#3 Cracking and fetching new chunks in parallel

[Lypsis]

Hope I don't misunderstand something here, but as I get it there is no parallel fetching of new chunks while cracking one.

This would increase performance for nodes on slow network connections. I for one have one node which has a huge CUDA GPU, but is on relatively slow internet connection. It is not possible to use this one as the moscrack server (as you suggest in the iFAQ).
The bottleneck for my node in this scenario is definitely the fetching of the chunks (which can be decreased in size of course, but as a consequence it heavily increases the overhead because it's cracking the little chunks in no time).

If you need any more information, please let me know.

Thanks and keep up your great work!

[Ryan Babchishin]

I just realized this ticket was here. Looking into it.

[Ryan Babchishin]

This is already done in parallel. The main process forks as it picks up chunks. Everything else is done in a separate process. It sounds like what you want is prefetching chunks. That's something I may consider doing if it's not too complicated. In the mean time, try turning on SCP compression and increasing the chunk size for that node.