Industry-first AI analog matrix processor launched
The M1108 AMP is offered as delivering for the first time an analog compute solution that achieves best-in-class performance and power with accuracy comparable to digital devices. The processor, says the company, is at the forefront of a major new trend in AI processing and opens up unparalleled possibilities for device deployment at the edge in a broad range of applications and markets, including smart home, AR/VR, drones, video surveillance, smart city, and automation on the factory floor.
While digital inference solutions based on SoCs, TPUs, CPUs, GPUs, and FPGAs can meet some of the challenges of running AI models at the edge rather than in the cloud, inherent limitations due to memory, clock speeds, and process technology create insurmountable trade-offs, with the result that only processors capable of high-performance analog compute – such as the M1108 Mythic AMP – can address them all, the company claims.
“This is a significant inflection point in the industry,” says Mike Henry, co-founder, CEO, and chairman of Mythic. “We are delivering technology that was previously thought to be impossible. Our Analog Compute Engine eliminates the memory bottlenecks that plague digital solutions by efficiently performing matrix multiplication directly inside the flash memory array itself. The high performance and low power of the Mythic AMP combine to open up AI technology to broader application areas and address product categories that are currently inaccessible to comparable digital solutions.”
The M1108 integrates 108 AMP tiles, each with a Mythic Analog Compute Engine (ACE) featuring an array of flash cells and analog-to-digital converters (ADCs), a 32-bit RISC-V nano-processor, a SIMD vector engine, SRAM, and a high-throughput Network-on-Chip (NOC) router. In addition, four control tiles provide a high-bandwidth PCIe 2.0 interface to a system host processor.
With 108 AMP tiles, the M1108 delivers up to 35 Trillion-Operations-per-Second (TOPS) enabling the power-efficient execution of complex AI models such as ResNet-50, YOLOv3, and OpenPose Body25 on a single-chip with extremely low latency. Typical power consumption of the M1108 while running complex AI models at peak throughput is approximately 4 W. And, says the company, with the inherent cost advantages of utilizing mature 40-nm technology and not requiring any external DRAM or SRAM, the M1108 has up to a 10-times cost advantage over comparable digital architectures
The M1108 AMP will be available in both PCIe M.2 and PCIe card form-factors. M1108 PCIe evaluation kits are available on request.
Related articles:
On-device AI startup adopts Codasip RISC-V processor
Neural networking startup raises funds for ‘local AI’ chip
Fastest AI inference chip for edge systems announced
AI chipset market on pace for rapid growth through 2025
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
