@ramcenterfun
RAM CENTERcreator rewards → gigabytes
Every creator reward this coin earns is claimed, converted and spent on real memory compute: RAM rented by the gigabyte-hour and allocated to a public node. Automatically, every cycle, for as long as the coin trades.
–
RAM CENTERround #–
The side quest: your device will lend spare memory to the pool from this tab and the round pot gets split by verified gigabyte-seconds. The real capacity lives on the engine.
waiting for a checkpoint0% of the 0.01 SOL payout threshold
RAM CENTERleaderboard
Who is lending, and how each round's pot is shared.
Share is your verified memory work in the period divided by everyone's. A checkpoint at page size d counts as 2d bytes served.
RAM CENTERstats
How the pool is doing: throughput, lenders, pots and payouts over time.
RAM CENTERrounds
Every settled round. Open one to see exactly how its pot was split and paid.
RAM CENTERbenchmark
Measure this device: 5 seconds of memory work on every CPU thread, then 5 seconds on the GPU. Nothing is lent or sent while it runs.
Self-reported by the browser that ran the test. Shown for comparison only; it never affects credit.
RAM CENTERledger
Every payout, with its transaction. Balances are paid at round close once they cross the threshold.
RAM CENTERmemory
The coin's creator rewards buy real RAM capacity from a compute provider and the gigabytes are allocated to a public node. Every hop is a receipt: Solana transactions, a provider allocation you can watch drain, and the node's capacity reported live.
The engine wallet claims the coin's creator rewards from pump.fun. SOL, on chain.
Everything above a gas reserve is swapped on Solana through Jupiter into a dollar stable (USDC), the unit the compute provider bills in. One transaction, the signature is the receipt. That balance is the compute budget.
The USDC is sent to the compute budget wallet, a separate public address shown below. It only ever receives: nothing is spent from it by hand. Its balance is the budget behind every gigabyte, and anyone can check it on chain.
The budget is priced against Vast.ai's live market: verified machines with at least the target RAM under a ceiling per hour, cheapest first. Each machine goes on the book covered for a full term at its live $/h; when the budget covers another one, another goes on. If the budget stops covering a machine, it comes off. Fees in, gigabytes out, no subscription.
Gigabytes online is the RAM of the machines on the book; paid-for time is the budget divided by the book's hourly burn; memory served accrues by the hour. Every number here comes from the wallet balance and the market, nothing is typed in by hand.
Every 10 minutes.
RAM CENTERhow it works
From a random seed to SOL in your wallet.
Your tab asks the server for a page bound to your address: a random 32-byte seed and a page size in bits. Tickets last two minutes.
Web Workers on your CPU and a WebGPU compute shader on your GPU fill memory pages derived from seed ‖ counter, looking for a page whose checksum starts with page size zero bits.
A hit is sent as its counter. The server checks it with a single checksum. Weak, reused, expired or borrowed checkpoints are rejected.
A checkpoint at page size d is credited 2d bytes served, the amount of memory it takes on average to find one.
Each address gets its own page size, tuned for about one checkpoint every few seconds, so a phone and a workstation are measured the same way.
At round close the pot is split by share of verified memory work. Balances over the threshold are paid out and logged in the ledger.