Intel releases new programmable acceleration card D5005, which has been adopted by HPE

Intel today announced the launch of the D5005 Programmable Acceleration Card based on the Stratix 10 SX FPGA, which has been used in the HPE ProLiant DL3809 Gen10 server.

Intel announced the launch of the Programmable Acceleration Card in 2017 as a comprehensive platform for using FPGAs in the data center, with the first product consisting of the 20nm Arria 10 GX. In 2018, Intel announced that it would launch the PAC with the newer 14nm FPGA, and today the product was officially unveiled, with Hewlett Packard Enterprise (HPE) jointly announcing the launch as a launch partner.

The new Intel FPGA PAC D5005 accelerator is the second generation in the PAC product family and contains a Stratix 10 SX FPGA. The PAC also comes with Intel's acceleration stack, which provides drivers, application programming interfaces (APIs), and an FPGA interface manager. It also contains four sets of 32GB DDR4 with ECC; it has two 100Gbps QSFP interfaces, and the card is connected via PCIe 3.0 x16. It has a 215W TDP.

Compared to the Arria 10 PAC, the programmable logic density is three times that of the Arria 10 PAC, the DDR4 memory is four times higher, and the Ethernet ports are twice as high, with speeds from 40GbE to 100GbE.

Intel's three accelerator cards, each with its own characteristics

Accelerator Card Product Comparison

Building a smart ecosystem for the future

HPE is the first company to announce that it will pre-certify the D5005 into its servers, with the OEM of the HPE ProLiant DL380 Gen10 server. Target workloads include streaming analytics, video transcoding, financial technology, genomics and artificial intelligence.

Currently, Intel is working closely with leading accelerator providers in various fields to create products that are best suited for FPGA acceleration and is committed to providing complete Intel solutions. These solutions take advantage of the powerful performance and versatility of FPGA acceleration and use the acceleration stack for Intel Xeon CPUs that include FPGAs. The acceleration stack provides a common developer interface for application and accelerator function developers. The acceleration stack works together with acceleration libraries and development tools to save developers time and enable them to reuse code on multiple Intel FPGA platforms.

As shown in the figure, Intel has partners providing acceleration stack applications in six major areas.

In addition to partner solutions, Intel also uses a broad ecosystem to unite a group of server OEMs and system integrators to ensure that customers and solution partners can take full advantage of powerful hardware acceleration and a flexible, easy-to-use software development environment to provide the broadest compatibility and enable large-scale deployment, such as HPE.

In addition, Intel has recently increased its cooperation in China, announcing the signing of long-term cooperation plans with Baidu and Lenovo, and will continue to cooperate in the fields of artificial intelligence, autonomous driving, 5G and high-performance computing to solve technological challenges in China and around the world.