Jump to content

How to speed up boot process under Windows Vista or Windows 7


Recommended Posts


The first boot captures the current state. If your Windows took 55s it shows it.

So your "feeling" and you unreliable way to measure the boot time may let you think that it is not faster.

Well, the stopwatch certainly doesn't measure my feeling - I don't see how any other method can be more reliable (for any practical use, not theoretical) since the stopwatch shows the practical boot time, it shows what I experience as a user. But maybe xbootmgr measures the time in some other way that is unknown to me. Can you tell me what exactly is measured as bootDoneViaExplorer? When does the measurement start and when exactly does it end?

Link to comment
Share on other sites

@magicandre

Need your help on this. This is a brand new sandy bridge laptop with a greater than 2 minute boot time.

The trace:

https://www.dropbox.com/s/n2uak80f78orbuo/boot_base%2Bcswitch%2Bdrivers%2Bpower_1.7z

And the log:

[2012/05/21-19:28:47.284] Detected Win7RTM physical (ReadyBoot) prefetcher; the SysMain service is auto-started.

[2012/05/21-19:28:47.300] Registering for auto-run with a 120000 msec delay...

[2012/05/21-19:28:47.300] Completed auto-run change.

[2012/05/21-19:28:47.331] Starting trace...

[2012/05/21-19:28:48.173] Enabled boot logging to 'C:\Windows\system32\boot_base+cswitch+drivers+power_1_km_premerge.etl'

[2012/05/21-19:31:38.349] Detected Win7RTM physical (ReadyBoot) prefetcher; the SysMain service is auto-started.

[2012/05/21-19:31:38.365] Stopping trace...

[2012/05/21-19:31:41.500] Stopped kernel logger.

[2012/05/21-19:31:41.719] Stopped user-mode logger.

[2012/05/21-19:31:41.719] The trace you are capturing "C:\Windows\system32\boot_base+cswitch+drivers+power_1.etl" may contain personally identifiable information, including but not necessarily limited to paths to files accessed, paths to registry accessed and process names. Exact information depends on the events that were logged. Please be aware of this when sharing out this trace with other people.

[2012/05/21-19:32:28.176] Merged ETL file to 'C:\Windows\system32\boot_base+cswitch+drivers+power_1.etl'

[2012/05/21-19:32:28.238] Waiting for prefetcher...

[2012/05/21-19:37:18.695] Gave up waiting for Win7RTM physical prefetcher after 300 seconds.

[2012/05/21-19:37:18.695] Could not wait for prefetcher.

[2012/05/21-19:37:18.695] Stopping trace...

[2012/05/21-19:37:18.789] Couldn't find kernel logger in active logger list.

[2012/05/21-19:37:18.789] Couldn't find user-mode logger in active logger list.

[2012/05/21-19:37:18.789] Unregistering auto-run...

[2012/05/21-19:37:18.789] Completed auto-run change.

Regards

Link to comment
Share on other sites

Can you tell me what exactly is measured as bootDoneViaExplorer? When does the measurement start and when exactly does it end?

The bootDoneViaExplorer key shows the duration of the boot (in milliseconds) until the start of Explorer.exe (desktop appears)

@magicandre

Need your help on this. This is a brand new sandy bridge laptop with a greater than 2 minute boot time.

The trace:

Your Windows boots in 59s to the desktop and is completely booted in 91 seconds.

There is a 10s delay caused by 2 drivers:


<pnpObject name="AFD" type="Driver" activity="Load" startTime="8579" endTime="15062" duration="6483" prePendTime="6483" />
<pnpObject name="SRTSP" type="Driver" activity="Load" startTime="4620" endTime="8379" duration="3759" prePendTime="3759" />

SRTSP semas to be Symantec and AFD = Windows Winsock driver (seems to be related to Symatec). So update Symantec and look if this fixes it.

Also the bluetooth service takes 10s to start:


serviceTransition name="Bluetooth Device Monitor" group="" transition="start" totalTransitionTimeDelta="10740"

update or disable bluetooth. Also Symantec service is slow to start. Norton Online Backup v2.0 also takes 3.6s to start. Next teamviewer service takes over 6s to start.

Also the launch of all startup programs takes over 30s:


postBootDisturbance="32700"

Do you really need all of this Sony applications? Try to disable some with autoruns or msconfig.exe

Link to comment
Share on other sites

The bootDoneViaExplorer key shows the duration of the boot (in milliseconds) until the start of Explorer.exe (desktop appears)

Then it looks like the counting starts the moment "Starting Windows" text appears with the animated logo and ends when the "Welcome" screen begins to fade out, which is about 1 second before the desktop becomes visible - then the measurement is consistent with the stopwatch (20s). But what xbootmgr reported as boot time before optimization - 55s - is not true at all. My system never took that long to boot, I know for certain because I measured boot time many times before when I wanted to see how installing/uninstalling certain programs impacted the boot time, and of course before the optimization I did a few more measurements so that I could see any improvement. It is therefore impossible that my measurements were off - certainly not by such a huge margin. Either xbootmgr must have measured the time in a wrong way or maybe there was a rare coinsidence of windows doing some extra unusual tasks, updates, optimizations, etc. - just when xbootmgr did its first reading.

Edited by LemonJuice
Link to comment
Share on other sites

I'm not an expert in this area such as Andre is, but it might be along the lines of what you implied at the end of your post. MS changed some things around compared to earlier versions of Windows to improve the user's "experience". They made the desktop appear quicker, even though it was still busy starting services, background programs, etc. So maybe the counting doesn't end until the desktop is usable rather than just visible? That could explain why your stopwatch, based on what was visible, was off compared to what was reported. I would think that the optimization was meant to improve the measurement to usability, but with MS you never can be sure.

Cheers and Regards

Link to comment
Share on other sites

I'm not an expert in this area such as Andre is, but it might be along the lines of what you implied at the end of your post. MS changed some things around compared to earlier versions of Windows to improve the user's "experience".

I was implying something different at the end of my last post - I meant that sometimes windows takes longer to boot than usual because it has some extra tasks to perform like install an update or run some maintenance task. Maybe that first boot that was measured by xbootmgr was longer because of that, I didn't measure it with my stopwatch so I can't tell. But 55s was certainly not the usual boot time I experienced every day.

They made the desktop appear quicker, even though it was still busy starting services, background programs, etc.

From my observation this was somewhat improved after win XP. I remember this was more of a problem in XP where I had to wait a minute or longer after seeing the desktop until it actually becomes responsive. In win 7 more services seem to start before the desktop appears so the waiting for responsiveness isn't that long - I'm not an expert on this but these are my observations.

So maybe the counting doesn't end until the desktop is usable rather than just visible? That could explain why your stopwatch, based on what was visible, was off compared to what was reported.

I also thought it could be like this but my observations seem to suggest Andre was right that the measurement stops when the desktop appears - not when all services and startup programs are loaded. The final xbootmgr report of 20s matches my stopwatch when I stop measuring just when the desktop appears, and certainly it is before all startup programs are loaded.

I would think that the optimization was meant to improve the measurement to usability, but with MS you never can be sure.

I don't think it would be easy to accurately measure when the moment of 'usability' happens since different services/programs start at different points in time (some are delayed, etc.), some are heavy on CPU usage while others have almost no impact running in the background so I guess it's much better to measure until desktop appearance since this is some definite point objectively.

But as I said earlier after the desktop appears it takes about 5s for my system to be responsive - I can't tell if it's 100% responsive but I can start Explorer and other programs right away without any delays and xbootmgr optimization didn't change that. I could believe if there was some 5-10s. improvement in starting up background services that don't affect performance much but the results suggest that the boot time went from 55 to 20s! This is a huge difference and I'm sure it wouldn't go unnoticed. Therefore I assume that either xbootmgr measured time incorrectly or that first boot took unusually long for whatever reasons. I believe xbootmgr might have optimized something but it hasn't affected my system much - certainly not more than 1 or 2s.

Link to comment
Share on other sites

  • 1 month later...

I tried the process you suggest in this guide.

It took me more than an hour to complete and now my restart time is around 320 secs.

I really have no idea on what can affect so much my reboot time.

I just gave a look to the trace produced during the 6th reboot of the process and I see that most of the time my 2 cpu are idle (93% of the time) and most of this time (81%) I see in windows performance Analizer Summary Table is connected to Module=<Heuristic> Function=<Low Power State>.

Does this have a meanng for you?

Is there something I can show you to have some help?

Thanks

Michele

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...