I am replacing my old Nocona-based server (2x 3.0GHz) with L5420 Harpertown (2x 2.5GHz) based machine. The problem is that I have a single-threated custom TCP server process that is maxing out Nocona 3.0GHz Xeon. "top" command shows 100% CPU usage for this process. I know that exact benchmarking depends on many factors, but can anybody predict whether this process will take less than 100% of single core utilization when it is run on a L5420 2.5GHz Xeon? Again, since this is a single threaded application, the server process will utilize only one core.
I need to know this asap, to decide whether I need to parallelize the TCP server application immediately, or I can wait a bit thanks to the CPU upgrade.
Well if the CPU us are currently using is a older one the Harpertown should increase performance quite a bit. As for if the program will use 100% of the CPU is hard to say since its up to the program. It may use all resources or only what it needs.
And if it does use all of it that means it will have more power to perform better.