Thursday, July 10, 2008

Measuring responsiveness 1

I'm interested in trying to measure the responsiveness of computer systems. This was because I'm amazed at the apparent quickness of my old DELL Dimension 4550 (2.53GHz Pentium 4, 1Gb DDR memory with Windows XP SP2)which I use for email, web browsing, Word, Excel, Sage bookkeeping, web designing and other administrative tasks. Yes, I know it is over five years old and needs replacing but it still works well and seems reallyresponsive often with many windows open and applications running.

I install quite a few new DELL systems with various configurations and you have to beef them up a bit with the faster Core 2 Duo processors and at least 2Gb of memory before they appear to be just as responsive as my old DELL. So I started to research how to measure responsiveness and discovered very little in the way of research or benchmarks. The closest is Futuremark's PCmark which at least simulates normal use of a computer and times the various operations. However, it still seems very CPU oriented.

Thirteen years ago I had an Intergraph TD-3 workstation with two Intel Pentium 90MHz processors and a SCSI disk drive running Windows NT. That was a very responsive system. You could even continue to work on it while it was printing or formatting a floppy disk. Today multi-core processors seem essentially linked to fast response. This means that when you double click a desktop icon the first time a window opens immediately with the respective application's menu structure and your document or data displayed within it, ready for your attention. A slow response gives three blank windows (because you ended up hitting the icon three times), the hard disk light comes on continually and the system grinds away trying to fill in the boxes.

No comments: