Larry Ruane

Reviewers, please take a fresh look at the latest force-push; this PR is much better than before. I'll explain from scratch, so it's not necessary to read any of the...

Just to clarify, the last bullet point in the [review club writeup](https://bitcoincore.reviews/16981.html): > Before this PR, we would always read entire blocks from disk into our buffer. If the predecessor...

@vasild, thanks for the timings, I like both of your suggestions ([here](https://github.com/bitcoin/bitcoin/pull/16981#discussion_r403694969) and [here](https://github.com/bitcoin/bitcoin/pull/16981#discussion_r403695617)) so I just force-pushed both of them. I also fixed the reindex functional test, which I...

@vasild > Do you think it would be possible to demonstrate that it produces identical result as `master`? That is - reindex with `master`, dump the leveldb database from `blocks/index`...

I had an idea for a further speedup, but it didn't work out as well as I had hoped. It's about 10% faster, but probably should not be merged because...

Thanks, @fjahr, good idea, I just force-pushed three commits to replace the previous single commit so the changes are separated out more clearly. I also just updated the description, which...

Force-pushed and rebased to fix merge conflicts. This PR had momentum for a while but that's faded. Reviewers may be put off by the number of comments, but the result...

@vasild, good suggestion, done (squashed and force-pushed). I re-ran the timing experiments again since a lot has changed. I started the client with `maxconnections=0` (so it wouldn't be bothered by...

Force-pushed rebase to eliminate merge conflict.

Force-push to fix comment ([review suggestion](https://github.com/bitcoin/bitcoin/pull/16981#discussion_r459756399)), also took the opportunity to improve the test by checking for expected text in `debug.log`.