The computing landscape continues to change and evolve, and new technology tools are being applied to challenges once thought insurmountable. Some of these are business and commercial applications, others in research and scientific breakthroughs. Ten years ago, who could have imagined what we can do with genomics sequencing and analysis today, where a complete DNA strand can be analyzed in a few hours?
For many of these compute challenges, general-purpose processors benefit from targeted acceleration to deliver the “step function” performance improvements needed to achieve a breakthrough. Field Programmable Gate Arrays (FPGAs) are the accelerators of choice for a growing range of technical, business and research problems, from financial tech to machine learning to big data analytics.
Acceleration-as-a-Service with FPGAs is Growing Rapidly
In recent months, several leading cloud service providers, including Amazon and Baidu, announced FPGA Acceleration-as-a-Service offerings. A public cloud environment can be a great place for developers to start the FPGA journey, with virtually no initial capital outlay and a low-risk environment for testing and experimentation. As workloads and applications are adapted to FPGA acceleration, cloud deployments can readily scale to meet the growing capacity.
Intel joined Alibaba Cloud recently to announce their new FPGA acceleration offering, based on the Intel® Arria® 10. “Intel® FPGAs offer us a more cost-effective way to accelerate cloud-based application performance for our customers running business applications and demanding data and scientific workloads,” said Jin Li, Vice President of Alibaba Cloud. “Another key value of FPGAs is they provide high performance at low power, and the flexibility for managing diverse computing workloads.”
Right now, Cloud Service Providers have a unique opportunity to get in on this growing trend and create differentiated FPGA services that build their portfolios and businesses. Customer needs are diverse, and they will be looking for FPGA service choices that meet their requirements.
Intel FPGAs Lead the Industry in Performance and Ease of Use
Intel gives cloud service providers a range of form factor options so they can optimize for their infrastructure needs. Our traditional FPGA components have been joined by easy-to-deploy, PCIe-based Intel Programmable Acceleration Cards that can be added to any Intel server, including our latest Intel® Xeon® Scalable platforms.
Intel’s Stratix® and Arria FPGA product families are ideal for cloud service providers looking to power their Acceleration-as-a-Service offering. Intel is the silicon manufacturing leader, and Stratix is shipping on 14-nanometer today, with 10-nanometer on our short-term roadmap. Silicon leadership translates to performance and power efficiency. New security features have been added, such as Intel® Virtualization Technology SR-IOV, an important isolation feature in a multi-tenant Acceleration-as-a-Service environment.
Intel also created a software framework that vastly simplifies FPGA programming. Historically, FPGA developers had to be fluent in hardware-descriptive languages, such as Verilog or VHDL, and have specific knowledge of FPGA hardware designs. With the Acceleration Stack for Intel Xeon CPU with FPGAs, a robust collection of software, firmware, and tools, designed and distributed by Intel, coding for Intel FPGAs moves into like familiar environments like RTL or OpenCL, abstracting away the hardware-specific intricacies of the past, and opens up FPGAs to mainstream developers. The Acceleration Stack provides optimized and simplified hardware interfaces and software APIs, saving the developer time so they can focus on the unique value-add of their solution.
At the heart of the Acceleration Stack is the Open Programmable Acceleration Engine (OPAE). In a nutshell, the Open Programmable Acceleration Engine (OPAE) is software programming layer that provides a consistent API across FPGA product generations and platforms. To foster an open ecosystem and encourage the use of FPGA acceleration for data center workloads, Intel has open sourced the technology for the industry and developer community.
Equipped with the OPAE framework, a rich ecosystem of third-party companies are developing a wide range of workload accelerators that can be used by application developers or cloud service providers. One example is Falcon Computing, a solution partner of Intel, who offers its FPGA-accelerated genome sequencing solution on Alibaba Cloud, an impressive 3X overall GATK pipeline speedup and 26X alignment algorithm speedup versus using the CPU alone.1
The Opportunity is Now
We are still in the initial ramp-up stage of Acceleration-as-a-Service with FPGAs, and cloud service providers have a clear window of opportunity to enhance their product portfolios in this area of rapidly growing demand. We look forward to working with them as we grow the business and solve the problems that will change the world. Join us in accelerating compute for the data demands of today and beyond.
- Intel FPGA Acceleration Hub
- Intel® Programmable Acceleration Card with Intel Arria® 10 GX
- Supercharge Your Data Center Video featuring new Intel FPGA Acceleration Solutions