We're stabilizing a new version that improves latency by (reduction of ~85%), while guaranteeing much higher execution throughput (up to 10x) during peak times.
But given the current volatility of #crypto markets not sure when we'll decide to release it.pic.twitter.com/zxlPF1uYTo
-
-
Show this threadThanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
yeah but how often do you restart your ethereum nodes
-
For sure more frequently ---- ps -eo pid,lstart,cmd | grep geth 23796 Thu Sep 17 13:14:40 2020 bin/geth... ---- While here's a screenshot of bitcoind when I finally decide to restart itpic.twitter.com/fOQBWh9zuW
End of conversation
New conversation -
-
-
That is insane uptime for a Raspberry Pi!
-
Lol! I test quite a lot the matching engine on a Raspberry PI 3B+. It's quite nice to test on slower hardware just to find bottlenecks more evidently.
- Show replies
New conversation -
-
-
I'm curious: what is the implementation language? Secondary: has the choice of implementation language contributed towards this uptime?
-
Core parts (bin trees, net stack) are in C++, while most of non-critical is java. Note hat usually most bottlenecks are actually not languages but developers' wrong choices (inefficient data delivery queues, wrong data structures, ...).
- Show replies
New conversation -
-
-
So 300 days with no update? ;)
-
No update nor restart
End of conversation
New conversation -
-
-
You have one matching engine process per pair or it's one for the whole exchange ?
-
Our trading engine is heavily configurable. At the moment we shard pairs across multiple processes/threads based on CRC32(pair) % X, where X is how many cores we want to reserve for sharding.
- Show replies
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.