Sign in with
Sign up | Sign in
Your question

Computer Performance over the years

Last response: in CPUs
Share
November 21, 2008 10:18:25 PM

Hello to all you fine folks,

I'm currently working on a project which deals in part with the trends of computer speeds and performance over the past ~10 years. Now, as strange as this sounds, I've had a hard time coming across any solid comparisons of new and older technologies.

What I'm really looking for is a graph of average computer performance over the last 10 years. I'm not big into computer hardware, so I really don't know if there's any standard metric or benchmark, but I certainly haven't been able to find one.

Does anyone know of a site/database/resource that could help me by either providing such graphs or the data so that I could make them myself? Any information is helpful - I've come across alot of dead ends.

Thanks!
November 21, 2008 11:29:05 PM

there is an option in Si-Soft Sandra which will give a good comparison over older tech by benching ur own equipment and comparing it with thech of the past

ATM this is my best guess, but will inform u of anything else when it comes to me
November 22, 2008 12:51:00 AM

Well errr the trend is whatsisfacesname's law more or less. look up Moore's law and you won't be far off.
Course it's not really a law for performance but the results pretty much lead to the same graph in performance.
Related resources
November 22, 2008 10:23:33 AM

Try this

http://www.cpubenchmark.net/

Its a synthetic cpu benchmark but has a huge list of cpus, going back to piii. Look at the high, mid and low cpu mark charts. You will have to put it in some sort of order though, thats going to take you a while. Also, it doesnt include the i7 cpus
November 22, 2008 10:25:52 AM

^^ Edit: whoops, yes it does include i7
November 22, 2008 2:52:57 PM

1998 - 2008?

For the first half of that stretch (1998 - 2003), performance went up in the usual almost-exponential manner observed over the previous 20+ years. But for the second half and for the foreseeable future, performance increases have been relying on widening the execution much more than speeding it up. That is a very distinct trend, and the common explanation is that some of the process geometries are hitting the atomic scale, meaning new processes are not complete shrinks of previous generations and thus don't show across-the-board performance improvements.

Also, every benchmark out there either tests a very limited set of CPU capabilities, or tests a preprogrammed mixture of tasks that may not accurately reflect a CPU's improvements over the prior generation.
!