I first noticed the issue at about 1am (25th Jan), when I was downloading a maintenance update to an Asterisk box.
From looking at the logs from another machine, the problem was present at or before 11.30am (24th Jan), but I can't be any more precise than that.
I dropped and restarted the PPP session, from my end several times, without any improvement, so I logged a service call.
The line re-sync'd at 4am and single threads were back up to full speed.
The building has another GEA- FTTC connection, provided by KCOM, from the same cabinet, which wasn't affected.
I ran the speed test from a Windows 10 laptop, but the Asterisk box runs under CentOS and the other machine a custom build of Debian and all were suffering, so not a Win 10 specific issue, although thanks for the suggestion.
Lizzie