Faster Spark ML training using accelerators

[inaccel <in...@inaccel.com>]

If you want to run much faster your Spark ML applications (up to 14x faster),
you can now use the Accelerated ML suite from  inaccel
<https://www.inaccel.com/>  on aws. The Accelerated ML suite allows to
offload the most computational intensive part of the ML tasks to FPGA
hardware accelerators without having to change your code at all (it supports
Scala, Java and python). Instead of running "spark-submit" you only have to
use "spark-submit --inaccel". The accelerated ML suite is packed as a docker
for seamless integration with Spark. You can test it on AWS using the f1
instances:  https://aws.amazon.com/marketplace/pp/B07KXQLTJ3and you can
achieve up to 14x faster training of Spark ML applications and up to 4x
lower cost. 



--
Sent from: http://apache-spark-user-list.1001560.n3.nabble.com/