Calculating the min requirements for dedicated brain
#1
Current project is on a single repurposed PC w/ 3 GTX 1080s running both client and server:
  • Itunes >10 Password (mode 14800)
  • Total hashrate 300 H/s
  • Crappy CPU
  • Ram 4GB (might be too low, should I set a --brain-client-features?)
  • SSD 100GB
  • Internet: 30 Mbps (potentially need to upgrade if I move the server over the network?)


I have my primary workstation on a different network which could add another 200 H/s but I figured I would need to relocate my current ldmp files and server to cloud instance like AWS but I'm having trouble figuring out min specs required for which aws tier I can go for without compromising performance.

Since the wiki mentioned 50 kH/s being the bottle neck I'm safely under that to make the brain worth it, but for my network it mentions "a single GPU can create around 5Mbit/s of upstream", would that mean I'm consistently using up 15Mbits/s ? I heard using --brain-client-features 2 may help but I'm still wrapping my head around the impacts of each setting.

Lastly I couldn't find how much RAM is recommended if I'm planning on having two clients pushing 500 H/s in total to a single server. Any calculations or more references to the wiki would be very helpful
Reply


Messages In This Thread
Calculating the min requirements for dedicated brain - by Raccoon - 03-31-2022, 04:54 AM