 |
July 19, 2015, 07:58:03 PM |
|
while reviewing code, and testing some things, I have identified that the current proof of stake is single threaded, and thus, when staking is enabled it only uses a single core, or single processor, single core. While running in a VM environment I was able to monitor this. I have the system assigned multiple vcpus and had one processor with 4 cores, or 4 single processors, or 2 processors 2 cores, and it always utilized a single core. Tested XenServer, and HyperV. Further research indicates that making this operate in parallel is the only option.
Lightstake code from PressTab on HyperStake only uses 1-2% of the cpu occasionally and does not hog the processor so that is an option. Their coin is different from ours. I was able to match some to NovaCoin, and NovaCoin does have a new stake engine, which could be utilized.
Does anyone who stakes other coins notice that when staking it will utilize multiple cores? if so please let me know, i would like to review options here.
I am virtualizing all my systems currently, and the processor in my server is slower than my main system, but has a lot more cores, so I am seeing staking performance issues on the wallet. If I use my main system, staking is quick and efficient, if I move it to an atom proc, or a celeron dual proc, or a old 2x quad xeon, the performance of the staking engine is obvious. So my solution is to get faster procs so the cores are faster, or resolve this issue. I plan to resolve.
much regard,
--o
|
