Feature Gate: SIMD-0207 - Raise Block Limits to 50M CUs

[apfitzge]

SIMD

https://github.com/solana-foundation/solana-improvement-documents/blob/main/proposals/0207-raise-block-limits-to-50M.md

Description

• Initial step in raising chain capacity
• Minimum cluster version at or above minimum supporting version in Agave and (Fire|Franken)dancer

Feature ID

5oMCU3JPaFLr8Zr4ct7yFA7jdk6Mw1RmB8K4u9ZbS42z

Activation Method

Single Core Contributor

Deployment Considerations

Tests to verify new limits applied upon feature activation and propagated to child banks.

Minimum Beta Version

No response

Minimum Stable Version

No response

Testnet Activation Epoch

No response

Devnet Activation Epoch

No response

Mainnet-Beta Activation Epoch

No response

[Chazzzzzzz]

Love to see CU increase.

But curious to know whether this helps at all.

Any idea why all validators' CU can't reach 48M at all?

https://solanacompass.com/validators/EARNynHRWg6GfyJCmrrizcZxARB3HVzcaasvNa8kBS72

[apfitzge]

Love to see CU increase.

But curious to know whether this helps at all.

Any idea why all validators' CU can't reach 48M at all?

solanacompass.com/validators/EARNynHRWg6GfyJCmrrizcZxARB3HVzcaasvNa8kBS72

There are inefficiencies near block-limits because of our reserve-execute-commit scheme (see #3528 as an alternative, with some description of issue).
But i'm also skeptical that solanacompass even uses the right measurement - historically many of these external sites seem to be using ONLY executed compute units (i.e. bpf + builtin) but not other costs such as signature, write-lock, etc, and so they are strictly undercounting. This is not really their issue since that is what is stored in the blockstore and available outside metrics