Hello! I want to test three computers in order to know which one has a better performance with floats operations. I found complex benchmarks that gives the FLOPS of a machine. But I don't need precision, I need a fast aproximation of the performance. Anyone know a simple benchmark for this problem? Or what kind of instructions should I manage to create my own simple benchmark?
Thank You
(excuse me for my english)
Any "simple" benchmark can be manipulated into producing whatever answer you want. You may as well just roll a dice and make your choice based on that.
One very crude "measurement" is to simply look at the clock speed of the processor. It might "win" on inline code, but if it has a deep pipeline and your code is heavy on the branches, then a slower processor with fewer pipeline stages will win (this really happened at a previous workplace, the slower laptops beat the desktops when running one particular program).
Your choice of OS, how many drivers and services are loaded, what other programs are running can affect the stats.
Your choice of file system, the performance of the physical disk, the state of fragmentation can affect the stats.
Running the program you actually want to run, with a representative data set is the only way to come up with a good answer.
If you dance barefoot on the broken glass of undefined behaviour, you've got to expect the occasional cut.
If at first you don't succeed, try writing your phone number on the exam paper.
I hate benchmarks. If you want to see how different machines compare when doing some task X, then try task X on each of the machines.Originally Posted by bartleby84
Using some benchmark Y to try to predict how fast task X will run is pointless.
