What is Kaspa max tps?

Key Context on Kaspa's TPSAchieved Peaks (Mainnet Records):September 14, 2025: 3,585 TPS September 17, 2025: 3,210 TPS September 18, 2025: 3,585 TPS (surpassing Solana's prior record of 2,909 TPS) September 28, 2025: 4,757 TPS October 2, 2025: 5,584 TPS (latest verified peak)

Theoretical vs. Real-World Limits:Kaspa's design supports sustained throughput around 3,000 TPS under normal conditions, with peaks exceeding this during high-activity bursts (e.g., mempool surges up to 500,000+ transactions). Unlike many blockchains, Kaspa's TPS figures are verifiable on-chain and exclude non-transaction messages (e.g., consensus artifacts), making them directly comparable to real user activity. Future upgrades (e.g., aiming for 32–100 blocks per second) could push theoretical max TPS to 10,000–32,000+, but as of today, 5,584 TPS represents the highest empirically achieved rate.

posted about 2 months ago

CC BY-SA 4.0

tty‭

1 reputation 0 2 4 1

Although @tty's answer is technically correct, it represents a peak TPS due to fluctuations in the block rate, which doesn't represent Kaspa's throughput over time. I think a more interesting metric is the average TPS, which can be calculated as BPS * MAX_TRANSACTIONS_PER_BLOCK (where BPS denotes "Blocks per second", aka the block rate).

To calculate what's MAX_TRANSACTIONS_PER_BLOCK, I'll try to explain a bit about a concept called "Block mass limit", which is similar to Bitcoin's block size limit.

In Kaspa, transaction "mass" is a measure of its externality on the Kaspa network according to several parameters:

1. Compute mass: this mass estimates the compute cost of processing a transaction. Some heavy operations such as signature verification will increase the compute mass (e.g. a multisig transaction will have a high compute mass).

2. Storage mass: This mass regulates the UTXO set growth. Generally speaking, a transaction that creates a lot of UTXOs, or some dust UTXOs will have a high storage mass. For more details see KIP 9.

3. Transient mass: this is a complete equivalent to transaction size, which is multiplies by 4 for normalization reasons. For more details see KIP 13.

The consensus enforces a limit of 500K per block for each one the above.

So after this long intro, let's do some math:

• The smallest possible transaction is a 1 input 0 outputs transaction, which has a transient mass of 584 (and compute mass 146, but it doesn't matter), so a block can carry 856 of those, and therefore if all blocks are filled with them the average TPS will be 8560 (BPS=10). But this type of transaction has very little use, so this figure doesn't say much.

• Most transaction have 1 input and 2 outputs, where it spends and creates P2PKs UTXOs, and this kind of transaction has compute mass of 2036, and therefore the network can support average TPS of 2450 of those transactions.

• Transactions that mostly utilize the payload (e.g. L2 transactions) will mostly have 1 input 1 output. If we assume a payload of 100 bytes, the transient mass will be 1452 and the network can support an average TPS of 3440 of those transactions.

To summarize, the average TPS can be as high as 2400-3400, depending on the scope of use of L2.

P.S.

Due to the BlockDAG's parallelism, some transactions are expected to be duplicate, where the number of duplicate transactions is expected to go down as long as the mempool backlog is bigger.

For more info read:

• https://ancapalex.medium.com/transaction-selection-games-in-blockdags-602177f0f726
• https://functor.network/user/912/entry/953

posted about 2 months ago

CC BY-SA 4.0

someone235‭

1 reputation 5 8 7 3