[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/biz/ - Business & Finance

Search:


View post   

>> No.52716176 [View]
File: 178 KB, 1077x939, icpthroughput.png [View same] [iqdb] [saucenao] [google]
52716176

>>52714146
Biggest piece of FUD which ICPshills can't squirm away from is a quite straightforward technical limitation. Because it is based on classical consensus protocols which have at minimum O(N^2) messaging complexity, throughput is limited to 500 user writes/s per subnet. Dfinity dev in the pic explains in more detail. The more nodes in a given subnet, the more messages need to be kept around and the lower the TPS performance gets. You actually saw this play out IRL recently with the SNS-1 launch. A couple subnets gagged including the NNS (which is the slowest of all because it's the biggest of all). There is no satisfactory way around this limitation. You can't just build a single dApp across multiple subnets because there's marked increase in latency as well as complexity, performance might be even lower than just a single subnet.

ICP also relies on permissioned nodes as a core of how the network operates and obtains sybil protection. There also is no getting around this without a drastic change of how the protocol works.

Nontechnically, Dom is super cringe and frankly seems unhinged when he goes off on his rants. He's fairly undereducated and the cryptoleaks/Christen Ager-Hansson shit is also absolute 5 year old tier tantrum throwing. And this dude is supposed to be in charge of the creation of "web 3.0" kek

>> No.52649138 [View]
File: 178 KB, 1077x939, icpthroughput.png [View same] [iqdb] [saucenao] [google]
52649138

>>52649081
Nope, that's the limit for any given subnet. What you're talking about was essentially a queueing issue.

Navigation
View posts[+24][+48][+96]