Answering
this deleted post, I'm currently running this:
for file in *.tsv.gz; do gunzip -c $file | grep -v is_from_coinbase | cut -f 7 >> /tmp/addresses.txt; done
It takes a while, but doesn't consume a lot of memory. When done, I'll sort | uniq the file and get the result. Now that I think about it, I could have piped the whole thing through sort in the same line instantly.
When I have a bit more time, I'll create daily updates for all used Bitcoin addresses and make it available for download. It's going to be a big file though.
Blockchair's /bitcoin/outputs currently takes 106 GB, and grows 2 GB per month. At 100 KB/s, it takes just under 2 weeks to download from Blockchair. For
$32 per year, I can run a VPS in Germany with 1 Gbit connection, enough disk space to keep up for a few years, and enough bandwidth to allow 9 downloads per month. If anyone can use this, let me know.
The list with all addresses is 49 GB in size. If you tried to load it to RAM, that's probably why you ran out of memory.
Total address count: 1,484,589,7491... address count: 1,039,899,708
3... address count: 343,485,961
bc1q... address count: 55,006,904
...-... (with a "dash") address count: 46,197,161
Unique address count:1... address count: 470,943,308
3... address count: 167,941,821
bc1q... address count: 39,137,878
...-... (with a "dash") weird address count: 15,157,808
And here it stops for now: after processing data for 5 hours, I made a mistake and accidentally overwrote my end-result. I'll restart later.
I'd like to see which address has received most transactions.