Every now and then I get a bit nostalgic for the days when my workaday computer had a nearly instantaneous response time.
That would be when that computer was a 6 MHz IBM PC/AT.
How could it have an instantaneous response time when it was several hundred times slower that the machine I’m using to write this blog entry? Because the software was designed to be fast and efficient. If I wanted a directory of files in a folder, it appeared like lightning. (Even faster when I hot-rodded that AT to run at 8 MHz with a faster crystal on the motherboard. )
Today, Windows 7 is all over the headlines, and one of the reasons people like it more than Vista is that it’s faster and more responsive. However, even Windows 7 on a state-of-the-art PC is a slug compared to my old AT running DOS. Does that matter? It turns out that it does.
Back in November 1982, an IBM researcher named Arvind Thadhani wrote a paper titled “The Economic Value of Rapid Response Time.” This paper challenged the conventional wisdom that improving computer response time past a second or two had no measurable impact on productivity. Thadhani’s work indicated that a programmer’s productivity improved dramatically – and non-linearly – when the system response time fell below one second. Heck, I wait several seconds sometimes just for Windows to enumerate the contents of a network folder.
My belief is that Thadhani was right, and the price we have paid for the huge overhead that Windows imposes is greater than we realize. Windows 7 is faster than Vista and we can all be glad about that. It’s still not anywhere near fast enough to enable us to work at our most productive. In that sense, at least, it’s not nearly as good as DOS!
Glenn Weadock is a longtime instructor for Global Knowledge and teaches Windows 7, Server 2008, and Active Directory. He has recently co-developed with Mark Wilkins two advanced Server 2008 classes in the Microsoft Official Curriculum. Glenn also consults through his Colorado-based company Independent Software, Inc. and is technical director of MarketCoach Investment Education Software LLC.
The Leader in Network Knowledge
I still agree
I wish I had read this article before my response to your next post. You had already said the basics of my comment, except that I went even further back in time - BPC (before the Personal Computer). At that time my "PC" was an IBM 4381 (small mainframe), which I was setting up to be a multi-user development system. I was very impressed by the directly connected terminals over a 1 megabyte per second channel and 3380 disks at 3 megabytes per second. It ran at 2.2 MIPS (single processor), 8MB RAM, and 2.5 GB of disk if memory serves. It was designed to support about 20 simultaneous users doing programming in assembler language. Big stuff for its day. I had sub-second response, but I was concerned about the responsibeness with more than a dozen users. My how times change. Now I support a large UNIX system with 40 CPUs, 200GB ram and 4Gbps FC disk arrays.
Keep up the good work...
Post new comment