5/ @bullx_io makes spotting early gems easy.
➥ Go to: https://t.co/kJYPdNjogw
➥ Press "Start," then "Login" to access the site.
Here’s my filter setup:
• Volume: $2,400+
• Market Cap: $7,000+
• Holders: 10+
• Developer’s Share: ≤ 2%
With this setup, you’ll filter out the noise and focus on real opportunities. 🚀🧵👇
Risk is the engine that drives progress In the world of high-stakes sports betting, losing isn't just about losing money – it's about learning from failure and
A feature of the XRP Ledger network was used in an unsuccessful exploit attempt on the prominent crypto exchange Bitfinex. By @shauryamalwa.
https://t.co/NCmKJXYFfO
Last weekend was amazing to say the least. I had the privilege of being able to honor one of our Founders, Founder Fannie Pettie Watts, in a tremendous way. As many if you know, Founder Watts and I had a very close relationship-truly honored
#zetaphibeta#zphib#flashbackfriday
Took a couple weeks longer than expected because I found other bugs along the way, but here's some numbers on the XDP work I've been doing.
tldr: turbine can now retransmit much faster and is not a bottleneck anymore, and does so using a fraction of the CPU we use today.
As I mentioned a while ago we've been stress testing turbine in preparation for making much larger blocks. In the tests we made turbine retransmit 1M+ shreds per second, with peaks of 2M+, at an avg of 15Gbps.
I attached profiles of before and after my fixes. Left all yellow screenshot is master, right screenshot is my PR.
In master you can see 12 cores (solRetransmitXX) running at 100% calling sendmmsg() transmitting shreds. 12 cores isn't enough, so shreds are actually propagated too late.
With my PR, there's 4 cores being used for retransmit which are _idle_ most of the time (when they're not doing the turbine shuffle). There's an extra thread (solRetransmitIO0) driving the NIC with XDP to send shreds, which is sleeping all the time because we have _a lot_ extra capacity. Shreds are transmitted in time, no added latency.
There are other minor bugs I found while working on this which I'm gonna fix next, but this was the largest bottleneck in the way of doing much much larger blocks, and it's now gone.