jamesob

Kicked off a bitcoinperf run; will have some results tomorrow.

Cool! Seeing a **~8% speedup** over a modern region of the chain with **lower memory usage**. ---  | bench name | command | |--------------------------------------------|-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|...

> When flushes would happen with e.g. lower dbcache size or longer range of blocks memory usage should be about equal, but then there should be an even larger performance...

Back with results for dbcache=1000; less noticeable speedup (5%) and increased memory usage (11%). ---  | bench name | command | |--------------------------------------------|-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------| | ibd.local.range.dbcache=1000.667200.697200...

Is this project not maintained?

I'm very interested in this feature and will be investigating use of Silverbullet more thoroughly in the coming week or so. If I continue to use SB, I'd be happy...

Wow, thanks @zefhemel! I'll give this a shot today.

This works great! Thanks.

Thanks for the look @Sjors. I also realized that there were two outlier calls where we weren't asserting that Flush()/Sync() has succeeded, so I've fixed those.

@Sjors how did you detect that RAM *wasn't* being freed? Did you verify that the flush did not happen based on an absence of logs? Measures of memory can be...