These are BENCHMARKS - which means we are not testing something random.
Worst cases for each compute scenario - so pushing clients limits.
Average block post-Glamsterdam with the 200MGas will be likely still processed in ~20ms = 10 GGas/s.
A lot of room left for next bumps.
Show more
The work behind @Nethermind Client is kind of wild.
1.39.3 HalfPath→master FlatDB eth_call performance:
• suite 1: 155.0 → 34.4 ms — 4.50×
• suite 2: 69.4 → 16.1 ms — 4.31×
• suite 3: 648.5 → 192.0 ms — 3.38×
Part of that lands in 2.0.0, rest will be shipped in 2.1.0.
Show more
