I am getting a free PC from my uncle which is quite old..
Its an Intel Pentium G3900, 2 cores - 2 threads, 2.8GHz, 16GB DDR4 RAM
Planning to run Debian v12.10 headless server OS.
How many nodes can I run without maxing out my CPU?
Anyone running on similar setup?
Thank you!
Start with one or two and see how that goes. If ok, add another couple etc until you find a limit.
Be patient and you’ll get there quicker 
If you’d push it I think you could do around 500 to maybe even a 1000 nodes.
I hate tinkering PC so I thought I could install X number of HDDs at once instead of adding gradually.
I tried running nodes on Dual core CPU which is Intel Pentium G4400 (Also 2 cores - 2 threads) on Windows 10 and at 25 nodes it was using 30-40% CPU. How do you estimate 500 nodes?
Probably because windows is much more demanding than debian. I’ve had several cpu’s running 2 ghz (in containers) that were capable of running 500 nodes per container. The 1 thread per core I’m not sure of though.
I think you can run about 120-130 nodes on 50% cpu load if you use anm that controls cpu load.
As that gives you some headroom if cpu load would spike in the future.
I run about 1500 nodes on a 5700g and that cpu is about 11x stronger than the one you use, so 1500/11 = 136 on 50% cpu load.
Can you tell on average RAM usage per node & Max per node?
About 22 GB per 1000 nodes, 22 MB per node.
1000 x 0.110 = 110GB no?
Also 110MB per node is the max? coz I am planning to fit 150 nodes in 16GB RAM.
Correct, made an error, about 22MB per node, 2200MB with 100 nodes, 22GB with 1000.
Thanks, looks like my CPU is the bottleneck now for 250 nodes.
I run 60 nodes on old MacBook Air with Core i5. It has a 2 cores/4 threads CPU and has ~5 avg load.
Nice! I have around 10TB of free HDDs…
If you don’t mind me asking, How much ANT tokens did you earn for running 60 nodes that’s roughly 1TB.
Last time I checked it was 6.2 ANT (all nodes altogether) per April running 24/7. And you don’t particularly need all those terrabytes as network is empty now. Nodes don’t reserve space upfront. Only logs eat space really.
I highly doubt that. Modern 4 core (4 thread) intel CPUs max out at about 400-500 nodes due to CPU usage at times being too high. 2 cores on older instruction sets will not get that high a count. Maybe 200 as the limit leaving room for increased CPU load when node activity occurs
Hi, Is it recommended to keep 50% CPU free for spikes?
Can I fill more for average 75% and leave 25% room for spikes? Or is it a bad idea?
Yes, I would claim that it is better to leave good room example 50% for spikes, we don’t know what is a secure level as the network has not seen drastic shocks as during the test networks.
During the test network 25% was a ok cpu level. If you have a custom monitor tool that can adjust nodes every second then you can have a higher % load but if you don’t have that then it is not recommended. As you will not make any huge peofits from 150-200 nodes then it is better finding a stable level that supports the network.
In addition. At this time the node activity is low because there is not much stored, and with so many nodes in the network each node has little work to be do. Once there is more data stored on nodes there would be expected to be more activity. Such as storing data, retrieving data, giving quotes, responding to clients asking for network discovery of other nodes your node knows of.
At this time having 75% CPU is fine, but in future the spikes will be higher and more often or even more sustained.
I don’t have the time today, but in the other topics you talk of ports being opened in the router. They will have to be port forwarded to the PC using those particular ports. If you have 3 PC then there will be ports forwarded to one PC and other ports forwarded to another PC and so on.
You have to manually set those port forwarding rules in your router or else your nodes will not be contactable by clients or the emissions quoting machines.