We run E10.0.700.3 at the moment (1 Server 2012 VM that runs SQL 2012, 1 Server 2012 VM that runs the E10 appserver, and 1 Server 2012 VM that runs RemoteApp for the E10 client - all on the same ESXi host) and will likely move to E10.1 in the next few months. We actually found that at sites with low internet bandwidth, RemoteApp is a little slower than when the client is installed locally. One site running a T1 told us that, and we confirmed it to be the case. It may have had to do with the nice metro interface in E10. The sites with 35/5 coax or 10/10 fiber seem to run RemoteApp about the same as installing the client locally.
I actually heard someone say at Epicor Insights 2015 that Epicor 10 ran better when installed locally on client machines connecting over site-to-site VPN. We never really had to test that theory other than at the site with a T1 that I mention here.