What exactly is changing?
Solana is cutting the target block production time (slot time) on mainnet from 400 milliseconds to 350 milliseconds. According to a CryptoSlate report, this change activated at the start of epoch 1019, but because of what is known as a one-epoch delay, the network keeps its existing parameters and the new 350ms target only takes effect from epoch 1020.
Just to clarify the terms: slot time is the target interval in which a single block is meant to be produced. An epoch is a longer time window made up of many slots, after which certain parameter changes are applied. Compute units are a measure of how much compute a block may contain.
Why is the per-block compute limit being lowered at the same time?
This is the heart of the whole story. According to CryptoSlate, blocks get a shorter target production interval, but they do not get a larger allowed amount of compute per second. In practice this means that when blocks are produced more often, each individual block has to be "lighter" so that the total load on the network and validators per unit of time stays roughly constant.
The logic is simple: speeding up blocks alone without adjusting the compute limit would mean more compute per second, which could strain validators and threaten network stability. So lowering the per-block limit is a safeguard, not a side effect.
How far along are the test networks?
According to CryptoSlate, the rollout is further ahead elsewhere than on mainnet:
| Network | Block time target status |
|---|---|
| Mainnet | transition from 400ms to 350ms (effectively from epoch 1020) |
| Testnet | effective target 200ms |
| Devnet | 300ms target, 250ms gate activated but not yet effective |
So Testnet and Devnet serve as a proving ground where more aggressive values are tested before they reach mainnet.
Where does this change come from?
CryptoSlate points to Solana's changelog from August 6, where these parameters are described. Gradually shortening block times is part of Solana's broader long-term push for lower network latency.
What is not yet clear?
The available source does not give the exact calendar date when epoch 1020 will occur, nor the specific numerical value of how much the per-block compute limit is being lowered. This source also does not document what real impact the change will have on transaction latency, fees, or network stability for end users. That will only become apparent from mainnet behavior after epoch 1020.
What to watch for with this type of change?
With protocol-level blockchain parameter changes, it is useful to watch three things: whether the change actually activates in the announced epoch, whether any unplanned outages or slowdowns in block production occur, and whether throughput (transactions per second) actually changes or stays stable. Precisely because Solana is deliberately holding compute per second constant here, the main metrics to verify are latency and stability rather than raw throughput.
charliedesk is noting this change so it can later be compared against what actually happens on the network after epoch 1020.