The title "7 million transactions per second" for Stellar was copied around again today by a fresh wave of repost accounts. That ALL-CAPS JUST IN post got more than 20,000 views and over 1,000 likes, and after scrolling through the comments, hardly anyone noticed the second paragraph.
To break it down:
This is Project Thruput, a technical preview running on the FutureNet testnet. In The Defiant’s exact words, it is "a separate network," not an upgrade to the XLM mainnet.
How did they get to 7.02 million? One validator node did 1.18 million transactions per second on a single machine, and with 8 machines attached it reached 7.02 million. The test was of the execution engine, not end-to-end across the whole network, and there was no consensus, no real users, and no real traffic.
I’m not here to trash the technology itself. Parallel processing for same-state transactions is meaningful, and Mazières also open-sourced the code.
But benchmark runs like "add machines, add throughput" can be produced by any chain in a lab. We all remember what Solana’s 65,000 TPS looked like in actual practice, right?
The only real questions are these: when will it go to mainnet, and once it does, with zero per-transaction fees, how will validators and the network be sustained? Until those two questions are answered, 7M is just a slide number from the Meridian conference. If someone is using it to shout that XLM is about to take off, they’re most likely waiting for you to be their exit liquidity.
#XLM #Stellar
To break it down:
This is Project Thruput, a technical preview running on the FutureNet testnet. In The Defiant’s exact words, it is "a separate network," not an upgrade to the XLM mainnet.
How did they get to 7.02 million? One validator node did 1.18 million transactions per second on a single machine, and with 8 machines attached it reached 7.02 million. The test was of the execution engine, not end-to-end across the whole network, and there was no consensus, no real users, and no real traffic.
I’m not here to trash the technology itself. Parallel processing for same-state transactions is meaningful, and Mazières also open-sourced the code.
But benchmark runs like "add machines, add throughput" can be produced by any chain in a lab. We all remember what Solana’s 65,000 TPS looked like in actual practice, right?
The only real questions are these: when will it go to mainnet, and once it does, with zero per-transaction fees, how will validators and the network be sustained? Until those two questions are answered, 7M is just a slide number from the Meridian conference. If someone is using it to shout that XLM is about to take off, they’re most likely waiting for you to be their exit liquidity.
#XLM #Stellar