The platform, which includes the Metis AI Processing Unit (AIPU) chip and the Voyager SDK software stack – available as cards, boards, and vision systems – looks to accelerate computer vision processing at the edge.
According to Axelera, the Metis AI Platform is intended to redefine affordability for Edge AI solutions without compromising on performance and is the first product in the market to offer hundreds of TeraOPS (TOPs) compute performance at an edge price-point.
The products will be available in different form factors such as: an Edge M.2 acceleration card, offering up to 214 TOPs as well as a PCIe card delivering up to 856 TOPs, powered by four Metis AIPUs. Separately, Axelera AI is working with partners to develop system ready solutions.
The software stack, Voyager SDK, includes an integrated compiler, a runtime stack and powerful optimisation tools that will enable customers to import their own models delivering class-leading performance on the Metis AI platform.
Application templates speed up development, with customers able to choose from Axelera’s curated Model Zoo.
The Metis AIPU delivers high performance at the Edge in a highly cost-effective compute platform and was taped out within 12 months after the successful tape out of the Thetis test vehicle that demonstrated the efficacy of Axelera’s SRAM-based D-IMC engine.
The Metis AIPU is a quad core architecture, and each core can execute all layers of a standard neural network without external interactions delivering up to 53.5 TOPs of AI processing power.
The AIPU can reach 214 TOPS and is said to be the most powerful and advanced chip for edge devices, outperforming many existing products.
The cores can either be combined to boost the throughput of a complex workload or operate independently on the same neural network to reduce latency or process concurrently different neural networks for applications featuring pipelines of neural networks.
Central to the operation of each core is Axelera’s D-IMC engine for accelerating matrix-vector multiplication operations, offering an unprecedented high energy efficiency of 15 TOPs/W at INT8 precision.
The four cores are integrated into a System-on-Chip (SoC), comprising RISC-V controller, PCIe interface, LPDDR4X controller, and a security complex connected via a high-speed Network-on-Chip (NoC). The NoC also connects the cores to a multi-level shared memory-hierarchy of more than 52MiB of on-chip high-speed memories, while the LPDDR4 controller connects to external memory, enabling support for much larger neural networks.
Finally, the PCIe interface provides a high-speed link to an external host, which will offload full neural network applications to the Metis AIPU.
“The Metis AIPU not only delivers record performance and energy efficiency but also achieves FP32 iso-accuracy without retraining. It demonstrates the effectiveness of Axelera’s D-IMC engine by achieving matrix-vector multiplications with high-precision and constant time-complexity, which in tandem with the advanced quantization algorithms, the large on-chip SRAM and the RISC-V technology are well-suited to address the stringent operational requirements of the AI applications at the Edge,” said Axelera AI co-founder and CTO Evangelos Eleftheriou. He continued, “Axelera's end-to-end integrated software stack, built on open, industry-standard frameworks and APIs, addresses the user pain points of AI application development at the Edge”.
The Metis AI Platform enables customers to rapidly develop and deploy computer vision applications. The Voyager SDK provides turnkey pipelines for state-of-the-art models, which a user may customise for their application, generate optimised code for, and deploy to Metis-enabled edge devices with one-click simplicity.
The platform supports solutions for a wide range of environments, ranging from fully embedded use cases to distributed processing of multiple 4K video streams across networked devices.
By delivering class-leading ease-of-use as part of the Metis AI Platform, Axelera enables software developers to focus on solving the application problem – no Deep Learning expertise is required –significantly reducing time to market.
Via the Axelera AI Model Zoo, customers can access Neural Networks to customise, fine-tune or import their own pre-trained neural networks. The Voyager SDK automatically quantizes and compiles neural networks that have been trained on different frameworks, generating code which runs on the Metis AI platform with industry-leading accuracy. Optimised networks running on Metis AIPU are indistinguishable from those running on systems with floating-point units.
“The Metis AI Platform represents an inflection point for businesses looking to deploy computer vision applications at the Edge,” said Paul Neil, VP Product Management at Axelera AI. “Tightly integrated hardware and software capability, a curated model zoo and customizable application templates enable customers to accelerate through the evaluation and development cycle and shorten time-to-revenue.”
Axelera AI has opened an Early Access Program, offering customers the opportunity to help shape the next generation of computer vision products at the Edge.
