Windows XP boot time / hardware

[Guest]

[LeoniAquila]

Based on your CPU, you've surpassed my boot-time in relative terms.  In absolute terms, I beat you by 1 or 2 seconds.

OK, that's fine. The guy I really want to beat is a good friend with a brand new top-of-the-line computer (some dual core thing, 4 GB RAM) but with a standard-bloated system. He also uses XP. I'll visit him in the evening for some goldf on Xbox 360, then I should measure his boot time. It'll be interesting. If he's close to mine or lower, I won't tell him which services to shut down (which he's asked me about).

What affects boot time? My guess would be the CPU and the speed of the HDD, rather than e.g. how much RAM there is. But it's just a guess. Besides, it's:



LA

Thread split after this post.

[Ragwing]

After all optimizing measures, the best boot time I seem to get is 26+5 seconds. Maybe if I cool the machine down a bit it'll perform even better (although it was quite near 20 degree Celsius throughout).

20 degrees Celsius?!



I guess I'll have to wipe away some dust...


Bit off topic but... How can my RAM has been manfactured week 56, 2006, when there's only 52,177457 weeks each year?

BTW now it's also physically clean - wiped it off with a rag. You should all do it! Remember the keyboard hygiene topic?

No need to clean it, it'll get physically unclean pretty fast

What affects boot time? My guess would be the CPU and the speed of the HDD, rather than e.g. how much RAM there is.

If you measure until the desktop is shown, and the cursor isn't a time glass, then RAM will matter. CPU won't. Speed of HDD should only some difference, seeing as it's not an extremely large amount of data that is read and written.
What really matters is drivers and executables (.exe, .dll) loaded on startup...

[LeoniAquila]

20 degrees Celsius?!

Yes, I put the machine on the balcony (13-14 degrees) for half an hour (shut off), then I had it in the room for a while so 20 degrees Celcius is an estimation.

Bit off topic but... How can my RAM has been manfactured week 56, 2006, when there's only 52,177457 weeks each year?

I guess it was made on another planet.

If you measure until the desktop is shown, and the cursor isn't a time glass, then RAM will matter. CPU won't. Speed of HDD should only some difference, seeing as it's not an extremely large amount of data that is read and written.
What really matters is drivers and executables (.exe, .dll) loaded on startup...

26 seconds was until desktop is visbile, the other 5 until cursor animation is finished. Are you sure the amount of RAM matters? I've disabled the page file, so my machine only uses the physical RAM. It's never close to using all of the 1.12 GB I have.

LA

[Soyabeaner]

CPU does matter.  My bro's PC has less RAM than mine but faster processor and his boot time was a bit faster than mine (before all this XP killing/bold fortune/craziness/etc.) and I know he has more junk programs. 

Do you notice real benefits in disabling the page file in the longer sessions of using various programs?

[Japo]

Fragmentation matters a lot here. And of course base hard drive speed.

[LeoniAquila]

I haven't yet used any RAM heavy programs. The only one I really use is Bryce 5.5. I started to make a scene some weeks ago, some day I'll continue and then it'll be important to monitor RAM usage. Don't want the system to crash.

LA

[Ragwing]

Yes, I put the machine on the balcony (13-14 degrees) for half an hour (shut off), then I had it in the room for a while so 20 degrees Celcius is an estimation.

Well, during the hottest days, it's like 35 degrees celsius in my room. But global warming seems to be on vacation or something...

Are you sure the amount of RAM matters?

Well, it does, if your computer doesn't have enough. If not, it'll use the pagefile, which is like 1,000 times slower than RAM...
I noticed a big difference from when I had 256 MB, and added 512 MB

I haven't yet used any RAM heavy programs. The only one I really use is Bryce 5.5. I started to make a scene some weeks ago, some day I'll continue and then it'll be important to monitor RAM usage. Don't want the system to crash.

That's why you have a pagefile...

Fragmentation matters a lot here. And of course base hard drive speed.

Yes, we forgot that. If you don't defragment your HDD for a while, you'll notice a difference if you do. It's like a library (defragmented) and a pile of books (fragmented). Which one is best if you want to find something fast? The library. That's why you defragment your HDD.

[Japo]

The hard drive is specially important when booting, because at first that's where everything is, threre's nothing in memory unless it's loaded from a disc first.

Besides the general level of fragmentation, it's specially important to keep it down for the OS files that must be loaded at startup.

PS: PageDefrag reported that all my system files were perfectly defragmented, and when they weren't a single sweep of JkDefrag did the trick every time. But recently I added 1 GB of RAM (for no special reason since I had 1 GB and didn't need over half of it), and now my \hiberfil.sys is broken in two fragments 'cause it's too damn big to be defragmented any further.

PPS: Pls don't suggest me to "tweak" Windows' memory management 'cuz I don't wanna. And don't worry about me, the wasn't for real.

 

[LeoniAquila]

Well, it does, if your computer doesn't have enough. If not, it'll use the pagefile, which is like 1,000 times slower than RAM...
I noticed a big difference from when I had 256 MB, and added 512 MB

Then it shouldn't really matter if the amount of RAM is enough and the page file is disabled. I use only a fraction of my 1.12 GB RAM.

LA

[Soyabeaner]

Any facts here?  I really want to know 2 things:
1. Any performance loss/gain on disabling the page file for 1 giggers+ of RAM with little usage like in LA's case?
2. Will using more physical RAM degrade it faster over time as it does for hard disk with frequent defragmentation?

[Japo]

All this is a misconception. Virtual memory is not hard disc space used as memory. It's a virtual space described by adresses, neither in RAM nor in the hard discs, but virtual. Windows assigns pages of virtual memory to every program. If there's enough physical RAM (as it should in most situations), the virtual memory will be assigned by Windows to physical RAM. Only if there's no physical RAM left, will Windows assign virtual memory addresses to discs (pagefiles).

If you're not short on physical RAM, hard discs won't be used as memory, no matter how big your pagefile is.

Alternatively, if you were short on RAM, there would of course no way you could make the programs use physical RAM that's already full, no matter your settings. Then it's when the pagefile (which is simply a reserved space) must be big enough to fit everything that doesn't fit in RAM. If it wasn't big enough and Windows is set to manage virtual memory automatically, Windows will try to extend the pagefile--I've seen this happen at work. If it wasn't big enough and you prevent Windows from extending it, I bet you'll have your Windows crashed catastrophically.

There's no way you can make Windows (3.x and  higher) use physical RAM directly. That belongs to the real mode of DOS days which I'm unsure modern processors will ever support, and believe me it sucked. Big. Time. Windows programs don't know how to interface with RAM to start with, they just ask Windows for memory, virtual memory that is. If the physical RAM isn't exhausted, they'll be loaded there, no matter your settings.

People ***** a lot about Windows' virtual memory without knowing what it is because of Microsoft's Black Legend. They hear and believe that MS makes their computers work slower just for the heck of it. But it so happens that Linux uses the same device--of course, it would be a worthless OS if it addressed to real memory--, and what's more it traditionally uses a whole partition instead of a pagefile, so it's more difficult for the user to tweak it.

The ONLY effect of lowering the size of the pagefile is saving hard disc space, and risking a crash if you ever use a program that would not fit in your RAM plus pagefile along with the rest of the load. There is NO performance effect of tweaking the pagefile, period.

Since I'm no expert, please correct me where I'm wrong. Here's some reference:

http://members.shaw.ca/bsanders/WindowsGeneralWeb/RAMVirtualMemoryPageFileEtc.htm

[LeoniAquila]

I hope you are right Japo (what you write does make sense to me). Using physical RAM only as far as possible would be the smartest thing, and then have a pagefile in case the need of memory exceeds the amount of physical RAM.

As it seems, my only performance benefit from disabling the pagefile, is that Windows doesn't have to create the file on every boot up. Maybe it's neglectable though.

LA

[Japo]

I don't think the pagefile is re-created at every boot, wouldn't make sense to me. I guess it sits there and is written only when needed and resized only when needed. It's only reserved space in case. If you want to be sure you could boot a second OS and look if pagefile.sys is there without Windows booted.

My pagefile is 1'5 GB and I very seldom see the hard drive usage lamp flashing during normal operation.

[Ragwing]

As it seems, my only performance benefit from disabling the pagefile, is that Windows doesn't have to create the file on every boot up. Maybe it's neglectable though.

Is it created every boot? I thought it always was there... Why would it have to be re-created each time you boot?
I'm not sure, but if it's not used, I think it just contain some garbage that fills the pagefile (so it can reserve the space)

If it wasn't big enough and you prevent Windows from extending it, I bet you'll have your Windows crashed catastrophically.

I think this only will happen if you use up all your RAM, and also a 4096 MB (max size) pagefile. I'm not sure what would happen if this occured, but I think you'll just get some error message...

Anyways, I disabled the pagefile and spawned several cmd.exe to flood the RAM, and I got this...


Fast translation: Could not initialize the program correct (stupid error code). Press OK to terminate the program.

[Japo]

Well at least Windows is clever and refuses to start a new process instead of crashing all. But what if the shortage is caused because of an already existing process that raises its memory demands?

I think this only will happen if you use up all your RAM, and also a 4096 MB (max size) pagefile. I'm not sure what would happen if this occured, but I think you'll just get some error message...

I think you're still getting it wrong, what is 4 GB (in 32 bits Windows) is the virtual memory space (that is memory addresses are 32-bits words, 2^32=4x1024^3), not the pagefile.

(Virtual memory equals NOT pagefile!)

So in any case a 32-bits Windows won't be able to make available more than 4 GB of memory--irrespective of how much of that can be placed in physical RAM and how much has to be placed on drives. That means that if you had 4 GB or more of physical RAM and a 32-bits Windows, memory would run out before the pagefile started to be noticeably used. And the system wouldn't be able to make more than 4 GB of memory available (irrespective of how much physical RAM you have) even if you manually made the pagefile larger than 4 GB.

[Tags:]