According to Steam hardware survey:

https://web.archive.org/web/20190409092426/https://store.steampowered.com/hwsurvey/videocard/

In 2018 for Turing:

...... Nov/ Dec / Jan

2060..: - / - / -

2070..: - / .17% / .33%

2080..: - / .22% / .31%

2080ti: - / - / .15%

Total.: - / .39% / .79%

2020 for Ampere:

......: Nov / Dec / Jan

3060Ti: - / - / .27%

3080..: .23% / .48% / .66%

3090..: - / - / .23%

Total.: .23% / .48% / 1.16%

I imagine there could be some lags and some issues using steam hardware survey, but has long has those issues are similar over time it should not matter too much. around 3 more time 3080 than 3090 do seem to match the little public numbers few resellers gave us in the past, 3070 should rapidly get close to the 3080 numbers once they show up.

The number of monthly active users grew from 90m to 120m from 2018 to 2020, so achieving to get 1% of the users now should be around 33% more of them.

1.16 * 1.33 / .79 = 1.953 (almost twice as much)

If Ampere is significantly more popular on non steam gaming machine like miners, it could mean more than twice as much cards.

Pascal was much more popular, 4 month after it's launch the 1080 had .64% of the market, same has the 3080, but the steam user base was around half of what it is now, making it probably smaller in absolute numbers.

In total market share of the whole pascal family in september 2016 they had:

1070: 1.03%

1060: .65%

1080: .64%

total: 2.32%

If the steam user base was half of what it is now that would be the same numbers of Ampere, if Ampere is more popular among miners now than Pascal was at launch, it could be an higher amount of them were sold.