CPU Management

[TheJam]

I have an XR1000 which has 3 CPU's, however, one is almost always idle with the other 2 frequently going up towards 100% which ends up bogging down the network and I have to restart the router fairly regularly. It would be great to have the option to use 3 CPU's for all traffic rather than having one dedicated to WiFi which really doesn't get utilised, the design decision doesn't really make sense to me. I'm guessing it's to mitigate complexity with managing 3 CPU threads with each other at a low level but it would really solve a lot of high load issues on the router if wired traffic had an extra 33% overhead (assuming all CPU's are the same).

Could that possibly be added as an option to override the default CPU behaviour and use all 3 for all traffic? 

[Netduma Liam]

42 minutes ago, TheJam said:

I have an XR1000 which has 3 CPU's, however, one is almost always idle with the other 2 frequently going up towards 100% which ends up bogging down the network and I have to restart the router fairly regularly. It would be great to have the option to use 3 CPU's for all traffic rather than having one dedicated to WiFi which really doesn't get utilised, the design decision doesn't really make sense to me. I'm guessing it's to mitigate complexity with managing 3 CPU threads with each other at a low level but it would really solve a lot of high load issues on the router if wired traffic had an extra 33% overhead (assuming all CPU's are the same).

Could that possibly be added as an option to override the default CPU behaviour and use all 3 for all traffic? 

I like the idea, it's unfortunately not something we at Netduma have any input on though. As it's NETGEAR's hardware, they or possibly even their manufacturer of the hardware take the decision as to what the CPU cores are utilised for. Are you able to reliably generate a lot of traffic, on demand, and replicate the issue? If so, can you let us know the exact steps your follow? We may be able to replicate it internally and optimise it for a future update.

[TheJam]

Thanks for the response, didn’t realise you guys couldn’t control the behaviour with hardware directly.   

It’s something I’ve only looked at when there’s a network slow down, I haven’t monitored it enough to see if it gradually increases or spike to 100% and gets stuck for some reason. I’ll go and do some investigation on it - if I can get some useful information for you guys to work from - I’ll share it in this thread.