Skip to content
Archive

Post

Back to @armaniferrante

A armaniferrante
Armani Ferrante
@armaniferrante

Stop FOMOing Hyperliquid. If you actually believe these things then rip out the SVM and colocate in a single data center. You need 50 microsecond blocktimes, not 200ms.

@

he's right. he got me. my proposal is opinionated. the truth is that I don't want solana to be a general purpose blockchain; i want solana to be the blockchain that eats up the world of finance. a general purpose compute platform does not really need fast finality. a competitive globally distributed financial system does. a general purpose compute platform does not need airtight censorship resistance guarantees. a competitive globally distributed financial system does. a general purpose compute platform does not need taker speed bumps or generalized application controlled execution. a competitive globally distributed financial system does. a general purpose compute platform does not need to be robust to extreme spikes in traffic, as it can just queue up jobs for later. a competitive globally distributed financial system does. to eat up the world of finance, solana will need to: 1. lower fees for market makers 2. add friction for takers 3. have enough bandwidth to support all financial activity 4. provide censorship resistance 5. provide 100% uptime with no degradation the resource fee, whose primary objective is to encourage developers to optimize their systems, partly addresses the first three in one fell swoop and is synergistic with future network upgrades. that is, the resource fee: 1. lowers fees for market makers for both spot (clob and prop) and perps (prop and spline). 2. raises fees on takers. 3. puts a firm incentive for apps to optimize their onchain computational footprint so yes, i am opinionated: i want 120 ms slots for solana. i want alpenglow for solana. i want ace for solana. i want mcp for solana. i want the resource fee for solana.

· 30K Views

5 Reposts 3 Quotes 191 Likes 21 Bookmarks
replies reposts likes
34 replies collected of 37 X reports
MAD888_맫가 ('mad Guy')⛏️ @mad_guy888 · B

@armaniferrante Solana is on a whole different level

original · ko

@armaniferrante 솔라나는 차원이달라

1
帅帅🐂🀄️ @ssovoovo · 31K A
1
老黑.. @yle43036 · 13K D

@armaniferrante Forget general purpose, go all‑in on finance use cases

1
스카이 DeFi @Defi_Lofi_ · D

@armaniferrante 50 microseconds gets you a matching engine, not a global financial system

Quackduck.Sol🎒 @quackduck_sol ·

@armaniferrante Please don’t dump on my bags, Backpack is very good and hyperliquid is very good, let’s just enjoy this for a bit And yes my backpack bag >>>>> hype pouch, but please don’t dump on that pouch Thank you for your attention to this matter

1 2
Capy Research @Capy_Research · D

@armaniferrante finance needs distribution too

1
great hylord cavey phd ∿ @cavemanloverboy · A

@armaniferrante hl does not have half of the things I mentioned and i didn't even mention hl at all. in this tweet, in this proposal, anywhere.

2 15
Solana allday @Solana_allday · 21K D

@armaniferrante Those block times change everything for traders.

1
Leon @Leon_Defi · 10K D

@armaniferrante microsecond latency is a pipe dream

1
Lou @0xLouz · 25K D

@armaniferrante 50 microseconds wins the colocation demo, not a globally distributed financial system

milly.hl @milly_rock7 · D

@armaniferrante If 50 microsecond blocktimes are the real need, then a single data center setup is the only way to get there thats not a blockchain anymore, thats a trading venue with extra steps

1
Sharptake @ssharptake · D

@armaniferrante so the real ask isnt general purpose speed its a colocated exchange with a chain attached

Knox @xknoxbt · D

@armaniferrante cmon bro don’t do that it’s racism

1
sayuj kb🦂 @sayuj_kumar · D

@armaniferrante colocating in one data center kills the whole point though. finance wants speed, everyone else wants decentralization, hard to serve both

1
Mad Professor 🎒 @ivancsicsakos · C

@armaniferrante Totally agree, if high-frequency trading is the goal, decentralization trade-offs don't make sense and true low latency requires dedicated, colocated infrastructure.

1
behzad @behzadnouri · D

@armaniferrante The problem is they are toast if they do not FOMO Hyperliquid. It is a lose-lose situation for solana.

1
Shubh @01Rajpurohit · C

@armaniferrante I would love your perspective on this: @fogo offers 40ms on SVM with DFBA(dual flow batch action) so what's something missing from it that it can't complete with Solana?

1
NasibLoc | On-chain Pass Draws @NasibLocapp · D

@armaniferrante Centralizing in a single data center yields sub-millisecond latency but destroys censorship resistance. Building high-performance DEXs requires optimizing state machine execution and deterministic clearing engines, not abandoning decentralized consensus to recreate AWS.

Kian ($/acc) @klirphyy · A

@armaniferrante I didn't see Hyperliquid on the quoted tweet or pertaining to it?

1
~Rex @Rex_Ogjy · D

@armaniferrante sol is the legend so far

1
django @oxngon · B

@armaniferrante Incidentally, exactly something @blinktrade is building afaik

1
theprocrastinator @pr0crastin80r · D

@armaniferrante Hyperliquid

1
DeepSOL @drdeepSOL · 12K D

@armaniferrante 50 microsecond blocktimes sounds like a single data center tradeoff that kind of speed usually means giving up the distributed part

Diego Rodriguez @tibia_roman · D

@armaniferrante 50 microsecond blocktimes only work if the whole system is colocated At that point you've traded decentralization for a single point of failure

Leo Isakov ⭐️ @daiteskorosti ·

@armaniferrante you’re preaching to the choir solana’s the venue, not the game

1
suckmydiscoteque.eth @bullrungenius ·

@armaniferrante Stop FOMOing Hyperliquid. Prepare FOMOing Backpack

1
Richard | ISC @Richard_ISC · C

@armaniferrante Yes pls. Global distribution isnt required for censorship resistance. You just need to be willing to move.

Olivier BRUNO @OlivierBRUNO13 ·

@armaniferrante The thing is, this SIMD doesn't actually make the network faster, it simply brings about a mechanical reduction in on-chain transactions via fees.

1 1
Dan @L_Earnings ·

@armaniferrante fogo spotted

1
Willd ⛏️ @willdxyz · C

@armaniferrante A surprisingly refreshing take - now talk about your own thing, is BP going to house all of finance?

1
MAD Vincent 🎒 @MadVincent666 · 31K A

@armaniferrante Wen like Nasdaq <1 ms? Askin for hft quant frens

1
Jay @jayendra_jog · 26K B

@armaniferrante whats the point of a blockchain at all if you're essentially going to run as an AWS server

13
Validatus.com @stakeingermany · D

@armaniferrante For finance, chasing the lowest possible latency probably matters more than trying to make the chain do everything.

Lisa Florentina @LisaFlorentina8 · 13K D

@armaniferrante Totally agree! Latency is key; optimizing block times can make a huge difference.

13 replies whose parent comment X withheld

Quackduck.Sol🎒 @quackduck_sol ·

@armaniferrante If I said what you said it would be saying, when YOU say it, it’s dumping, your words hold a lot more weight

1 1
Armani Ferrante @armaniferrante · 166K D

@quackduck_sol I’m very confused

2 2
great hylord cavey phd ∿ @cavemanloverboy · A

@armaniferrante what the hell are you talking about brother this conversation was started because there is an enormous amount of anchor slop on chain. thank you for that

3 18
great hylord cavey phd ∿ @cavemanloverboy · A

@armaniferrante literally cleaning up your mess

1 3
Quackduck.Sol🎒 @quackduck_sol ·

@armaniferrante Yea I know, double standards from me, but it is what it is. 😂

1
Armani Ferrante @armaniferrante · 166K D

@cavemanloverboy Has nothing to do with your original tweet but ok 👍

1 5
Alessandro Decina @alessandrod · B

@cavemanloverboy @armaniferrante but who cares that there is anchor slop? things don't slow down because of anchor, they slow down because of mispriced syscalls and other bs

1 1
great hylord cavey phd ∿ @cavemanloverboy · A

@alessandrod @armaniferrante things slow down because of anchor. please measure the wall clock time of 1 anchor (classic, not 2.0) amm swap versus a more optimized contract

1
Alessandro Decina @alessandrod · B

@cavemanloverboy @armaniferrante but does it show up as a bottleneck anywhere?

1
Alessandro Decina @alessandrod · B

@cavemanloverboy @armaniferrante Not sure how this is showing that anchor is a bottleneck?

1
great hylord cavey phd ∿ @cavemanloverboy · A

@alessandrod @armaniferrante the #1 program on this chain by cu (35%) uses anchor brother

1

These were collected in full; the comment they answer was not returned by X.