Would an SSD make for faster start-ups?

Nildo lost track of its imaginary kill file friend...



Path: aioe.org!news.mixmin.net!news2.arglkargh.de!feeder .erje.net!eu.feeder.erje.net!fu-berlin.de!uni-berlin.de!individual.net!not-for-mail
From: Nil
Newsgroups: alt.windows7.general
Subject: Would an SSD make for faster start-ups?
Date: Tue, 10 Jun 2014 23:52:16 -0400
Organization: (?!)
Lines: 11
Message-ID:
References:
X-Trace: individual.net C9fRpa+33YVe165drFPuvgWdNwOmJz66//uFTsH/qBWdrL2JdK
Cancel-Lock: sha1:FDpBC9bjX7Iv4EAmJl9Y4PGIS4M=
User-Agent: Xnews/2006.08.24
X-Face: esm\a~e7BW-JD"t0\Ww_~\t!z_p0}xokJ"]a4/!ZtMGxQt_J`\IuTO++qOqVx0&Y.=z(B!:d?HNxL}yTuIS^5T8 W\iGv_s'oSFfLp%X|naUNr
Xref: aioe.org alt.windows7.general:87734




On 11/06/2014 04:52, Nil wrote:
On 10 Jun 2014, BobbyM wrote in
alt.windows7.general:

It does if many of them are starting up when Windows does. If
only 10% of them are starting, that's 30 programs that have to
load before startup is complete.
Well yes, of course. But the issue is having too many of them start
automatically with the system, not their mere presence on the computer,
which is what the OP seems to believe. I often run across this
misconception.

In message , Nil
writes:
[]
Well yes, of course. But the issue is having too many of them start
automatically with the system, not their mere presence on the computer,
which is what the OP seems to believe. I often run across this
misconception.

And it can be quite deep-rooted: I've had cases where I've carefully
explained that the mere _presence_ on the disc of prog.s doesn't affect
startup, but I can clearly tell that the person is not convinced.
--
J. P. Gilliver. UMRA: 1960/1985 MB++G()AL-IS-Ch++(p)Ar@T+H+Sh0!:`)DNAf

live your dash. ... On your tombstone, there's the date you're born and the
date you die - and in between there's a dash. - a friend quoted by Dustin
Hoffman in Radio Times, 5-11 January 2013

In article ,
says...

Win7 SP1 64bit

My 1TB HDD has so many programs on it that the
computer takes over 5 minutes to start up.

The used space on this drive is about 300GB and is
almost all programs and operating system. I store
data on other drives.

Would a SSD give significantly faster start-ups?

Peter

What are you using as your benchmark to the point you consider the
computer 'started up'? To the login screen? Quickly logging in and then
waiting for the desktop? Waiting for all disk activity to end?

Also, have you defragged the drive recently?

--
If there is a no_junk in my address, please REMOVE it before replying!
All junk mail senders will be prosecuted to the fullest extent of the
law!!
http://home.comcast.net/~andyross

Nym-shifting troll...

--
John Doe wrote:

Path: eternal-september.org!news.eternal-september.org!feeder.eternal-september.org!aioe.org!.POSTED!not-for-mail
From: John Doe
Newsgroups: alt.windows7.general
Subject: Would an SSD make for faster start-ups?
Date: Wed, 11 Jun 2014 21:36:12 +0100
Organization: A noiseless patient Spider
Lines: 34
Message-ID:
References:
Reply-To: lid
NNTP-Posting-Host: bdqO927Xn0CCME09+eoAuQ.user.speranza.aioe.org
Mime-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
X-Complaints-To:
User-Agent: Mozilla/5.0 (Windows NT 6.3; WOW64; rv:24.0) Gecko/20100101 Thunderbird/24.3.0
X-Notice: Filtered by postfilter v. 0.8.2
Xref: news.eternal-september.org alt.windows7.general:101813

Nildo lost track of its imaginary kill file friend...



Path: aioe.org!news.mixmin.net!news2.arglkargh.de!feeder .erje.net!eu.feeder.erje.net!fu-berlin.de!uni-berlin.de!individual.net!not-for-mail
From: Nil
Newsgroups: alt.windows7.general
Subject: Would an SSD make for faster start-ups?
Date: Tue, 10 Jun 2014 23:52:16 -0400
Organization: (?!)
Lines: 11
Message-ID:
References:
X-Trace: individual.net C9fRpa+33YVe165drFPuvgWdNwOmJz66//uFTsH/qBWdrL2JdK
Cancel-Lock: sha1:FDpBC9bjX7Iv4EAmJl9Y4PGIS4M=
User-Agent: Xnews/2006.08.24
X-Face: esm\a~e7BW-JD"t0\Ww_~\t!z_p0}xokJ"]a4/!ZtMGxQt_J`\IuTO++qOqVx0&Y.=z(B!:d?HNxL}yTuIS^5T8 W\iGv_s'oSFfLp%X|naUNr
Xref: aioe.org alt.windows7.general:87734




On 11/06/2014 04:52, Nil wrote:
On 10 Jun 2014, BobbyM wrote in
alt.windows7.general:

It does if many of them are starting up when Windows does. If
only 10% of them are starting, that's 30 programs that have to
load before startup is complete.
Well yes, of course. But the issue is having too many of them start
automatically with the system, not their mere presence on the computer,
which is what the OP seems to believe. I often run across this
misconception.



program

Peter Jason wrote:
On Tue, 10 Jun 2014 20:00:42 -0400, Paul
wrote:

Peter Jason wrote:
Win7 SP1 64bit

My 1TB HDD has so many programs on it that the
computer takes over 5 minutes to start up.

The used space on this drive is about 300GB and is
almost all programs and operating system. I store
data on other drives.

Would a SSD give significantly faster start-ups?

Peter
I would want to understand why it was slow first.

WinXP had "BootVis", as a utility for displaying
the timing of the boot sequence. It made charts
and graphs for you. Microsoft discontinued support
for it, so it is no longer distributed.

This is the closest thing I can find to that.

"Windows Performance Tools"

https://web.archive.org/web/20080212...perftools.mspx

"xbootmgr boot trace capture tool"

The hardest part of using that, is going to be finding
a good version to download. The one for Windows 8
is screwed up by apparently being part of Win8 SDK.
So it'll be the usual raw deal, of dealing with the
crappy Microsoft delivery mechanisms for downloads.
We can't put them all in one place, on one web page,
like this.

Vista (Windows Performance Tools) Download_here
Win7 (Windows Performance Tools) Download_here
Win8 (Windows Performance Tools) Download_here

But I suppose that is too much to last. Much easier
to just delete all references to the older versions,
then make half-assed references to having put the
Tools into some SDK, then lie about which SDK they're
inside, and so on... The "Microsoft Way".

Paul

I dowloaded this and it presents in the Program
Folder as:
"C:\Program Files\Microsoft Windows Performance
Toolkit\xperfview.exe"

But nothing happens; the window wants a file
loaded.
There's a ReadMe file thus:
***

1.b. I cannot decode symbols. Why is that and
how can I fix it?

The xperf/xperfview trace symbol decoding support
is quite complex.
Please ensure the following requirements are met:

i. You have specified -symbols on the xperf
command line, or
selected "Load Symbols" in xperfview *before*
you opened a summary
table.

ii. You configured the environment variables as
described in the online
help (xperf -help symbols)

iii. The ETW kernel trace file has been stopped
and merged with xperf's
option -d or merged on the same machine it
was taken with xperf's
option -merge. (xperf performs a special
image identification
process during its custom trace merge.)

iv. The ETW user trace file is processed by xperf
in conjunction
with a kernel trace file taken in the same
time on the same machine
and merged as explained at point iv. above.

v. You have access to the binary and symbol
sources specified on
_NT_SYMBOL_PATH. If you use symbol server,
please note that many
times the symbol server is just a redirector,
and you need to have
access to both the symbol server itself and
the site actually hosting
the binaries and/or symbols.

vi. _NT_SYMBOL_PATH points to the right files. If
the files are from a
different build or architecture they will not
work. If you replace
the binary or symbols you will not be able to
decode anymore symbols
for traces recording activity of the old
binaries.

To rule out a symbol mismatch, use symchk
from the Windows Debugging
Tools distribution to ensure the symbols
match the binaries on the
machine on which the trace was taken:

symchk /v local_file /s
sympath_to_name.pdb

To rule out a binary mismatch, use fc /b to
ensure the binaries on
the machine on which the trace was taken
match the ones on the drop
sha

fc /b local_file drop_share_file

vii. The ETW kernel trace was captured with at
least PROC_THREAD+LOADER.
These flags provide basic information about
process lifetime and
image ranges in process memory, which are
instrumental in decoding
virtual addresses to images and symbols.

To verify that these flags have been enabled
in the ETW kernel trace,
check that Process events (Create, Delete,
Start Rundown, End Rundown)
and Image events (Load, Unload, Start
Rundown, End Rundown) are present
in the table generated by "xperf -i
kernel.etl -a tracestats -detail".


1.c. I'd like to have ETW log the context stacks
for certain events. How
can I do that?

Please refer to "xperf -help stackwalk".
***

Confused.

OK, I got it working.

The xbootmgr.exe is used to start a trace. When you
run that command, the machine is going to reboot immediately
afterwards, so any open files should be saved, programs
put away and so on. Then run the xbootmgr command.

After the reboot, and the xbootmgr dialog appears to have
finished, there will be captured .etl files in your working
directory. (That's the directory you set with CD in command
prompt, just before running xbootmgr.exe.)

The xperfview.exe, draws the graph. The output isn't
quite as nice as BootVis. Maybe it's telling us the
same thing, but I vaguely remember BootVis having a mark on
the graph, as to "when the OS was finished". The xperfview
leaves that to your imagination. In my trace, you can see the
Search Indexer gets busy after the OS has booted, so the end
of boot is somewhere to the left of the WSearch rectangle
near the bottom of the graph.

Now, one weirdness of the xperfview.exe, is there is no
"print" option! And the Save As doesn't have a way to print
off a graph either. I had to take a screen shot to get my
stinking results. And also note that, when the screen shot
is taken, the decoration at the top of the window ends up
distorted. I don't really like taking screenshots like this,
due to ClearType screwing up the colors, but this is the only
proof I can offer, of a working test run.

http://i58.tinypic.com/2vc7zuw.gif

I'll write another post, with the steps (and mis-steps) to
get to this point. It sounds to me, like you're very close.
If you'd double clicked the WindowsPerformanceToolkit.chm
help file, then looked under "xbootmgr", you'd have found
this, and this is what reboots the computer and collects
the ETL trace.

cd C:\my_result_folder --- holds the .etl files

xbootmgr -trace rebootCycle -noPrepReboot

When it complains about "no loggers" being available or
attached or something, I feel this is because the tool
is designed for both remote and local profiling. An IT
guy can have his "logger" at his desk, and point it
at the xbootmgr running on a user desktop. That allows
collecting "slow boot" info from a user machine. When
I ran xbootmgr here, the analysis is taking place on
the same machine as the capture. And I have no "remote
loggers" in my situation. So just ignore those complaints.

I don't see anything in the results (yet) to suggest
why it needs a symbol server.

HTH,
Paul

On Wed, 11 Jun 2014 15:21:06 -0400, Nil wrote:

On 11 Jun 2014, "s|b" wrote in
alt.windows7.general:

On Tue, 10 Jun 2014 21:43:19 -0400, Zaidy036 wrote:

Yes, an SSD would speed things up, but, first:
1. Reduce the number of Icons on your Desktop by placing them in
folders.

Shortcuts on the Desktop reduce the startup speed? First time I'm
reading this... :-?

No, he's wrong about that. Any impact that might possibly have would be
so slight as to be undetectable.

Not in my case.

But it's still too slow.

--
Gene E. Bloch (Stumbling Bloch)

On 6/11/2014 5:16 PM, Andrew Rossmann wrote:
In article ,
says...
Win7 SP1 64bit

My 1TB HDD has so many programs on it that the
computer takes over 5 minutes to start up.

The used space on this drive is about 300GB and is
almost all programs and operating system. I store
data on other drives.

Would a SSD give significantly faster start-ups?

Peter
What are you using as your benchmark to the point you consider the
computer 'started up'? To the login screen? Quickly logging in and then
waiting for the desktop? Waiting for all disk activity to end?

Also, have you defragged the drive recently?

Hi Peter,
I've read the thread but don't recall you mentioning which
operating system you're on, how much RAM you have and what your CPU is
that you're running. Those things also factor into how long it will take
for your system to start. If RAM or CPU, or both are below standard then
it could play a significant role in the start up time.

Jan Alter

Paul contined:

OK, here's a brief log of the steps to set up boot analysis.

1) Download .NET 4.0 if you don't already have it. This is a
"gatekeeper install", so you can't install the SDK on the
"wrong" OS. I doubt the tools really really needed this.
But we have to humor Microsoft and pay our .NET tax.

.net 4.0
http://www.microsoft.com/en-us/downl....aspx?id=17718

dotNetFx40_Full_x86_x64.exe 2/21/2011 48.1 MB

Later OSes, when they want to enforce the OS version, they
use an even later .NET which Windows 7 SP1 can't use.

The time taken for the install, is not to install the files.
The time is consumed by an NGEN run, to recompile the .NET
assemblies on the computer. This can take five to ten minutes.

2) Now get the Win7 SDK stub loader. This probably isn't the
only way to get the Performance Tools package, but it's the
way I did it. I think I was hoping "symchk.exe" would be in
here, which might account for why I was suckered into doing this.

Win7 SDK
http://msdn.microsoft.com/en-us/wind...dware/hh852365

http://www.microsoft.com/en-us/downl...n.aspx?id=8279

509,264 bytes
http://download.microsoft.com/downlo...winsdk_web.exe

3) Run winsdk_web.exe

Select a subset of all files. Don't bother with Visual Studio stuff
for example. Note that, I took this picture much later, and the
size estimates are now screwed up. This will take a while. Maybe
several hundred megabytes (because I've selected the redistributables,
and that turned out to be a happy accident). Anything not in view
here, is turned off.

http://i61.tinypic.com/2u91hja.gif

4) The installer pretends to do a good job. When you review
one of the program folders, it's virtually empty. Now, you need
to find the Windows Performance Toolkit redistributable file,
right-click it and select Install. It will say Repair as an option
as well (as if it was an Office installer), but just install it.
This gives me copies of xbootmgr.exe (capture) and xperfview.exe (view).

5) Now you can set up your symbols. My Windows 7 SP1 is x64. If you
had a 32 bit OS, you'd want the x86 retail one instead.

(symbols)
http://msdn.microsoft.com/en-us/wind.../gg463028.aspx

301,812,736 bytes
http://download.microsoft.com/downlo...RE.Symbols.msi

The received file is a .MSI. It's effectively a ZIP file of sorts,
with an auto-expander. It will ask you for a destination for symbols.
On my machine, it selected D:\symbols and I selected C:\symbols instead.
That keeps the symbol files (4000+ of them) on the C: drive.

You could also profit from setting a system environment variable.

http://i58.tinypic.com/28r2u7b.gif

I set _NT_SYMBOL_PATH system variable to
SRV*C:\symbols*http://msdl.microsoft.com/download/symbols

The second field, is where you dumped your 4000+ symbol files.
The third field, is a link to the Microsoft server, in case some
tool needs to locate files which are not present in the cache.
For example, after Security Updates, some of those 4000+ files
will be invalidated, and more will need to be automatically
downloaded.

6) Now, I created a folder to hold the .etl trace files. Open a
Command Prompt window, cd to that folder, then run the
command.

cd C:\my_result_folder

xbootmgr -trace rebootCycle -noPrepReboot

The computer immediately does a reboot at this point, so put
all your files and tools away, before the xbootmgr command.

A couple of .etl files will be put in your results folder,
and one of them will be 100MB+ in size.

7) After the reboot, you can use xperfview to open the
C:\my_result_folder stuff. The tool will complain about
"no loggers" or the like, but this is because the tool (xperfview)
is designed to also work with remote computers. And this
run was a totally local run (same machine views results, as
collected the results).

6) You get the picture of how your system booted.

http://i58.tinypic.com/2vc7zuw.gif

Mine appears to be "mostly booted" at 32 seconds.
At 35 seconds, the SearchIndexer seems to be starting up,
because we all know how important it is to add new files
to the search index.

If you're running an AV, there might be a long bar for the
AV in that picture, somewhere.

It's not BootVis, but it's the best I got so far, for Windows 7.

And not once, did I get any evidence it was using
the contents of C:\symbols. So the entire symbols exercise
might have been a red herring. It'll still come in handy
for a WinDbg session some day.

HTH,
Paul

Il Wed, 11 Jun 2014 08:58:04 +1000, Peter Jason ha scritto:

My 1TB HDD has so many programs on it that the computer takes over 5
minutes to start up.

it's not because of the space on the disk (well, just in a very few cases)

my old laptop was booting linux lxle in about 40-45 seconds, and it was
good when compared to some other o.s. here around

I put a kingston 120gb sdd drive, reinstalled the whole stuff and now the
boot is completed in 9-10 seconds

i tested also a brand new fujitsu laptop with ssd: windows 8 boots in
less than 10 seconds too



--
/-\ /\/\ /\/\ /-\ /\/\ /\/\ /-\ T /-\ ... ¯\_(ツ)_/¯