How to Benchmark an Intel CPU
by andy70707 in Circuits > Computers
71555 Views, 4 Favorites, 0 Comments
How to Benchmark an Intel CPU
Have you ever wanted a good comparative method to test the actual speed of your CPU? You can measure its physical speed in GHz, but it won't give you much idea as to how fast it can actually solve problems. For example, you can clock a Pentium 4 at 4GHz, but an i7 at 1GHz will probably be still a lot faster. There are various other benchmarks, like CPU Marks , ok for comparing your CPU to others, but they are otherwise meaningless, and require payware to test your own configuration. If you want to benchmark you CPU, it is best to use GFLOPS (Giga [Billion] FLoating-point Operations per Seconds). It is what all supercomputers will use to test their speed and a very good measure of the processors true speed.
NOTE: This only works on Intel CPUs. You can probably find one for AMD if you google it, but as I don't have any, haven't really looked. Only thing I could find was a linux version you have to compile yourself.
NOTE: This only works on Intel CPUs. You can probably find one for AMD if you google it, but as I don't have any, haven't really looked. Only thing I could find was a linux version you have to compile yourself.
Get Linpack
Using Linpack
Once downloaded, open the "benchmarks" folder, then the "linpack" folder. Run linpack_xeon64 (or 32 if you have an older 32-bit cpu).
You should now have a command prompt/terminal window open. Press return once so it says "Number of equations to solve". Input a large number like 5000, and hit enter. Then, for the leading dimensions for array, enter a larger number, for example, 10000. Note: you may want to run less equations for older CPUs, otherwise they might take over a minute just to solve one! Next, enter about 200 trials, and then the amount of ram for it to use (in kb). You will want to change this according to how much ram you have free, I gave it 100000, which is just under 1GB. If done correctly, it should then start giving you benchmarks. I have found mine varies quite a bit, but I have managed to get up to 43GFLOPS on Ubuntu with it overclocked to 3.3GHz. Feel free to post your benchmarks in the comments!
You should now have a command prompt/terminal window open. Press return once so it says "Number of equations to solve". Input a large number like 5000, and hit enter. Then, for the leading dimensions for array, enter a larger number, for example, 10000. Note: you may want to run less equations for older CPUs, otherwise they might take over a minute just to solve one! Next, enter about 200 trials, and then the amount of ram for it to use (in kb). You will want to change this according to how much ram you have free, I gave it 100000, which is just under 1GB. If done correctly, it should then start giving you benchmarks. I have found mine varies quite a bit, but I have managed to get up to 43GFLOPS on Ubuntu with it overclocked to 3.3GHz. Feel free to post your benchmarks in the comments!