Interesting read, i thought i would post a new thread as the other one has degraded into being completely ridiculous (as Brandon might put it)

http://blogs.msdn.com/e7/archive/2008/08/2...erformance.aspx

I fixed their diagram.

Nice graph

Interesting, but It seems clear that depending on your hardware setup and how you've tweaked or added startup services/apps to Windows it will make your boot times long.

There is only so much tweaking and performance boosting MS can do to windows. I know lots of systems that would take long to start up because they had loads of apps being loaded and so on. The systray would be full of things when it was all over. God only knows what else was started in the background.

I'm glad they are working on boot time, sadly the OEM's will probably negate the improvements with all the start up crap they install. My vista install already only takes like 30 seconds to boot, but if you get a pc with an OEM vista install it takes FOREVER.

+1
Boot time of Windows itself isn't that problematic, but an OEMs seriously cripple the machines they sell. The Windows team must work close with them and try to solve this issue.

Probably the most interesting paragraph in my mind:

Interesting. But if a service can reduce its footprint to something negligable until it's doing something, is it really worth it? Wouldn't there be more of a performance penalty in the stuff involved in starting the service than having the service just sitting there not doing anything until its needed?

I'd rather waste 5-10 megs of memory and no CPU time with it idling than have a spike in CPU usage when it has to start.

I have 4 gigs, and discounting cached memory, nothing ever comes anywhere near using it all. Processor usage, however, can spike to 100% easily when I'm doing a lot of stuff.

The way I see it: Memory is cheap, CPU time isn't.

Even an idle services is going to use some cycles to process messages and to do whatever it is its supposed to be doing (like watching for new hardware, or a particular kind of hardware, or whatever). Mainly efforts like this involve finding cases where multiple services are doing the same thing... ie. both waiting for new hardware to be inserted. Then factor that code into one place and have it spawn the right service in response to that trigger depending on the situation.

Also keep in mind that this article is about boot performance. If you can reduce the amount of code that needs to be loaded and the number of processes that need to be started, you can shave time off of start-up (not to mention shutdown and other tasks), in addition to reducing the OS's footprint. There are other benefits too, like battery life for example.

Yep i've read this article and i've been suggesting this 4 years to the Windows Team up today, windows should have a type of self service management,if a service isn't used,Windows disable it automatically with no user intervention very convenient eh?.The whole idea of this is that Windows must have the capability to self manage "intelligently" in benefit of the End user.

This is just a part of the idea i've been suggesting to Microsoft and i think other people of course,however i haven't seen anyone suggest something similar of what i've been saying.The future of Windows could rely on something that i call "Computarized AI".

Finally my voice was heard

I'm not exactly sure how energy efficient it really is, but i use "standby" as oppose to shutting down my computer. The only time i'm booting up is if an update or some software was installed and it is required. Other than that it only takes a few seconds (<10) to wake up from standby and go into standby.