Blaize delivers breakthrough in AI Edge computing

13 August 2020

Blaize has announced its first AI computing hardware and software products built that look to address rhe compute and productization of AI applications at the edge.

With multiple feature advancements, the Blaize Pathfinder and Xplorer platforms coupled with the company’s AI Software Suite will enable developers to develop more practical and commercially viable edge AI products across a wide range of edge use cases and industries.

“Today’s edge solutions are either too small to compute the load or too costly and too hard to productize,” explained Dinakar Munagala, Co-founder and CEO, Blaize. “Our AI edge computing products overcome these limitations of power, complexity and cost to unleash the adoption of AI at the edge, facilitating the migration of AI computing out of the data centre to the edge.”

Edge AI markets are projected to grow rapidly as demand for AI at the edge transforms entire industries and AI functionality becomes a “must-have” requirement for new products. According to Omdia Research, edge chipset revenue are set to reach $51bn in 2025.

The Blaize embedded and accelerator platforms are built on the company’s Graph Streaming Processor (GSP) architecture designed for the demands of edge computing. With 16 GSP cores and 16TOPS of AI inference performance within a 7W power envelope, the GSP is able to deliver up to 60x better system-level efficiency vs. GPU/CPUs for edge AI applications.

In addition, the GSP enables 50x less memory bandwidth, and 10x lower latency. This efficiency comes without compromising flexibility – the Blaize GSP is a 100% programmable processor – and features advanced capabilities including multi-threading and streaming.

The P1600 embedded Systems on Module (SOM) brings the programmability and efficiency benefits of the GSP to embedded edge AI applications deployed at the sensor edge, or on the network edge. No host processor is needed for the P1600, just plug it in and go.

The Xplorer platform has been built to accelerate AI applications at the edge of the enterprise via easy plugin to the PCIe slot in a host server or appliance. The X1600E is a small form factor accelerator platform for small and power-constrained environments. This is an accelerator form factor uniquely available from Blaize due to the efficiency of the GSP – and can be easily added to accelerate AI apps in industrial PCs or as a rack of cards in a small 1U server. The X1600P is a standard PCIe-based accelerator in a half-height, half-width form factor and can be used to replace a power-hungry desktop GPU in edge servers and provide anywhere from 16-64TOPS of AI inference performance within a very low power envelope.

The Blaize AI software suite is built on open standards and provides access for product developers helping them to move away from being locked into proprietary development environments.

Comprised of the Picasso software development kit (SDK), and AI Studio, a completely code-free visual interface, the software suite offers tools for both traditional developers and non-coder domain experts. In addition, both tools sets make use of the Blaize Netdeploy, a Blaize innovation with edge-aware algorithms to get better accuracy and performance for edge deployments.

Example applications include:

• Industrial: Running five independents neural networks at 50FPS with less than 100ms aggregate latency monitoring human and robot pose detection & position, product ID & count, and product quality.
• Smart City: Three independent neural networks running at 50FPS monitoring human detection, pose, position, auto detection and traffic intersection safety & security.
• Sensor Fusion: High resolution FHD video and Lidar/Radar sensor fusion running as a complete graphnative application on Blaize GSP.
• Last mile delivery: Sensor fusion of HD cameras at 30FPS & lidar/radar with less than 100ms latency and less than 15W power.
• Retail: Four independent neural networks running at 50FPS monitoring human detection, position, mask, and theft for store safety and security, and product detection.

Author
Neil Tyler