DOs and DONTs

[Narayanan]

Hello all,

I plan to use the "performance test v6" to analyse the computers in our organisation. The computers are very hetrogenous regarding their hardware and software.

All these machines will have variety of applications running (either as application or as service in background).

I would like to benchmark all these computers. But I fear very much about the reliablity of the results especially when run on these 'used' systems. I know, the theoretically correct way to run the benchmark is to format the systems with identical OS settings and run benchmark. But this is not a feasible solution.

I just want to know whether my fear is valid -or- am i just getting over cautious?

I feel that BIOS settings (like AGP memory allocation, PIO/DMA modes for drives), OS settings (like swap file size, drivers, PIO mode/DMA mode, cache), applications (like virus scan running in background, services) will affect the result of the final score. Am i correct? if so how can i provide a true baseline to compare my system with others?

Any inputs will help me to get out of this problem...

Thanks in advance,
Regards,
Narayanan

[David (PassMark)]

Depends what you are trying to measure. Maybe you want these factors to affect the results. It might indicate which PCs are most in need of some attention. e.g. a defrag of the disk, removal of unnecessary background tasks or loading of the correct video card drivers.

Some BIOS settings can impact performance, but the default settings for any particular board are normally fairly good.

O/S settings can also have an impact. But there are too many to list.

I think I would be tempted to have a quick look for really obvious problems (e.g. a background task using 50% of the CPU), then benchmark the PCs. Then later on, have a closer look at the machines that seemed particularly bad in one area of another.

------
David