous generations of Efinix FPGAs and delivers performance without precedent for embedded computing applications with space and power limitations.”
The new device has a hardened LPDDR4 interface for good bandwidth to external memory. A Linux-compatible quad-core RISC-V processor makes the Ti180 suitable as a platform for high-performance embedded computing applications where high-speed access to large amounts of local storage is paramount. A low-power version without the LPDDR4 interface is also available for applications where power consumption is the main limitation. A key part of the product proposition is the ability to enable hardware-accelerated AI and ML applications with zero NRE, helping to accelerate the development of solutions for this market.
Efinix’s titanium FPGAs are based on what the company calls it’s Quantum Fabric, which in turn is made up of configurable tiles called “interchangeable logic and routing” (XLR) cells. These cells optimize routing speed and efficiency while achieving high utilization rates. The matrix also features highly configurable on-board memory blocks as well as dedicated high-speed DSP blocks. Together, these features are designed to deliver optimal performance for applications ranging from edge computing to industrial automation and video processing.
The Ti180 is manufactured using the TSMC 16nm process and uses the cost-effective energy efficiency of this process to bring this extension to the Titanium range. Lucas Tsai, Director of Market Development and Emerging Business Management at TSMC North America, said: “We are delighted to partner with Efinix to enable its Ti180 to benefit from the superior performance and power consumption of our industry-leading 16nm process. We look forward to continuing our partnership with Efinix to unlock more silicon innovations for next-generation FPGA technology, especially in AI and ML applications.
- Artificial Intelligence in the Fourth Industrial Revolution
- Adversarial Machine Learning: Definition, Types, Defense, and Fight with Artificial Intelligence
- How structural proteomics is being advanced by artificial intelligence