Splatter

@0xSharples this looks like it wasn't even intentional (aside from being a normal arb bot)

Looks like someone exploited a pricing bug on ZeroFi's DRAM/USDC pool yesterday. Over $4.7b was traded yesterday in this market, with around 400k transactions utilizing flash loans to buy DRAM and then immediately sell it back for a small profit. In total, around $45k was stolen from a trading volume of about $4.7b

@eightbitsim @mert @alessandrod yes. a bundle usually does that

@SplatterSOL @mert @alessandrod source: solscan.io/tx/3A92skc3iUv… backrun: solscan.io/tx/2iK2kpsZqHs… (n, n+1) every time good luck out there @SplatterSOL

so much midcurving around the helius backrunning thing it's crazy > 100% of the block producers do MEV, not all MEV is the same > a lot of stake doesn't even accept TPU transactions anymore > it would have been so much easier for them to switch to a MEV block builder ages ago, delay, rev strat, maxxextract, etc > instead heroically they DID NOT, and remained pretty much the only big validator to stick to the default agave scheduler, no delayers, no bs, this has been GREAT for the network and has objectively lost them money > they've been of HUGE help in improving agave perf. I've profiled and looked at their validator countless times, and all along I've been telling liam I'd personally FUD him to death if I caught him doing sketchy stuff > now they find a way to increase rewards _without_ doing all the dodgy shit everyone else is doing, with a KYCd service which arguably does the least worst MEV > people on twitter are mad ????

@eightbitsim @mert @alessandrod You're not making much sense. also you have shared no evidence that every arb is even caught..

@SplatterSOL @mert @alessandrod every single time, is this randomness?

first of all, you never HAVE to hold a tx. The fact you see it early already gives you the advantage (unless a crank arb bot gets lucky) -- the amount of profitable crank bots would also decrease if more stake would backrun :p I also highly doubt anyone is holding transactions (even for just a few ms). They simply submit their arb in a bundle with the backrunnable tx. That is just jito / harmonic doing it's job. And for the 0 prio fee ones, it's likely just a private way of inclusion that skips the regular fee auction.

This ends now. Even if I must drag the gods down with it. lets go over simple set of assumptions i am making @mert let me know which one doesn't make sense. I bill 10SOL per slot for consults. 1. the validator has to periodically release shreds [thus txns reaching the TPU have to pass through without delaying] - Network requirement/Social consensus 2. ⁠Backrunning any txn - Involves route searching across the chain state - takes 10-20 ms atleast [including latency even if co located] 3. ⁠A backrun can be done in two ways -> either the source is passed through and the searcher only gets a signal and backrun goes separately [CASE 1] or you bundle the source and backrun [CASE 2] 4. ⁠in CASE 1 [10-20 ms delay = txn will be several positions apart] CASE 2 [src,backrun will be n and n+1] 5. ⁠As multiple txns in this thread have shown, there are CASE 2s with this searcher on bald slots. 6. ⁠Thus to do Case 2; you have to hold the source for 10-20 ms 7. ⁠Which means, the price of execution [within users slippage tolerance] - would differ than what he could have gotten in case 1 8. ⁠CASE 1 is fine, CASE 2 is not 9. ⁠@toly and you both are trying to make a fool of your audience, letting them think you are doing CASE 1, but doing CASE 2. Because 2 makes much more money per backrun vs CASE 1 [coz your guy mriya, gets a priority lane behind the user] coming up with straight answers is rough, requires acceptance, balls, agency & courage. hope you'll use it this time. @trentdotsol @sonic_from_ny @toly @uriklarman @bencirc @0xrafal @alessandrod @niteshnath @bigtimetapin @helius

@eightbitsim @solana @Helius Right but this deal is backrun only. If anything this helps the chain by making crank arbs less profitable.

I understand the economics behind the deal, but the question is should we be promoting out of protocol deals or not at @solana Out-of-protocol deals aren't limited to backruns. They extend far beyond them and can easily introduce toxic behavior. By definition, this falls outside the scope of the protocol.

Two @solana validator operators are caught red handed running private inclusion lanes for one of the biggest MEV bots on chain. The backrunner MRiYA4oN3158fCV8evhuCofrDzbHyYvYnGZUDJvoCsa for the backruns analyzed: - Pays zero priority fee → 84% success on @Helius leaders, 80% on Kiln - Same bot, same txs, every other validator: 25% Not SWQoS. Not luck.

@eightbitsim @solana @Helius Sure, but is it someone right to land just because they’re high paying and efficient? The only reason a validator would take a deal like this, is if they earn more than they’d make on priority fees for those same opportunities

@SplatterSOL @solana @Helius backruns shouldn't be a problem as much prioritized backruns is, someone efficient and high paying is not able to land due to these cartel deals.

Well well well 😂

@cavemanloverboy I guess it’d depend on the scheduler. But to me it seems like they’d get more priority with less base fees this way

@SplatterSOL they all pay now. no revert protect

WE ARE GOING TO CREATE SO MUCH REV, YOU ARE GOING TO GET TIRED OF ALL THE REV. PEOPLE WILL SAY, "PLEASE, SIR, NO MORE REV." I WILL SAY, "WE HAVE TO KEEP REVVING." github.com/solana-foundat…

@cavemanloverboy shoutout to the llms

@SplatterSOL they do know how

the crank arb bots are **checks notes** 1. requesting 262kb of heap 2. over-requesting CUs 3. write locking 33 accounts 4. loading 19 MB of accounts merge the resource fee SIMD with haste lol solscan.io/tx/5KSqpceBD4u…

hey validators, lets make solana faster by enabling xdp. xdp improves turbine packet handling by cutting packet overhead, copies and context switches. more shred propagation headroom = better validator performance under load and more room for solana to scale. add: --experimental-poh-pinned-cpu-core \ --experimental-retransmit-xdp-cpu-cores \ --experimental-retransmit-xdp-zero-copy bnxt_en: --experimental-poh-pinned-cpu-core \ --experimental-retransmit-xdp-cpu-cores \ --experimental-retransmit-xdp-interface \ extend your systemd service: AmbientCapabilities=CAP_NET_RAW CAP_NET_ADMIN CAP_BPF CAP_PERFMON CapabilityBoundingSet=CAP_NET_RAW CAP_NET_ADMIN CAP_BPF CAP_PERFMON LimitMEMLOCK=2000000000 notes: - dont run poh and xdp on the same physical core - keep poh away from the xdp cores shared L3 / numa domain - validate your actual topology with lscpu -e/hwloc or solanahcl/solanahcl/blob/main/public/topology/ easy rule of thumb: - 7965wx: 24c across 4x ccds -> 6 cores/ccd first ccd: 0–5c, next ccd starts at 6c - 9375f: 32c across 8x ccds -> 4 cores/ccd first ccd: 0–3c, next ccd starts at 4c common xdp-capable drivers: i40e, ixgbe, ice, igc, mlx5_core, mlx4_core, bnxt_en dont use bonded interface with xdp. for broadcom / bnxt_en: dont enable zero-copy. copy mode is the safe path. if you run into "huge page alloc failed", remove the zero-copy flag and set the xdp interface to the physical interface: --experimental-retransmit-xdp-interface check the ring sizes: it needs to be power of 2 IFACE=$(ip route get 1.1.1.1 | awk '{for(i=1;i<=NF;i++) if($i=="dev") print $(i+1)}') sudo ethtool -g "$IFACE" # if you see values like rx: 511 or tx: 511, thats likely the issue. sudo ethtool -G "$IFACE" rx 512 tx 512 restart and voila, you are not a potato anymore.

@adlonymous @faremeterxyz returns all the data you need in the extra field so clients don't need an rpc connection

a pain point of giving autonomous agents a solana wallet rn if i want it to sign x402 transactions, it needs an RPC, which it can access with x402 per-call but it needs to pay for the x402 endpoint, which it can't pay because it doesn't have an rpc endpoint yet

@durov This seems a bit cherry picked gng

🏁 TON leads Layer-1 blockchains in finality time. Sources: telegra.ph/Comparison-of-…

@jarxiao @toly @solana How did this Happen? Did they censor prop amm oracle updates in the prior blocks?

This validator made 7.5k in 1 slot and 17k over the full leader window. The majority of validators are good actors, but @solana needs to be resilient to malicious nodes. In-protocol ordering rules and enforced censorship resistance are neccessary for a healthy on-chain economy

we still get looksmaxxed on frontend a little but we IQmog hard now

why. in 2026. does your webiste show me a phantom login popup each time I click something..

@0xCypher_ @temporal_xyz Big

joined @temporal_xyz. best engineering team in the space. excited to build things that are good for us and good for solana. i think those can be the same thing more often than people assume.

Market data over the internet works. Solana shreds over DoubleZero Edge wins. DoubleZero Edge beta is live. ↓