California startup Kneron recently rolls out its newest creation, a neural processing unit designed for artificial intelligence.
A neural processing unit (NPU) is a specialized piece of hardware designed for accelerating artificial intelligence applications. Kneron claims that its newest flagship hardware is twice as powerful as the competition, and uses less power.
A cut above the rest
Kneron calls its newest hardware as the KL720 NPU. The hardware is capable of processing artificial intelligence applications that do not require the cloud.
Kneron is taking the community by storm, following its recent Series A round of funding. The firm received funding from major tech companies like Alibaba, Qualcomm, and SparkLabs Taiwan.
The firm claims that its newest hardware is comparable to that of Intel, and even Google. In comparison, Kneron claims that the KL730 NPU is twice as energy-efficient as that of Intel Movidius. The firm adds that its hardware is four times as efficient as that of Google Coral Edge TPU.
AI company, Kneron, deceived a payment system with photo-realistic masks, revealing how devices are exposed to these surroundings daily, increasing the chances of attacks. https://t.co/Ls60rpH6FT#PUFsecurity #cybersecurity #facialrecognition #AI #Kneron pic.twitter.com/cj6L6MVvxk
— PUFsecurity (@PuFsecurity) January 15, 2020
Kneron said that its new hardware is capable of processing 4K images and 1080p videos. The firm claims that the KL720 is also equipped with a new audio recognition program. This allows the hardware to bypass standard wake words, which means that users can directly converse with the device.
In terms of performance, Kneron claims that the KL720 is capable of reaching up to 1.5 teraflops of compute power. At the heart of the hardware is an Arm Cortex M4 control unit. In terms of power efficiency, it uses around 1.2W of power at full capacity.
Additional specs
The Cortex M4 control unit is backed up by a Cadence DSP clocked at 500MHz. This hardware acts like a co-processor, which adds efficiency and more compute power.
The chipset has a massive 128MB LPDDR memory. It supports several interfaces including USB, UART, SDIO, SPI, and CSI-2. The chip comes in two form factors, a 9x9mm model and a 11x11mm model.
In a statement, Kneron founder and CEO Albert Liu says:
“KL72- combines power with unmatched energy-efficiency and Kneron’s industry-leading AI algorithms to enable a new era for smart devices.”
Virtual assistants like Siri and Alexa rely mostly on powerful cloud-based platforms for their processing power. Due to massive compute requirements, it is virtually impossible to develop virtual assistants without the help of cloud computing.
Kneron claims that the new KL720 NPU has enough power to handle such processing requirements. This essentially means that its hardware can process artificial intelligence applications locally.
Image courtesy of Hitesh Choudhary/Unsplash