I ordered BT package after finding the solution to this thread describing how BT offer "standard", "stable" and "super stable" options on their customers internet connections to better optimise their connection depending on what services they like using, especially cloud gaming services.
Modern cloud gaming services use highly variable throughput UDP traffic which stresses internet connections out in a different way compared to more traditional traffic. Which is possibly why those "stable" and "super stable" options have been provided. BT were first ISP to offer cloud gaming service package with broadband in Europe so they may have more insight into the needs of cloud gaming services compared to other ISPs.
But i am concerned that many other more recent posts in BT Community with folks asking for help with weird ping spikes, packet loss and high latency simply get ignored and nobody in BT Community recommends asking for "stable" or "super stable" options.
So was hoping fellow cloud gamers who are currently using BT can show me their GeForce Now network test results to make me more confident that my new internet connection is indeed going to prove to be an investment as my current situation looks like
For me traditional speed test (ones not optimised for the larger packet sizes and higher frequency for cloud gaming scenario) say my internet connection is good.
Is streaming over BT network recommended by GeForce Now?
Did you need the "stable" or "super stable" option enabled?
Solved! Go to Solution.
What device did you run the rest on? Also what is "ootb"?
I just ran a test on my shield (ethernet) and my phone (wifi) and I still get reduced bandwidth and high packet loss on my Full Fibre connection
In screenie above I was using windows 10 pc . "ootb" is intended to mean "Out Of The Box" as in it just worked without doing anything on my end.
Here is what Nvidia Games Hub (GFN)'s native speed test looks like on old Shield TV Pro (2015) using BT connection
TL:DR Did you ask BT to set "super stable" option on your connection - like what is described in threads linked to in OP?
Have to admit i did mention to Openreach engineer when he doing wiring in home i needed "super stable" QoS on my connection as i was going to be cloud gaming. He then remotely configured my router and internet connection due to COVID restrictions so not sure if he applied this "super stable " thing or not or did anything at local box or local exchange that could have helped.
If you have a windows pc try using winmtr to see route taken. Zayo.com seems to be cause or common denominator where folks are seeing lower than expected bandwidth along with packet loss in GFN service. Zayo.com is Equinix peering partner where Equinix is Nvidias peering partner.
Changing ISP change route and removed zayo.com hops from the equation and fixed the lower than expected bandwidth and packet loss issues - completely.
Nvidia GFN community like to blame ISPs but Nvidia could possibly help more than they are.
So this is what winmtr says route is taken when pinging and tracert'ing GFN london server from my end when using BT connection
Using another ISP shows packet loss along with lower than expected bandwidth being reported in GFN network test coincides with hops involving zayo.com
Oddly using the "analysis" button in thinkbroadbands speedtest shows my old ISP scores very good for down and up bufferbloat but not so good for streaming (single thread aka tbbx1) whereas BT connection scores not so well for up and down bufferbloat but has awesome streaming (single thread/tbbx1) bufferbloat score
And even more oddly GFN games still report massive packet loss if you do not streaming quality parameters knocked way down below available bandwidth - regardless of what GFN network test says - again showing there are server side issues and/or issues with GFN service causing packet loss and weird networking related issues.
To get 0 PL and 0 FL
i had to set "streaming quality" too
as setting "max bitrate" to anything higher than 30 mbps causes PL and FL in game session.
Nvidia support really suck