High Frequency Vs Bare Metal vs Dedicated Cloud for CPU Performance
I started a ticket but realized that this info could benefit others so I will post this here and update it with more info down the road as I receive it:
My objective is to find the solution that offers the best sustained CPU output with single thread being a priority and then multithread performance coming close behind that.
I'm looking for benchmarks on sustained cpu performance compared between these 3 packages (4 core) as well as read/write speeds and read/write latency
I'm aware of the surface level advantages of the dedicated cloud and bare metal, not being affected by other users and what not but I'd like to know some information on the underlying differences in specifications between machines in all of these packages. What are the clock speeds, single threaded and multithreaded benchhmarks of the dedicated cloud instances vs HF instances? Do the Dedicated cloud instances have NVME drives like HF instances? If not what kind of read/write benchmarks can they achieve?
Comments
It's been a bit difficult to get the answers out of support that I want so I did some digging and the best resource I can find so gar are from Geekbench.com
2018 Bare Metal Benchmarks: https://browser.geekbench.com/v4/cpu/9595167
2019 High Frequency Benchmarks: https://browser.geekbench.com/v4/cpu/14786234
Not much for the dedicated cloud
I believe my final findings were that bare metal, for sustained cpu usage, will be 15-25% faster than the HF instances. I haven't tested this yet but will in the coming weeks/months