First off, this list is an experiment.
I currently DO NOT recommend you use it to decide on what parts to build your computer with.
This is an experiment to see if I can map a mathematical equation to the performance of processors. I've updated the equation to take into account the different memory speeds during benchmarking. The data I used to make the equation is from the benchmark 'WinRAR - Version 3.6 BETA 4' on the CPU comparison charts on Tom's Hardware.
This is the current equation that is used to calculate the estimated time for the benchmark:
Time = 441.4 - L2*0.2567 + L2^2*0.0001555 - Cores*77.81 - Freq*0.04785 - Freq^2*0.000001469 + (800MHz)*0.478 - (800MHz)*0.0004355
To calculate the performance index for the processors, I assumed the processors be used with 800MHz RAM. (I know this isn't physically possible on some processors, but it is to find relative performance.)
Please discuss the future possibilities of this type of data extrapolation and any ways you think I could make it better.
Note: The prices and processor information is from Newegg.com.
I think WinRAR is a weak basis for CPU-bound benchmarking. The application is not affected by just the CPU but by the different cache levels, bus bandwidth, RAM, and overall data latency.
Holding the RAM and MB steady would incorrectly stabilize large variations in CPU performance.
THG has already worked on this matter, though, by calculating an index of several CPU-based benchmarks and then offering a price-performance ratio for various processors, not necessarily in the same family.
Incidentally, your equation suggests that by waiting for AMD to accelerate enough L1 cache to a sufficient but finite clock speed, the time to pack/unpack RAR files would reach zero.
Thanks for the information about WinRAR being not only CPU dependent. I'll look into more CPU only benchmarks to see if I can get a better regression equation.
I've incorporated the RAM speed into the regression equation.
Once I made the equation, I assumed the RAM speed to be 800MHz to calculate the speed to complete the benchmark.
This has resulted in pretty accurate numbers. I'm sure the numbers would be much more accurate on a CPU only test, but this is showing a proof-of-concept that most of the variables can be accounted for.
You are about to answer a thread that has been inactive for more than 6 months. If you still wish to proceed, please ensure that your posting is original and does not duplicate or overlap any prior responses to this thread.