What are the measurements for an OS to be described as the 'World's fastest OS'? MorphOS is very fast, so this could be a good starting point for promoting it, but we need to have certain values to prove that. QNX is also fast but how do we compare operating systems? It is clear that Windows is not the fastest, but what is fast?

Fast,in my opinion,is that when you USE the OS, you don't have to wait ages for something to happen.
Windows is the worst example of this.. especially when it comes to harddisk i/o .. sometimes it doesn't let you continue working until if has finished doing god knows what with the harddisk.
The use of virtual memory is also pretty fekked in Windows.. one could imagine with half a gig memory virtual memory doesn't need to be used much...

Thanks, we all know it IS fast. But how do we show/prove it?

Montoya is also fast, but Schumacher is faster. And he's got records to prove it.

Are there benchmarks for OS's? Such as application launching Benchmark etc?

You can prove it with some real scenarios and from there make event-time diagrams.
Use UML which is standard Modeling Language to make your diagrams.
Unified Modeling Language is known for development of OOP applications and any kind of processes.

http://www-306.ibm.com/software/rational/uml/

(I'm using it too for Software Development).

The best way to prove it, is to let them use it. This can be best achieved by visiting demoparties and other similar events.. take your Pegasos with you and show it to people.

Quote:
WRONG!

Schumacher's got faster car.

And same works on computers too.

Is WindowsXP faster than Windos98?

I guess many PC users would answer "yes" to that question, just becouse they had slower computer back then, when they were running Windows98...

Quote:

Sorry, but Schumacher did win atleast 3 out his 6 titles with an
inferior car, while Montoya has a tendency to throw away points by
trying to win races with brute force.

Quote:
Well, he's also won many times when:

-He wasn't given any penalty when any other driver would have been given one.

-He didn't do anything about the penalty he was given.

But you got my point... Faster hardware doesn't really make software faster

@liquidbit

Thanks for the answer. Anyway, I believe I couldn't explain correctly what I meant, I just wanted to know if there is some scientific approach to measure the speed of an OS.

@Jupp3

Did not wanna start a F1 discussion, but you are right, MS Word is being executed on my XP PC in the same time as in my 386SX from 1995.

WindowsXP? <<Shudders>> I hope you don't use that thing on the 'net!

"Are there benchmarks for OS's? Such as application launching Benchmark etc?"

Nope. IMHO, at least not any good ones.
OS benchmark is hard to implement. One really needs to experience the responsivenes, I think it's very hard to simulate in any single automatic benchmark.

About F1....

Without the McLaren hampering Kimi's speed, the old (Rally) saying would rule.
"You need a Finn if you want to win."

Then getting back on-topic: MOS has a few Finns in the crew, so no worries.

I think what we want to measure is "user efficiency", that is, the effectiveness of the interface and the response time. RTOSes should be Real-Time. Is MorphOS a RTOS?

Anyway, we should pick some typical applications, like the web browser or our favourite paint application. Of course, the applications differ from Windows to MorphOS. Just disable plugins in Windows to be fare, or something like that.

Then, take some measurements, like:

* time to open/close the program 100 times;
* maximum number of threads we can open without using VM;
* time to switch between applications (USER time may also be included to also evaluate the interface! can we do ALT+TAB and directly type? or click with the mouse is required?)
* time to boot
* time to close the system
* time required for the installation of the system
* space in disk required
[...]
* time to surpass other systems on the RC distributed.net stuff??? ;)

[...]
Anyway, I can not think of anything else... May this be useful?

As always, very late at night, and a lot of things to do... (-_-)
oyasumi!

If you are looking for something serious look at this site: http://www.spec.org/

I think the biggest problem with SPEC benchmarks is that they cost a lot to perform. Usually they are adapted for the system you run on in order to achieve comparable results.