Page 1 of 1

User Account Control (UAC) and Performance

Posted: Thu Apr 14, 2011 9:31 pm
by DairyCreamer
Pardon me if I didn't search hard enough and this has already been discussed extensively.

Can someone explain what performance gains (or lack thereof) there are in disabling UAC?

Any difference between Vista and 7 in this regard? I know there is a slider setting in 7... but practically speaking anyway.

Thanks for the info. Sorry if I'm beating a dead horse.

Re: User Account Control (UAC) and Performance

Posted: Thu Apr 14, 2011 9:43 pm
by just brew it!
AFAIK there's no performance difference. You're trading off security against convenience.

Re: User Account Control (UAC) and Performance

Posted: Fri Apr 15, 2011 12:18 am
by Flying Fox
The only "performance difference" is that the first time the UAC prompt is raised for the application, you will notice a delay and the screen goes all dark on you with the dialog, which is intended behaviour to alert you that the application is asking for elevated access. Something that requires your attention. After you grant elevation access, the application should function the same. One time per application instance.

Re: User Account Control (UAC) and Performance

Posted: Fri Apr 15, 2011 1:11 am
by Meadows
Flying Fox wrote:
a delay and the screen goes all dark on you with the dialog, which is intended behaviour to alert you that the application is asking for elevated access. Something that requires your attention.

The delay is more than a delay. It's also intended to secure the input channels and prevent "malicious" permissions, for example via mouse cursor hijacking. Also, while the UAC prompt is in display, Aero is disabled and all windows and screen updates are suspended.

This makes it seem like the computer "jerks" back to work after you make the prompt go away with whichever answer, but UAC itself actually has little effect on performance.

There is, however, the problem with "checking" executables - the larger the .exe, the longer it will take for UAC to finally pop up. If you have a one-gig Windows service pack setup for example, it may take over 30 seconds between the double click and the actual running of the program, and this is directly a performance result of UAC. Without it, all executables are loaded instantly. (This is especially irritating over networks with a slow switch or otherwise modest connections, because even medium-sized setup files take next to forever, if you're installing them from the network. For that, you should copy them over to the end machine first.)

Re: User Account Control (UAC) and Performance

Posted: Fri Apr 15, 2011 5:59 am
by no51
I did the exact opposite and made UAC ask me for a password everytime. It allows me to be angrier at crappy programs requiring elevation for absolutely no reason, plus I can pretend I'm in a *nix environment.

Re: User Account Control (UAC) and Performance

Posted: Fri Apr 15, 2011 11:17 am
by Flying Fox
no51 wrote:
I did the exact opposite and made UAC ask me for a password everytime. It allows me to be angrier at crappy programs requiring elevation for absolutely no reason, plus I can pretend I'm in a *nix environment.

Instead of turning UAC up to 11, wouldn't it be easier to just run as a standard user?

Re: User Account Control (UAC) and Performance

Posted: Fri Apr 15, 2011 11:47 am
by SuperSpy
Meadows wrote:
There is, however, the problem with "checking" executables - the larger the .exe, the longer it will take for UAC to finally pop up. If you have a one-gig Windows service pack setup for example, it may take over 30 seconds between the double click and the actual running of the program, and this is directly a performance result of UAC. Without it, all executables are loaded instantly. (This is especially irritating over networks with a slow switch or otherwise modest connections, because even medium-sized setup files take next to forever, if you're installing them from the network. For that, you should copy them over to the end machine first.)


I'm pretty sure the signature checking happens regardless of having UAC on or not. Explorer on XP machines even has this behavior.

Re: User Account Control (UAC) and Performance

Posted: Fri Apr 15, 2011 12:02 pm
by Meadows
SuperSpy wrote:
I'm pretty sure the signature checking happens regardless of having UAC on or not. Explorer on XP machines even has this behavior.

Yes, but this isn't it. Launching executables is much slower under UAC, if the executable is enormous enough. Once they run however, there's no difference, so that's the answer to the OP.

Re: User Account Control (UAC) and Performance

Posted: Fri Apr 15, 2011 12:14 pm
by Sargent Duck
no51 wrote:
I did the exact opposite and made UAC ask me for a password everytime. It allows me to be angrier at crappy programs requiring elevation for absolutely no reason, plus I can pretend I'm in a *nix environment.


That works too. But what are you doing with crappy programs on your box?

Re: User Account Control (UAC) and Performance

Posted: Fri Apr 15, 2011 1:14 pm
by Ryu Connor
Flying Fox wrote:
no51 wrote:
I did the exact opposite and made UAC ask me for a password everytime. It allows me to be angrier at crappy programs requiring elevation for absolutely no reason, plus I can pretend I'm in a *nix environment.

Instead of turning UAC up to 11, wouldn't it be easier to just run as a standard user?


OTS (over-the-shoulder) elevation is the default behavior for standard users - I know I run as a standard user. Enabling OTS for the Admin accounts is hardly turning UAC up to 11.

There is an even higher secure desktop interaction setting available for UAC. Where upon arriving at the secure desktop you must send the SAS (sequre attention sequence) before you can enter in your Administative password.

http://www.windowsitpro.com/article/sec ... n-with-uac

Re: User Account Control (UAC) and Performance

Posted: Fri Apr 15, 2011 7:39 pm
by no51
Sargent Duck wrote:
no51 wrote:
I did the exact opposite and made UAC ask me for a password everytime. It allows me to be angrier at crappy programs requiring elevation for absolutely no reason, plus I can pretend I'm in a *nix environment.


That works too. But what are you doing with crappy programs on your box?


That's a good question. Honestly, it's only xfire that does it and I don't know why I keep it anymore but I've had it forever.