What is CEVA?
CEVA is a licensor of phones, multimedia applications, affinity technologies, semiconductor companies and OEMs that are helping mobile consumers, automotive and IoT markets. CEVA DSP IP group includes widespread platforms for multimode 2G/3G/LTE/LTE-A baseband processing in terminals and infrastructure.
CEVA further includes computer visualization and computational photography for any camera-using device, audio/voice/speech and ultra-low power always-on/sensing applications for numerous markets.
For connectivity issue, CEVA offer the industry’s most widely adopted IPs for Bluetooth (Smart Ready), (802.11 b/g/n/ac up to 4×4 Wi-Fi) and (SATA and serial storage SAS). One in every three phones sold worldwide is power-driven by CEVA, from many of the world’s leading OEMs counting Samsung, HTC, LG, Huawei, Xiaomi, Lenovo, ZTE, Micromax, Coolpad, and Meizu.
Find what tensor flow brings to you
Tensor Flow™ is an open resource software collection for numerical calculation using data stream graphs. Nodes in the graph represent statistical operations, while the graph boundaries symbolize the multidimensional data arrays (tensors) communicated between them. The lithe architecture allows you to organize computation to one or more CPU or GPUs in a desktop, server, or portable device with a single API. CEVA announced its second generation deep neutral network software structure.
Use of CDNN2 (Ceva Deep Neural Network)
The CDNN2 brings support for Google’s tensor flow to embedded systems. CDNN2 can ease on-device deep learning-based video analytics in real time, saving bandwidth and storage compared to running the similar analytics in the cloud. The CDNN2 software structure is part of an on-device deep learning, which could construct devices much faster even without a connection to the Internet. It’s also a big profit to those who don’t faith third-party providers with applications such as analyzing video feeds of their homes.
The CDNN2 is harmonizing with Ceva’s own “intellectual vision processor,” the Ceva-XMP4, which can facilitate 3D vision, computational photography, image observation and analytics.
Benefits of tensor flow support with other system
One of the major accompaniments of Ceva’s second-generation profound neural network software scaffold is support for Google’s Tensor Flow, which are speedily turning out to be one of the most trendy machine learning software libraries.
- CDNN2 also brings hold up networks, which permit any given network to work with any input declaration, as well as improved capabilities and presentation for the newest network typologies and layers.
- Google reimbursement from having everyone approves Tensor Flow because if there are more Tensor Flow developers, the supply of chips that natively support Tensor Flow will cultivate as well.
- Eventually it gives Google the prospect to acquire cheaper Tensor Flow-optimised chips for its data centers.
How The Embedded Platforms Deliver Extremely High Performance On Very Low Power Consumption
Various embedded platforms can deliver extremely high performance on very low power consumption.
One of the foremost features of the CDNN2 is that it contains an offline Ceva Network Generator that can alter a pre-trained network into one that is extremely optimized for embedded device driver development with fixed-point arithmetic hardware.
Usually only grown-up or ultra-low-cost embedded chips lack hold for a floating-point component. According to CEVA, a network that’s reform for these chips can be generated at the shove of a button and power consumption will not be a problem anymore.
