FPGA '18 - Fpga

. .

Résumé :

We are delighted to welcome you to the 2018 ACM International Symposium on Field-Programmable Gate Arrays (ACM FPGA 2018). ACM FPGA is the premiere forum for the presentation of new and exciting research on all aspects of FPGA technology, which include:

• Novel FPGA architectures and circuits
• Advances in CAD tools for FPGAs, in areas such as technology mapping, placement, routing, and others.
• High-level design methodologies that permit FPGA design at higher levels of abstraction.
• New applications for FPGAs, particularly their use as accelerators for achieving higher computational throughput and energy efficiency.

Aside from the technical sessions, the conference provides the opportunity for FPGA researchers and practitioners from around the world to connect with long-time friends, meet new ones, and network with one another in beautiful Monterey, California, famous worldwide for its spectacular coast, Fisherman's Wharf, and Cannery Row.

This year we received 116 submissions, of which 26 were accepted as full research papers (10 pages) to appear in the main conference or the pre-conference special-session on deep learning, and 4 papers were accepted as short research papers (6 pages). All full and short papers appear in these proceedings. In addition, 31 submissions were selected to be presented as posters; abstracts of these appear in these proceedings.

At FPGA 2018, we continue to see an emphasis on the two themes causing tremendous interest in the FPGA industry: 1) the potential role for FPGAs in deep learning, and 2) the recent introduction of FPGAs in the cloud by increasing numbers of large companies, including Microsoft, Amazon, Baidu, Alibaba, Huawei, IBM and others. The panel discussion, at the Monday evening banquet, will consider whether FPGAs (as opposed to GPUs and CPUs) will ultimately succeed in becoming a widespread computing platform for deep learning. We expect a lively exchange among the panelists!

The symposium kicks off with back-to-back Sunday morning workshops on cloud-deployed FPGAs, and a parallel workshop session on using FPGAs for packet processing as specified in the P4 language. Sunday afternoon will focus on how FPGAs can be used for deep learning, with both invited presentations and research presentations. The use of low computational precision will figure prominently in the deep-learning research presented.