Inaccel presented the new version of the scalable accelerators for Apache Spark on October 2nd at Xilinx's premier event XDF in San Jose. The new version of the ML accelerators for Apache Spark are scalable to a high number of FPGA nodes without and it is fully integrated with Spark. The new version of the accelerators in the cloud allows up to 3x speedup and 2x lower OpEx compared to contemporary processors.
Check the presentation from Inaccel at XDF in the following link:
Reload document 
Open in new tab