2018-06-30 00:20:58
I think 1024MB is decent for "small"

Clueless
2018-06-30 04:10:07
it's big for an array... if the JVM doesn't break up arrays (I don't think it does) it means the OS needs to find that much contiguous memory.

Tyler Boone
2018-06-30 04:39:23
do you think your users are scum?

Fireduck
2018-06-30 04:39:52
this code is basically insane

Fireduck
2018-06-30 04:53:00
Do we think ImmutableMap is fast or should I use arrays?

Fireduck
2018-06-30 04:53:11
I am double thinking everything probably in a not useful way

Fireduck
2018-06-30 04:55:15
run a profiler nub

Tyler Boone
2018-06-30 04:55:26
I don't have time for that

Fireduck
2018-06-30 04:55:47
when I profiled the existing miner all the time was spent deep in the hash libraries

Tyler Boone
2018-06-30 04:56:03
yeah, and I want to keep it that way

Fireduck
2018-06-30 04:56:14
meaning not fuck up and add dumb shit on top of that

Fireduck
2018-06-30 04:56:59
MinMaxPriorityQueue looks sweet

Fireduck
2018-06-30 04:57:11
https://google.github.io/guava/releases/19.0/api/docs/com/google/common/collect/MinMaxPriorityQueue.html

Fireduck
2018-06-30 04:57:14
exactly what I want

Fireduck
2018-06-30 04:59:33
some time was spent copying bits into bit holders

Tyler Boone
2018-06-30 04:59:46
that is where we keep them

Fireduck
2018-06-30 05:00:00
ideally you would pass the existing bit holder into the hasher directly

Tyler Boone
2018-06-30 05:00:17
I do that where I can

Fireduck
2018-06-30 19:55:11
ok, tuning this thing is like riding a stack of unicycles

Fireduck