InAccel is a world-leader in application acceleration using FPGAs in the cloud.
We embrace cutting-edge technology to help companies speedup mission-critical applications in the cloud and reduce their the total cost of operations (OpEx).
We provide both readily available accelerator cores and customized solutions based on the customer’s requirements. The accelerators are provided in the form of IP cores and can be used either on premises or in the Cloud. FPGA accelerators can massively accelerate computational intensive algorithms and that is why they are a perfect fit for a plethora of cutting edge fields including data analytics, machine learning, compression algorithms etc.
InAccel accelerators are fully compatible with Amazon AWS, Alibaba Cloud and Huawei Cloud FPGA instances. They are also compatible for both Intel and Xilinx FPGAs.
FPGA Resource Manager-
Inaccel's FPGA Resource Manager provides transparent acceleration to the user. The whole idea lies in the concept of client - server architecture. The FPGA Resource Manager serves as the server entity waiting for any user requests (clients). A user can simply implement and run accelerated tasks using the Manager's API.
Containerized environment and Flawless Scalability
InAccel offers its FPGA Manager in a Docker container that can be downloaded right from the Docker Store. This way we guarantee easy deployment and an isolated environment that will not interfere with your system's settings. And if you have more than one FPGAs running on your system, then that is not a problem for us. FPGA manager scales out managing multiple FPGAs at the same time.
We do care for the seamless integration of our accelerators with widely used frameworks like Apache Spark, Tensorflow etc. Acceleration of tasks as a technique isn't a new factor but it always came at the price of understanding how accelerators work and messing up with very low level information. We believe that acceleration is acceptable only when it comes at no further costs. That is why we also provide a framework that seamlessly offers acceleration for Apache Spark ML library. Just install an FPGA card on your system or choose an FPGA instance on the Cloud and let our framework take acceleration to a whole new level.
Zero code changes
Using the term seamless integration we mean zero code changes in your applications. Just install our framework and run your Spark ML applications as you would in the past.