For some reason, this was even more evident to me when I looked again at recent work from CPDN. I raised it on the project boards and have come up with this based on a suggestion from one of the moderators.

FWIW, I exit boinc, edit the p_fpops value in client_state.xml for a more realistic floating point benchmark value. For example, my i7 4790K had a WINE fpops benchmark of about 1200. I edited it to 5200 and much better task estimates were the result.

On this computer I get the following when I look at the computer's details page on the CPDN site.
Measured floating point speed 1424.39 million ops/sec under WINE.
Measured floating point speed 2707.12 million ops/sec with native LINUX client.