Andrew Fitzgerald

#### Problem I was curious about the main differences between clap 2 and 3. #### Summary of Changes Converted store-tool to clap 3 as an exercise Fixes #

#### Problem Locking is bad for performance. If the root bank hasn't changed we do not need to lock again. #### Summary of Changes * root_slot on BankForks is now...

#### Problem lamporots #### Summary of Changes lamports Fixes #

Feature Impl: cost model uses number of requested write locks

2

#### Problem Implementation of #34819 #### Summary of Changes - Some write-locks are demoted to read-locks during transaction sanitization - Change the cost model to apply write lock costs for...

### SIMD N/A ### Description - Some write-locks are demoted to read-locks during transaction sanitization - Change the cost model to apply write lock costs for the number of requested...

BankingStage: set reasonable bound on max received packets per call

1

#### Problem If the buffer is empty when many packets are in the sigverify channel, a single receive call may receive too many packets and lead to that thread missing...

#### Problem #### Summary of Changes Fixes #

Add program cache timing to replay

1

#### Problem No replay timing on program-cache updating #### Summary of Changes Add a replay metric for time spent upating the program cache Fixes #

Refactor: SchedulerController tests call receive then schedule

1

#### Problem - scheduler controller tests are spawning separate thread for controller. these threads sleep while waiting for packets, and thus can cause flakiness in these tests if the OS...

#### Problem - functions are unused - tests rely on actual value of `priority` - value of `priority` changing with #34888, which breaks tests of these fns #### Summary of...