The AI Revolution

Artificial Intelligence, or AI, is seeing an explosion of growth and applications. The flood of data that is generated and collected every day and the computations required to allow smarter algorithms to sift through that data more cost-effectively drive AI development. These new applications have the potential to transform how we live, work, and play, driving new insights and providing more intuitive interactions with our smart devices. Some of these applications include facial recognition, real-time language translation, autonomous driving, faster and more effective drug discovery, and precision agriculture. 

FPGA 101

Field programmable gate arrays (FPGAs) are programmable integrated circuits containing logic elements, DSP blocks, on die memory, and flexible I/O. These building blocks enable the developer or end user to implement any number of functions directly in the hardware. Users can efficiently manage data locality by taking the data directly from the source and tightly coupling the compute with the results between computations directly on the FPGA. In many cases, running computations on a Intel® FPGA is faster, lower power, lower latency, and has higher throughput than using software. Intel® FPGAs can also be reprogrammed, even after they are installed, enabling hardware reuse and reconfiguration.  

Benefits

Intel® FPGAs offer the ultimate flexibility and performance, providing unique value propositions across all vertical markets for AI deployments. 

Performance

Hardware flexibility

Workload flexibility

Power Efficiency

Intel FPGAs provide improved (or low) power efficiency for running AI workloads, such as AlexNet and GoogleNet algorithm, thus reducing overall power consumption and total cost of ownership.   

Deployment Flexibility

An FPGA enables you to run in “offload” mode taking data to and from a CPU or “in-line” where data is processed directly in the FPGA before going to the host processor.

Precision

FPGAs can run any precision and data type, from 64 bit floating point to integer to binary, further allowing you to customize your implementation to your exact needs.

Speed

FPGAs are fast. When response time matters, Intel FPGAs provide excellent speed and deterministic low latency, supporting faster decision making and offering a better customer experience.

I/O Types

FPGA transceivers allow direct interface to any data source and interface standard, including cameras, storage devices, or the network.

Future Algorithms

Because FPGAs are reprogrammable, they are not only useful for today’s AI workloads and algorithms, but are adaptable to the AI architectures of the future.

Throughput

Intel FPGAs can increase the throughput of your system, allowing you to do more with less.

Power envelope

Besides being power efficient, Intel FPGA designs are power scalable, so you can design your workload to use as much power as you need.

Multi-functionality

Intel FPGAs can combine AI algorithms with other important functions in your system into a single chip resulting in low power, low deterministic latency, high throughput, and a low total cost of ownership.