Thinkpads Forum

I'm eagerly anticipating the arrival of my new T400 and wanted to ask for some opinions about the pre-installed software. Is there anything you think isn't worth its weight? I've often wondered about all of the stuff that runs in the background of my T40p so I wanted to ask for advice on this new one.

Thank you!





Intel Core 2 Duo processor T9600
Genuine Windows Vista Business 64
14.1 WXGA TFT High Nit, w/ LED Backlight, Camera
ATI Mobility Radeon 3470 with 256MB
4 GB PC3-8500 DDR3 SDRAM 1067MHz SODIMM Memory

Start by reading thru the link (and associated sub-links) and tell us what you think might be useful for your needs ...
http://www-307.ibm.com/pc/support/site. ... START.html

Since your laptop is coming with Vista, you can do yourself a favor upon the first launch and disable prefetch/superfetch, it´s one of my favorite Vista tweaks One has to do this in 2 steps:
1. Disable superfetch service from the service console
2. Disable superfetch/prefetch from the registry: HKLM -> System -> CurrentControlSet -> Control -> Session Manager -> Memory Management -> Prefetch parameters -> set both EnablePrefetcher and EnableSuperfetch to 0

Happy ThinkPadding!

Marin

Those are both helpful. I'm digging on the Lenovo site right now. I'll be new to the machine and the OS, so I've got lots to learn.

Other recommendations are welcome and appreciated.

Thank you!

disable prefetch/superfetch

I've seen this suggestion a number of times. What is it supposed to accomplish?

Kylan wrote:Those are both helpful.

Disabling prefetch/superfetch is of very questionable real value, so you might want to think carefully about that.

Cheers,

Bill B.

I'm currently running vista business 64, everything seems alright, and the speed, honestly, was felt to be faster.

I wouldn't disable superfetch and prefetch, what it does is at the expense of memory caches your regularly used programs so that they boot faster, for example Firefox in my case. It's quite a useful feature.

I agree. Both are useful enabled.

Furthermore, I suggest you do not alter the default state of other services, as the performance increase is marginal at best, while there is a definite risk of disabling something you may need later. Many a user has 'tweaked' their system, only to encounter problems such as features/software not working as expected when actually using their system thereafter.

I suggest you just start using your system as normal and only make changes if you yourself encounter problems, lest you create them.

Enjoy your new system I love my T400

A few points regarding (disabling) prefetech/superfetch:

1. Ever wondered what causes the ridiculously long HD activity (even with new 7200rpm HDs) right after Vista has booted up -> it´s prefetch/superfetch I/Os, not Vista´s indexing. Vista´s indexing has comparatively insignificant amount of transfers. This HD activity noticeably increases the time interval from power-on to usable desktop and hence reduces productivity.

2. Prefetch/superfetch was designed to reduce application load times, particularly those of M$ applications. It dramatically fails this purpose. The time it takes to first-launch an application after boot is just the same with both prefetch/superfetch disabled and enabled.

3. With prefetch/superfetch disabled, second (and so on) launches of applications are still fast due to Vista´s basic caching: check out Task Manager -> Physical Memory -> cached. With prefetch/superfetch enabled, this doesn´t become faster.

4. Because of 2. and 3., prefetch/superfetch doesn´t reduce application load times in any case. Dead1nside, for that matter I´m almost sure that FF3 will first-load just as fast as with prefetch/superfetch disabled. Similarly, it will second-load just as fast as with prefetch/superfetch disabled. However, please note that first-load and second-load times (after boot) don´t need to be equal (and they are not so in most cases).

5. Nothing of the above applies to slower HDs like my Toshiba MK1234GSX 120 GB 5400rpm. In fact, prefetch/superfetch functionality is highly valuable in case of slower HDs, with positive gain in overall productivity that is IMO worth the sacrifice of a few minutes to completely usable desktop.

6. Disabled Prefetch/superfetch functionality is default setting in Server 2008. Even with "only" full Aero added, I´m pretty certain most people find or would find Server 2008 quite snappy.

I don´t see anything questionable in that at all!

Cheers,

Marin

Off-topic: Unlike with Vista, I found Win 7 prefetch/superfetch functionality to be greatly improved. For plenty of programs, it makes their load times shorter. What is more, Win 7 doesn´t suffer from those same after-boot HD thrashing symptoms as Vista did (not anymore for me ) because the prefetch/superfetch transfers in Win 7 are actually "stretched" over certain time interval after the boot in smaller transfers.

My experience with superfetch enabled (and the vast majority of suggestions from the experience and wisdom of jdhurst found here: http://forum.thinkpads.com/viewtopic.php?f=38&t=62321):

Point 1: My time from power on to “useable desktop” in less than a minute. (Nevertheless, boot time is an irrelevant statistic to me.) HD activity in the background does not prevent using the machine almost immediately after the desktop appears.

Point 2: Over time (specifically, January, 2009 to today) first loading of all apps (even non MS apps) has become quicker. Granted, not dramatically so but quicker. Granted, not instantaneously. If I recall, I noticed it sometime in February so call it a month with superfetch enabled before a noticeable improvement.

Point 3: My understanding is superfetch is designed to decrease first load times. If that’s the case, second launch times are irrelevant in a discussion of the benefits/disadvantages of superfetch. Re task manager, Reliability and performance monitor (Performance information and tools ->advanced tools->reliability and performance monitor) generally shows something less than 1/3 of my 3gb ram used.

Point 4: See my point 2.

Point 5: I don’t have different speed drives to test, but I doubt the assertion. I doubt a 7200 drive reads data faster than it is read out of memory. (I’d guess that’s true even with an SSD)

Point 6: No knowledge of the standard Server 2008 configuration, but it’s somewhat like comparing apples to oranges.

My advice, based on my experience, would be the opposite. That’s my data set. If there are data sets that are contrary, where superfetch didn’t work, I’d like to know.

dsvochak wrote:Point 1: My time from power on to “usable desktop” in less than a minute. (Nevertheless, boot time is an irrelevant statistic to me.) HD activity in the background does not prevent using the machine almost immediately after the desktop appears.

No, it doesn´t prevent using the machine, but it significantly slows it down (like when starting (prefetched ) programs, opening files etc). How long one is willing to wait for a file to open is then rather a matter of personal preference. In any way, during this time interval of more intensive HD activity, load times are longer than without this HD activity.

dsvochak wrote:Point 2: Over time (specifically, January, 2009 to today) first loading of all apps (even non MS apps) has become quicker. Granted, not dramatically so but quicker. Granted, not instantaneously. If I recall, I noticed it sometime in February so call it a month with superfetch enabled before a noticeable improvement.

I assume here it´s not just the user feeling over time with Vista that the loading times are becoming shorter... I have been using Vista for more than an year and have never been able to tell any difference between both scenarios, neither with Hitachi 7K200 nor with Seagate 7200.3. You did it in a month!

dsvochak wrote:Point 3: My understanding is superfetch is designed to decrease first load times. If that’s the case, second launch times are irrelevant in a discussion of the benefits/disadvantages of superfetch. Re task manager, Reliability and performance monitor (Performance information and tools ->advanced tools->reliability and performance monitor) generally shows something less than 1/3 of my 3gb ram used.

Yes, your understanding is basically correct. If enabled, prefetch/superfetch also includes the "basic" Vista caching functionality, so in a way, if enabled, prefetch/superfetch are indirectly responsible for second load times too (rather formal distinction). Apart from this, second launch times were relevant in my post only as far as the post of Dead1nside is concerned about FF3 loading fast because it was inconclusive to me in what way he meant FF3 was loading fast. As for Task Manager, with Vista (and Win 7) the raw memory usage is completely irrelevant. What matters is cached memory: how many of it is left, what can be cached, what has been already cached, what is not anymore cached.

dsvochak wrote:I don’t have different speed drives to test, but I doubt the assertion. I doubt a 7200 drive reads data faster than it is read out of memory. (I’d guess that’s true even with an SSD)

That is certainly true. However, prefetch/superfetch is not a complete ersatz of the HD, it doesn´t load straight-forward everything into system memory, it loads only certain components (I guess that pretty much depends on the available system memory). First application load always requires HD participation, that is the general bottleneck. The point is that prefetch/superfetch doesn´t work well enough to overcome that bottleneck in case of HDs with better performance.

dsvochak wrote:Point 6: No knowledge of the standard Server 2008 configuration, but it’s somewhat like comparing apples to oranges.

Same kernel, added server functionality, installs on a ThinkPad with most things working without problems, needs some tweaking though as default settings are optimized/rationalized for server systems. So, why "somewhat like comparing apples to oranges"

I would be happy if you share your system configuration including startup programs and OS install (e.g. clean or Lenovo preloaded).

Have you ever actually tried running (clean install of) Vista with prefetch/superfetch disabled from the very beginning on?...

Cheers,

Marin

Marin85 wrote:I don´t see anything questionable in that at all!

My reality is that your observations are not reproduced on my Vista systems, therefore they are quetsionable.

Cheers,

Bill B.

I mean startup performance of the application, how long the program it takes to load and become usable. You're saying that this is at the expense of a usable system at boot time. However with 2GB RAM and a dual-core 2Ghz processor my T61 both boots quickly and launches applications quickly, almost instantaneously - so I see no need to disable prefetch.

Edit: That's with a 7200RPM hard drive, as per my signature.

bill bolton wrote:My reality is that your observations are not reproduced on my Vista systems, therefore they are quetsionable.

So, you have tried out running (clean install of) Vista with prefetch/superfetch disabled from the beginning on?
I don´t know, maybe my system is defective in some way, but I have been always able to reproduce these observations with 2 different 7200rpm HDs and both Vista Business x64 SP1 (MSDN media) and Vista Ultimate 32-bit with SP1 installed via windows update (installation media from my university). My point here is not so much the HD thrashing after boot, I can imagine this might be different for different users, depending on machine specs, user activity and amount of installed programs (in my case quite a lot of them, so no wonder if prefetch/superfetch would be trying to preload a bit of every). My point is rather that Vista is just as fast with prefetch/superfetch disabled (from the beginning of a clean installation on).

Cheers,

Marin

1) If the same services, processes and programs are running, why would there be a difference related to whether one had a clean installation or the preload?

2) With Server 2008, as Marin noted, “default settings are optimized/rationalized for server systems”. Server OS v Desktop OS = Apple v Orange.

3) “I assume here it´s not just the user feeling over time with Vista that the loading times are becoming shorter...” One of the first things I did upon getting the machine set up was to time, via hand operated stopwatch, how long it took Word 2003 to load a 50 page Court of Appeals Brief (a double spaced, 12pt font, 1 inch margins document with 50 footnotes). Initially, it took 3-4 seconds as the first use of Word after boot. Second load of the same document (closing Word entirely not just closing the document) could not be timed. It was too quick to start and stop the stopwatch. The load time has reduced from 3-4 to 2-3 seconds. I’ve noted a similar result with other Office 2003 programs, Word Perfect Office, Lotus SmartSuite, Paperport, PDFill and various utilities. As I said, it’s not a dramatic improvement, but it’s quicker.

4) I’m not sure what constitutes “quite a lot” of installed programs. Regardless, I don’t think superfetch is designed to “preload a bit of every” program available. My understanding is superfetch makes an “educated guess” of what’s likely to be used rather than what is simply available.

5) Per request: T61 7659-1LU, T7100 (1.8GHz), 3GB RAM, 120GB Seagate ST9120817A 5400rpm HD, Vista Business 32 (The machine is the forum project T61). OS: Lenovo preload done by Harryc 11/21/08.

Services: Vista Business defaults except Tablet Services, Offline Files, Tvtsched service and System Restore disabled. Lenovo apps removed from startup and/or uninstalled: AC, CSS, R&R, System Update, Presentation Director, Easy Eject, First run offers launcher, and Keyboard customizer. (It’s not that any of the items disabled are bad, I just never use them)

Others removed from startup or off: Windows meeting space, Diskeeper, Acrobat launcher, Java update, and the Intel tray application

Current startup programs (beyond drivers and normal windows stuff): Core temp, Windows defender, power manager, hotkeys, on-screen display, thinkpad productivity center, Paperport and Avira AV.

As previously noted: boot time from off to usable desktop < 60 seconds.

6) The HD working does not increase the time to load the Brief referenced above.

1) Not sure what kind of difference you mean here (general differences or only differences related to prefetch/superfetch functionality) Thanks in advance for clarifying this. Maybe just one general note: Preloaded images have history whereas clean installations don´t.

2) Not really "Apples vs. Oranges" because one changes most default settings of the Server OS to make it "look and feel" just like the ordinary OS (depends though on one´s needs). My point was only that Server 2008 loads applications just as fast as ordinary Vista, but unlike ordinary Vista it has prefetch/superfetch disabled per default.

3) You have used stopwatch, controlled by human hand, and you have measured difference from about 1 second... With all due respect, I now really have some doubts as to the correctness of your measurements.

4) Correct, prefetch/superfetch supposedly makes an educated guess, based on monitored user activity.I was only referring to programs I use on regular basis as they are the most relevant for prefetch/superfetch, and I have more than just 5-6 such programs. Most of them come with many plug-ins and shared components.

5) Thanks for mentioning your specs. To be honest, I have hard time to believe that your Vista installation boots up to usable desktop in less than 60 seconds. It´s true that one can click on an icon as soon as the desktop has appeared, but that doesn´t make it really usable IMO. Usable would be if you can start browsing the internet or open word document right away without so noticeable delay (what is noticeable delay, is matter of personal preferences though).

6) Any additional HD activity does increase the time to first-load an application (except for the really tiny ones), difference could be sec or msec, depending on how heavy the application in question is, for the simple reason that first-loading an application under Vista needs significant access to HD. That´s why first loads are always slower than second loads (and for that matter prefetch/superfetch is actually supposed to make first loads feel more or less like second ones, so you tell me, does first-loading any bigger application feel that close to instant?...). The bigger the application, the more noticeable the delay.

Cheers,

Marin

One is entitled to doubt the accuracy of my measurements. The general result, however, is consistent and not doubtful: over time, initial loading time for apps has declined. I’ve provided the data available which supports my statements. With all due respect, no data supporting contrary statements has been presented.

“Usable would be if you can start browsing the internet or open word document right away without so noticeable delay”. It appears there is agreement on the definition of “usable”.

I wouldn´t call anything provided in that thread "data" that would support one´s statement. Similarly, my "measurements" don´t show any improvement in load times whatsoever for more than a year of constant use of certain applications.

I forgot to mention in my previous post that looking at your specs I see that you have a 120 GB 5400rpm Seagate HD. I´m not sure if this belongs to the performance category of HDs like Hitachi 7K200 and Seagate 7200.3. As stated earlier in this thread, I have seen the supposed performance gain with my Toshiba 120 GB 5400rpm, so this might well explain why we have different experience (you have already asked me why HD performance would make a difference here, and I believe I have responded to that somewhere above).

Kylan wrote:<snip>I'm eagerly anticipating the arrival of my new T400 and wanted to ask for some opinions about the pre-installed software. Is there anything you think isn't worth its weight? I've often wondered about all of the stuff that runs in the background of my T40p so I wanted to ask for advice on this new one.

1) First I would follow this great advice from Harryc:

" Start by reading thru the link (and associated sub-links) and tell us what you think might be useful for your needs ..."
http://www-307.ibm.com/pc/support/site. ... START.html

2) Then I would use your new ThinkPad just as it is, do not change it, for at least a week or two.

The benefit of that is if something goes wrong, it's not due to something that has been changed. I buy a huge number of new ThinkPads and find that if anything goes wrong it usually happens in the first week or two. If it does, it's usually something simple that is easy to resolve by using our help here. If you have not changed anything, it makes it very easy for us to help you isolate your problem. Then once your new laptop has a track record of reliability (not that they don't) and you are familiar and comfortable with it, then begin your mods.

Congrats on your new ThinkPad... Enjoy it!

Cheers...

In this thread we have Marin85 advocating disabling superfetch while bill Bolton, Dead1nside, neenee and myself indicate, as bill put it, “Disabling prefetch/superfetch is of very questionable real value”. As it appears these positions are fixed, I don’t see a point in continuing the discussion. Thanks.

Nobody here knows the memory architecture of Vista expertly enough to predict how and why enabling or disabling superfetch would be useful for particular systems--unless there are Microsoft employees who worked on the Vista team in here... all we have are the PC's we own and can test personally.

I was intrigued by this tweak, and performed it on my X61 Thinkpad. Relevant data: 2GB memory, 100GB 7200RPM HD, Core 2 Duo T7300 chipset, with Vista Business. Not applicable to a T400 PC then, but just another data point. The tweak eliminated the excessive hard disk spinning that slowed my OS to a standstill for 3-4 minutes after start-up. Did it hurt performance as measured by program start-up times? Crude stop-watch measurements were indefinite. However, I did not observe any noticeable slowdown when opening up Word, Excel, Outlook, Photoshop, and other memory-intensive programs several times. I've used it for days now and still cannot time a noticeable slowdown in opening up these software executables at various time-points (after boot; after hibernation and wake-up; after opening and closing other programs; after defragmentation; after watching a movie; etc).

My conclusion based upon this one machine is that the tweak succeeded in eliminating the excessive hard disk spinning that virtually crippled Vista Business after booting up. It may have improved or worsened load times for large, memory-intensive programs, but I could not detect a noticeable difference using either my own perception or a stopwatch (I normalized data to give me a standard deviation of time as acceptable error--ie, .15 seconds either way was considered not significant for loading time of Outlook).

It appears this tweak is machine-dependent. It remains to be seen whether performing the tweak can potentially cripple some machine configurations, but it would be logical to assume that provided the T400 has the memory, hard disk speed and space, processor capacity, and OS capability comparable to my X61, then disabling Superfetch will not drastically reduce performance.

I'd be interested in getting more T400 users to test it out on their machines, since I'm about to get one myself!

The superfetch option is an interesting discussion topic, it is highly discussed in low speed ssd forums (tweaking windows for cheap mlc ssds).

It appears this tweak is machine-dependent.

Perhaps this is the statement that best sums up the discussion. As the saying goes YMMV. Thanks for the input.

I say Kudos to the participants that have passionately brought forward their experience, perspective, and shared their knowledge. Personally, I've found this discussion quite interesting, educational and lively. Fulfilling the very purpose of a forum like ours.

At the end of the day, we all know the intricacies of the topic and the variables in involved. This has been like a good brainstorming session to a certain degree, and I'm enjoying it.

Cheers...

I had some free time yesterday so I experimented some, adjusting services, to see whether I could impact boot time. I did find one that does: disable ReadyBoost. Unfortunately, the effect was negative: boot times about doubled. After some searching, I found this explanation:

“Not to be confused with ReadyBoost, ReadyBoot is another feature designed to use memory to optimize the boot process, but ReadyBoot uses normal system RAM to do this rather than an external device. After every bootup, ReadyBoot calculates a caching plan for the next boot and stores part of this information under the \Windows\Prefetch\ ReadyBoot folder, and part in the Registry. The end result is that each time you boot up Vista, ReadyBoot can improve boot times through use of this cache. After bootup the memory used for caching is automatically freed up after 90 seconds, or sooner if required.
Importantly, ReadyBoot relies on the ReadyBoost Service to function, so you should not disable this service even if you don't use ReadyBoost; it should be left at Automatic”

Quote from, Tweaking Companion for Windows Vista by Koroush Ghazi, Version 2.3, page 121 (Copyright 2009) http://www.tweakguides.com/

Immediately above the quote above, Mr. Ghazi writes “Note that if you’ve disabled SuperFetch then ReadyBoost will have no impact.” Whether SuperFetch was disabled or not seemed to have no impact. Disabling ReadyBoost, as noted had a significant impact on boot time.

Do those who have SuperFetch disabled also have ReadyBoost disabled? If you do, what happens to boot time if you enable ReadyBoost?

dsvochak wrote:Do those who have SuperFetch disabled also have ReadyBoost disabled? If you do, what happens to boot time if you enable ReadyBoost?

I´m not sure about Vista, even though a friend of mine once tried out the famous Readyboost feature on my ThinkPad (still with Vista at that time), but on my Win 7 x64 setup I can confirm that stopping/disabling Superfetch service disables any Readyboost functionality.

Perhaps I was less than 100% clear. Both the quote and my post relate to the ReadyBoost service and are not related to using a memory device as ReadyBoost memory.

dsvochak wrote:Perhaps I was less than 100% clear. Both the quote and my post relate to the ReadyBoost service and are not related to using a memory device as ReadyBoost memory.

In Win 7 there isn´t the Readyboost service anymore. In Vista, if Readyboost service is running, then one can use Readyboost drives. If Readyboost service is not running, one cannot use Readyboost drives. The service dependencies of Readyboost service are as follows: reference -> Readyboost service at blackviper

Since superfetch service isn´t mentioned among Readboost service dependencies, I don´t think that disabling superfetch service in Vista disables Readyboost as well. Applying the superfetch/prefetch registry "tweak" from above doesn´t disable Readyboost service or functionality in Vista either.

To quote again, emphasized this time:

"Not to be confused with ReadyBoost, ReadyBoot is another feature designed to use memory to optimize the boot process"

As the quote indicates, turning off ReadyBoost service also impacts ReadyBoot.

My question is not whether disabling SuperFetch disables the ReadyBoost service. It is whether those who have disabled SuperFetch also disabled ReadyBoost, and, if so, whether enabling the ReadyBoost service impacts their boot time.