34 replies to this topic

[antonino61]

2gb dynamic vhd 600mb free space - filedisk - bootmgr booting 20secs - freespace decreasing over time

2gb dynamic vhd 600mb free space - ramdisk - preramloading + booting 26secs - freespace decreasing over time

2gb fixed vhd 600mb free space - filedisk - bootmgr booting 17secs - freespace basically staying the same

2gb fixed vhd 600mb free space - ramdisk - preramloading + booting 21secs - freespace basically staying the same

4gb fixed vhd 2.4gb free space - filedisk - bootmgr booting 16.5secs - freespace basically staying the same

4gb fixed vhd 2.4gb free space - ramdisk - preramloading + booting 20.5secs - freespace basically staying the same

1.5gb fixed vhd 270mb free space - filedisk - bootmgr booting 17secs - freespace basically staying the same

1.5gb fixed vhd 270mb free space - ramdisk - preramloading + booting 21secs - freespace basically staying the same

 

the above mentioned tests were made on an x299 cpu i9 7900x 64gb 2400mhz ram nvidia gtx 1070 gpu. the used space has always been occupied by the same content (sheer win10 setup, debloated, programs and features installed 32 entries basically consisting of drivers, visual c++, directx, primocache and primoramdisk). the rest of the apps are portable so they have little or no influence on registry and overall c:\ drive content. the most noticeable difference in favor of the smallest sizes is one of space occupied on disk and in ram, not of time taken to preramload and boot.

nino

 

 

[alacran]

@ antonino61

 

It seems to me you can't follow any instructions, or maybe do not understand fine what I said or the strong reasons for it:

 

 

concentrate only on time to load in Ram since it is the only thing you can more or less cotrol by means of file size to be loaded, the total booting time is very much more influenced with each program to be loaded during boot and can vary a lot on each user.

 

Also I have dissabled pagefile and swap file, and superfetch too, they are unusefull if Rambooting.

 

I allready have datas from VHDs in these way and in order to compare with anybody else datas those datas require to be extracted from a VHD made exactly same way if not we can't compare apples and oranges.

 

Also you said you have installed primocache and primoramdisk that also influence the size of VHD and loading time.

 

Another thing I don't know is if you have enabled or not any pagefile and swap file.

 

So only way we can compare our info is if you make a complete from scratch new build without pagefile and swap file and also without primocache and primoramdisk installed, just for the purpose of this test, then our OSs on Wimboot VHD would be more similar, and VHD used size and time to load on Ram could be compared in a more aceptable way, (latter, after the tests you can comeback to use your actual VHD if you want).

 

I can't see any benefit on having primocache and primoramdisk running on a system that is allready loaded on Ram, both only add additional unnecesary garbage on the Ram + the time required to load it on Ram and latter during boot, even if you are not using at the moment any of the cached info.  Those programs may be of some benefit (if any) on a system installed and running from a very slow HD and with a lot of faster Ram available. This could explain the very big space used on your VHD.

 

I allready explained very clearly the reasons for this requirements.

 

I'm sorry my friend but unless you sintonize in the the same frequency I am, I'm not going to answer or make any comment to info that I can´t value against my datas. Since it is a waste of time for both of us.

 

Your friend:

 

alacran

[antonino61]

my dear friend alacrán, pagefile and swapfile have never been there for the past 10 years here, whether on plain system or on vhd or rambooting or wimbooting or wimrambooting etc. as regards primoramdisk and primocache, they have always been there, for the following reasons: primo ramdisk enables me to save life to ssd by keeping temporary files in ram; primocache betters write speeds so it does affect disk speed even in the case of ssd (even though I agree with u does not affect read speeds) and does not cache the vhd when the system is preloaded into ram (so it is practically kept for the other disks, not for the system disk). I do not know how to disable superfetch, so pls tell me how to do it.

pls dont imitate wonko with your claiming that u cannot compare the results. You can compare the results, as the results have been obtained by changing sizes not content, which as I have already pointed out stays the same. read more carefully - it is not potatoes and onions, it is potatoes of different sizes or onions of different sizes, so u can compare these situations here. I agree with u that u could not compare ur results with mine, that would be potatoes and onions; my result a with my result b with my result c with my result d u can compare. come on.

As for the preramloading time that u wanted, there u have it, no significant difference for so small amounts (6secs), if u wanna compare these amounts to 20gb preramloading, then u do have sound difference.

nino

[alacran]

You talk too much about things you don't fully understand:

 

JFYI:

 

Primoramdisk redirects %USERPROFILE%\AppData\Local\Temp (on 7, and also some other on 10 wich I don't remember at the moment) to a location on a Ramdisk, same that can be done with any other Ramdik program as SoftPerfect Ramdisk (there is a free version I use on some PCs), but only if I have the OS installed and booted the standard way on HDD or SSD it reduces wear on both, but totally unnecesary when Rambooting.

I use SoftPerfect Ramdisk (on standard OS installs only) to also hold Firefox Browser cache (an average of 2 or 3 GB a day) and Avast definitions (defs folder) about 180 MB (splited on hundreds of 1 to 3 KB files written to HD almost every hour) by means of a Juntion Link.

TEMP and TMP usual location on Windows OSs is  %USERPROFILE%\AppData\Local\Temp = C:\Users\Your User Name\AppData\Local\Temp But during Ramboot your OS (C:\) is loaded on Ram, so no need any additional program to redirect to Ram something that will be allready on Ram, sounds stupid to load another program to do just exactly what your OS will do, and if you could need more space on your Rambooted VHD for your TEMP files just make it biger (you have plenty of Ram), as it will be empty and compressed on every boot there is no additional load/time during loading and booting. According with your datas on first post 4 GB is the best size option with your actual OS. But if you follow my instuctions we may find another size that fits better.

 

I had the need to create a 512 MB NTFS Compressed Ramdisk (A:\) to hold FireFox Profile (loaded every boot by a very little program I made using Bat to Exe Converter) and FireFox cache on a PC with only 4 GB of Ram because a 2 GB VHD can't be Rambooted on it, AFAIR VHD of 1,700 MB is the Rambootable VHD size limit found during my tests when having 4 GB of Ram but this is in no way your case.

Primocache is just a lie to scam fool people, Windows itself has a tool to do this since XP (prefetch) and improved on Win7 (prefetch + Superfetch) and latter, it does exactly same thing as superfetch does, it preloads to Ram all program you have ever used on your PC in the past on every boot, even if you are not using them at the moment, and I think superfetch has a better handling of the datas since it is part of the OS.

All SSD programs made by SSD builders to control/improve SSD performance and live, usually dissable Superfetch, and I'm not totally sure at the moment but AFAIR Win10 does it by default on SSDs too, since with the high speed of the SSD there is no need to preload to Ram anything, and this way they also reduce wear on the SSds since only what is needed will be readed, remember not only writing but also reading wears the SSD + since every time a program is readed Windows update info in it as Last acces time just to mention one. So using this program you are not protecting your SSD, on the contrary you are inducing excesive wear on it with this stupid program.

To disable superfetch or make sure it is disabled just open Services (services.msc) and on Win10 Find SysMain and disable it, on Win7 find Superfetch and disable it.

Or run System_Info from wimb and select Drivers + Services >>> Services: And find and disable the service.

 

So please follow instructions on my previous post for a better understanding of the info, since then we will be able to compare your info with my info and it will benefit not only us but all future readers.

 

alacran

[antonino61]

My dear Alacrán, 

talking too much about things one does not understand is not always a problem; it might also be a solution. I would never have gathered all this info from u or anyone if I had not talked too much. Thank u for ur patience, I will never stop repeating this. now that I have increased my knowledge a little bit, pls tell me how much bigger my vhd should be on account of these temp directories back on it for daily operations, including browser cache.

As for primocache, I appreciate your learned comment (btw, u r not the only one who has told me that), but I cannot fail to report a noticeable difference in testing between read and write speeds of the disks "cached" and read and write speeds of the same disks "uncached" (I am talking about the other disks, not c:\, of course, as it is not cached by primocache when it is ramloaded). the differences are detected both with crystal disk mark and AS-SSD-Benchmark2.0.6821. now I am longing for ur learned answer and thank u in advance.

nino

[alacran]

My dear Alacrán, 

pls tell me how much bigger my vhd should be on account of these temp directories back on it for daily operations, including browser cache.

 

As for primocache, I appreciate your learned comment (btw, u r not the only one who has told me that), but I cannot fail to report a noticeable difference in testing between read and write speeds of the disks "cached" and read and write speeds of the same disks "uncached"

 

nino

 

Maybe I was editing my post when you were writing, the answer to start with is in my opinion:

 

 

According with your datas on first post 4 GB is the best size option with your actual OS. But if you follow my instuctions we may find another size that fits better.

 

About your comment of primocache:

 

 

I cannot fail to report a noticeable difference in testing between read and write speeds of the disks "cached" and read and write speeds of the same disks "uncached"

 

My opinion is:

I prefer to sacrify a few speed to avoid wear and preserve the live of my SSD where all my preciated info resides.

I mean this way the info will be read and used only when needed, not every boot and then inducing a lot of unnecesary wear.

 

As I have told you several times before make a new build from scratch exactly the way I suggested just to test (and keep safe your actual VHD so you can comeback to it if after testing the new build if you prefer the old way), and then if you publish your new datas just the way I told you we both together can find what better fits to your needs. Remember only VHD full size on MB, VHD used space on MB and loading to Ram times are required, you may use your Celphone Chronometer to measure time, and if not sure of timing repeat the test.

 

Please excuse me if I have been rude sometimes but you desperate me when I have repeated something several times and you seem not to understand.

 

alacran

[antonino61]

my dear alacrán, I have 2 copies of vhd's, so I have picked 1 up and made the changes that u suggested. No primocache, no primoramdisk, I rebooted and ran a disk speed test whose results are shown in capture.png. then I rebooted off the unchanged vhd (primos included) and ran the SAME disk speed test whose results are shown in capture1.png. Besides, the unchanged vhd boots 1 second faster than the one without the primos.

Now would u pls explain to me how this could be? what do I have here that differs from what u say?

nino

ps: I then lz4'd the primoless one and ramloaded it --> the usual 6 more seconds to preramload + 18 secs to reach the interface (1sec longer than the primo vhd (6+17secs). how so?

Attached Files

•  Capture1.PNG   353.57KB   0 downloads
•  Capture.PNG   508.5KB   0 downloads

[antonino61]

it is not a minor difference in the pictures. the difference might be minor in the booting times, but paradoxically in favor of the old build! as for rudeness, we r not talking characters and personalities, we are talking facts! 

 

while u answer my last, I would like to try it with primocache and without primo ramdisk and see what changes (the primoless vhd shows no bulge, but actually a steady used space, which means the temps on it do not vary that much, in spite of my usual browsing). 

nino

[alacran]

No you didn't make the a new build from scratch, you deleted the programs wich is not the same as they never errase completly and there are always remains on the register and other places, so that is not what I said.

 

And with that speed test you are only fulling yourself measuring Ram vs SSD performace, of course it is obvious Ram will be always faster.  Do not waste your time testing this anymore Ram speed will always win.

 

There is something you really don't understand, OSs have timings preset for everythig like an example opening the menu has a certan delay, you may change it (AFAIR changing it from 400 to 100 ms is what I used to do on XP) and make people think you PC is very fast, during real life there is a point where the maximum possible speeds a device is capable to do something do not matter so much because the OS itself and also each program or another hardware will limit the real results or the human perception will not notice any diference.  When dealing with this case the human perception is what realy matters. If it is fast enought for you, then it is OK.  And also of course do all required to keep our HDD or SSD on good shape to avoid wear with the risk to lose our valuable info.

 

alacran

[antonino61]

done. as u can see, capture3.png yields a slightly better scenario. besides, half a second faster boot time. so far, what u said on primo ramdisk proves u right and what u said on primocache proves u wrong. how so?

I have also devoted fewer ram resources to primocache, only 4gigs, which gives the best disk speed results, paradoxically better than giving it 8gigs.

as a result, I have only 16% ram usage. do u think my system would benefit from taking 2 16gb ramstix off and making it 32gs as opposed to the present 64gb? I might keep them off the rig for later replacement if their presence on the rig is counterproductive speedwise. what do u reckon?

nino

ps: I have just lz4'd the primoramdiskless primocache vhd and rebooted it --> 6secs preramloading + 16secs boot completion = 22secs overall, which seems the best scenario to me so far.

Attached Files

•  Capture3.PNG   348.81KB   0 downloads

[alacran]

As I said before:

 

 

Ram will be always faster.  Do not waste your time testing this anymore, Ram speed will always win.

 

As allways it is a matter of preferences, but I'm quite sure you don't need 64 GB.

 

I started using PCs long time ago with an Apple II: NO HDD, only two 5 1/4 " Floppy drives, one for Apple Works (the program) and the other for your files/folders and AFAIR just 4 KB or maybe less of Ram and all was made by command line (then from there is my hate to command line) latter had a Win 3.11 PC with a very little and slow HDD (don't remember more info) and it was a big diference since almost all was graphic, no more typing very prone no commet mistakes, but this is only to tell you latter had an HP 386 at 33 MHZ and I had a dos game that had a demo you can run to show you how to use it, well it was all fine until my next HP 386 at 66 MHZ then on that PC it was really not possible to see the demo running, you just start it and it was finished and you saw nothing.

 

I think that this is an example why Windows and also all programs have introduced delates to let the humans interact with PCs.

 

So as said before:

 

 

If it is fast enought for you, then it is OK

 

Just remember almost all on PC business is to make you buy the new model faster than light speed and with a bunch of unnecessary Ram and processors, (resourses that you will never need unless you work for NASA or similar)

 

Even today not all programs are capable to use more than 2 processors with 2 threads each = a total of 4 threads.

And very few programs can use big quantities of Ram, excluding only some games and the resource waster Autocad that used to be about 400 MB on XP and worked very fine on 128 MB of Ram.

 

alacran

[antonino61]

My dear alacran, what makes u so sure I erased the primos instead of uninstalling them? Or that I said that I have more satisfactory results in ramloading, when l actually have the same timing except for the last case? Oh I see, the part we cannot control. The part u CAN control takes 6secs all the time here. Who told u that my timings are based on perception instead e of a cell phone stopwatch? Who told u th
at what u say of pc software, which is correct, applies differently to equal situations. Why are u prepared to assert the unscientific in order to pontificate? U want a system from scratch? U will have the results on that too, as soon as I get home, because I have it. It is the source of all my Wimboots. 11gigs takes 60secs to load on ram. I will carry out the same tests and let u know, nota bene the same tests as I have, not other tests,that would be unscientific. I hope wont tell me that I got those results instead of what I should have got because I did not wimboot it
First. Later

My dear alacran, what makes u so sure I erased the primos instead of uninstalling them? Or that I said that I have more satisfactory results in ramloading, when l actually have the same timing except for the last case? Oh I see, the part we cannot control. The part u CAN control takes 6secs all the time here. Who told u that my timings are based on perception instead e of a cell phone stopwatch? Who told u th
at what u say of pc software, which is correct, applies differently to equal situations. Why are u prepared to assert the unscientific in order to pontificate? U want a system from scratch? U will have the results on that too, as soon as I get home, because I have it. It is the source of all my Wimboots. 11gigs takes 60secs to load on ram. I will carry out the same tests and let u know, nota bene the same tests as I have, not other tests,that would be unscientific. I hope you wont tell me that I got those results instead of what I should have got because I did not wimboot it
First. Later

[Wonko the Sane]

 

pls dont imitate wonko with your claiming that u cannot compare the results. 

 

Why are you taking my name in vain?

Being the only and inimitable seems to be not enough to avoid people accusing other members to attempt (again vainly) to imitate me. 

 

 

So please follow instructions on my previous post for a better understanding of the info, since then we will be able to compare your info with my info and it will benefit not only us but all future readers.

 

To be fair, you could instead add primocache and primoramdisk to your build and still make a valid comparison.

 

 

Wonko

[antonino61]

Of course, ur last remark makes it all the more understandable. My congrats, what else can I say? I will continue using primocache as long as it (albeit inexplicably) increases non- system disk speeds. I have already stopped using primo ramdisk as it proved messier than the ordinary windows temp location. No other reasons besides the above. As for characters and personalities, if there is any room left and for all anyone should care, u once pontificated out of pure assumption on non-facts and even alacran told u and now he does the same on me;well, as long as I do not do it, everything is fine. Me I would stick to tests.

[Wonko the Sane]

 u once pontificated out of pure assumption on non-facts and even alacran told u 

Did I?

Did he?

Where/when?

 

Wonko

[alacran]

To be fair, you could instead add primocache and primoramdisk to your build and still make a valid comparison.

 

 

Wonko

 

No, since I will not have same info he has loaded/attached on both.

 

Please Wonko I didn't expect this foolish assumption from you, I start thinking the age is affecting you my friend. HA, Ha, Ha

[alacran]

@ antonino61

 

I answered because you asked, no more no less.

 

You are free to do whatever you like/prefer on your install, after all it is your PC.

 

alacran

[antonino61]

I guess u both are getting the wrong end of the stick. It is not a matter of authorization or susceptibility; it is one of opportunity and convenience. I do not mind testing, but I find it hard to accept that my tests are inaccurate or unreliable because I have changed only 1 variable and left the others intact. Whether there are 0,1,2,3,30,300 or 3000 of them, changing only 1 is actually what should make them reliable! If anybody thinks otherwise, that is pontificating! It has happened at the end of 1 test on the smallest wimboot vhd and alacran replied to Wonko in almost the same way as a few minutes ago. Anyway this is absolutely not the point that I am at pains trying to make. Now I will test IN THE SAME WAY, not in another way, what I have already promised to: whether alacran, Wonko or any info-tech pundit will care to take them into account or not, I am nobody to say whether it is worth their while affording to miss out on them - they have all the experience to see what suits them best.

[Wonko the Sane]

No, since I will not have same info he has loaded/attached on both.

 

Please Wonko I didn't expect this foolish assumption from you, I start thinking the age is affecting you my friend. HA, Ha, Ha

Ow, come on, come off it.

 

@antonino

Your tests are not inaccurate or unreliable, I firmly believe that you are accurate and reliable, the issue is that even if a teeny-tiny variable, setting or *whatever* changes then the tests are not comparable.

 

The point I was trying to raise was that not only your build/settings are "wrong"[1], but also alacran's ones are "far from being right"[2], and unless you find an EXACT way (as an example using a script) to replicate two IDENTICAL builds with SAME EXACT programs/settings the comparisons will always be "skewed" or however not meaningful.

 

Making tests (and comparison tests particularly) of an OS speed/behaviour is among the most difficult thing to do exactly because there are so many variables and - particularly when in the experimenting phases - there is little knowledge on the actual consequences of those that may appear as very minor differences in the setup.

 

Wonko

 

[1] in alacran's opinion, that is

[2] IMNSHO, that is, because alacran may have applied a number of tweaks or particular settings in his builds, or to the VHD, or to *whatever* without providing an EXACT way to reproduce them 

[antonino61]

my dear info-tech pundits, a big apology. My latest findings prove wonko and alacrán right and, alas, wimb wrong. Yesterday I promised I would perform the same tests as I had done lately on a virgin vhd (no wimbooting this time) I had saved someplace on a hard drive. well, I had to chuck it away because it was not updated. so I set out to reversebake a new virgin vhd from my present wim. this wim has allowed me to bake and run primoramdiskless vhds associated to it, so I thought now I would get a virgin unassociated vhd by simply applying it with gimagex (or other, that's not the point here). to my surprise, the resulting unassociated vhd included both primo cache and primo ramdisk!. this means that whatever change is made on the vhd in setup or configuration terms might as well not be mirrored in the wim. this is a serious problem, as one will easily lose sight of the changes. this is surely why alacran would shout "no no no, u did not start it from scratch!". well this is the reason for my delay, pls let me right this wrong and I will perform my tests, for u to compare among each other (an intra-comparison), not for u to compare it with urs (an inter-comparison), which would prove wonko right. I find intracomparisons indirectly useful if interpreted heuristically.

nino

later 

[antonino61]

here u go, as I said, no significant difference between vhd+wim and full vhd in terms of booting time 18vs17secs, NB not 17vs18secs, understood alacran?

ramloading test will follow, but I can anticipate only a 6sec delay (preramloading here).

Let us see.

in a bit

nino

[alacran]

@ Wonko

It has never crosed for my mind my build is perfect, that's why I want to compare to another build similar to it. And try to improve it and the info for future readers if any.

Only thing I have been asking for is this:

From Post No. 2

So only way we can compare our info is if you make a complete from scratch new build without pagefile and swap file and also without primocache and primoramdisk installed, just for the purpose of this test, then our OSs on Wimboot VHD would be more similar, and VHD used size and time to load on Ram could be compared in a more aceptable way, (latter, after the tests you can comeback to use your actual VHD if you want).

I highly suspect that during my tests the loading to Ram time didn't reduce as much as I expected using a LZ4 compressed VHD compared with an expandable VHD, because I reached the speed limits of reading the HDD and writing to Ram of my PC.

And since we know each PC specs we may check several things like:

1 - Is Rambooting from a LZ4 compressed fixed size VHD a real improvement compared with an expandable VHD or not?

2 - Is the time consumed to build a LZ4 compressed VHD a valid investment or not?

3 - From my tests I know 1.5 GB is good size for PCs with a minimum of 4 GB of Ram, but is there a way to find a size that fits better in accordance with the Ram available on each PC?

4 - Is Rambooting from a LZ4 compressed fixed size VHD better for?:

   - Booting from HDD.

   - Booting from SSD.

5 - Is Rambooting from an expandable VHD better for?:

   - Booting from HDD.

   - Booting from SSD.

 

I really do not care so much about certain difference on other programs that will not load things to Ram since they are on the source.wim wich is outside of the VHD loaded on Ram, but Primocache and PrimoRamdisk have a high influence on loading to Ram times.

And also do not care about total booting to desk time, since it will depend of installed programs and Apps that will load only god knows what to Ram during final stages of boot, (disabling Superfetch may mitigate this).

IMHO only loading to Ram time is valid in order to compare since this excludes almost all software loading times during final stages of boot.

And finally I was joking about your age, JFYI I'm 66.

alacran

[antonino61]

i think I can answer some of ur questions on my side, which obviously differs from urs or anyone's. this is not the mac world with only one machine a year or so.

nino

[antonino61]

ramloading proved to preboot in, u dont really wanna know, 60secs (9.25gig plain vhd), but the booting proper was 3secs faster. now let me wimboot test it and try to answer ur questions.

[antonino61]

so there, these are my sizes and timings now (the smaller, the better).

 

1gb fixed vhd 200mb free space - filedisk - bootmgr booting 17secs - freespace basically staying the same

its 302mb lz4 version expanding the same full size and free space as above - ramdisk - preramloading 3secs + booting 14secs = the same 17secs - freespace basically staying the same

 

the only scenario here is the last I have communicated: no primo ramdisk (windows seems to manage its temps much more efficiently, especially in terms of space), primocache (which still features much higher mass storage read and write speeds than ordinary windows caching). the nvidia files musta gotten embedded in the wim, which allowed me to shrink the vhd to the present extent.

 

now alacran's questions:

 

1 - Is Rambooting from a LZ4 compressed fixed size VHD a real improvement compared with an expandable VHD or not?
definitely so, it is, at least here (4secs faster loading time at least).

2 - Is the time consumed to build a LZ4 compressed VHD a valid investment or not?

it is, at least here, and of course, the smaller the faster.

3 - From my tests I know 1.5 GB is good size for PCs with a minimum of 4 GB of Ram, but is there a way to find a size that fits better in accordance with the Ram available on each PC?
the smallest size possible is the best, which depends on free space size: as long as it stays the same it is not prone to bulge c:\ to red. as long as c:\ stays turquoise, it stays fast.

4 - Is Rambooting from a LZ4 compressed fixed size VHD better for?:
better for everything but u need to resort to the plain source vhd for installing software or modifying settings. when u see the free space shring, it is time to bake another wim off the vhd and apply the wim to send the exeeding stuff on the wim.

   - Booting from HDD

I can't answer that

   - Booting from SSD.
whatever I said above definitely applies to ssd's

5 - Is Rambooting from an expandable VHD better for?:
hell no! no matter what wimb says. the expandibility of the vhd is nothing user-friendly (it does not expand the size of c:\ during operation; it just makes free space doddery, which is a nuisance that often leads to further manual expanding of the vhd); so I would call it user-unfriendly.

   - Booting from HDD.
cannot answer that
   - Booting from SSD.

same as above

 

I really do not care so much about certain difference on other programs that will not load things to Ram since they are on the source.wim wich is outside of the VHD loaded on Ram, but Primocache and PrimoRamdisk have a high influence on loading to Ram times.

I already answered that

And also do not care about total booting to desk time, since it will depend of installed programs and Apps that will load only god knows what to Ram during final stages of boot, (disabling Superfetch may mitigate this).

I do not think so, because, even if it had been true, U have something installed because u need it (unless told that something already existing does the job better --> verify and decide; what I find operationally crucial is vhd size for both preramloading and booting. let me verify the prefetch and superfetch kinda thing, since I have primocache, and I will answer more accurately, but if I remember correctly it might reduce loading time by just 1 second or so.

IMHO only loading to Ram time is valid in order to compare since this excludes almost all software loading times during final stages of boot.

And here is where we will never agree: it is easy to reduce preramloading time, just reduce the size of the thing that wants preramloading, which is the vhd size. 

 

prefetch and superfetch kinda thing --> just verified. it does not reduce loading times, which stays the same as above, but it definitely leaves the \windows\prefetch directory empty, which does not contribute to free space erosion.