|
You shouldn't disable the pagefile.
Everything in the kernel expects it to exist and has been built around that assumption. so if you remove it you will actually get worse performance. Even if your RAM is more than enough.
You can probably make it a lot smaller though.
If you go to the control panel -> System -> advanced system settings -> Select the advanced tab -> Performance settings -> Advanced tab -> click Virtual memory - change, you'll get a recommended size that will quite probably work just fine for you.
|
|
|
|
|
|
Jörgen Andersson wrote: You shouldn't disable the pagefile.
Everything in the kernel expects it to exist and has been built around that assumption. so if you remove it you will actually get worse performance.
I am struggling to understand which kind of OS kernel functions are speeded up by the presence of a pagefile. It certainly isn't obvious to me.
Could you provide a couple of examples (including an explanation of how the presence of a page file speeds it up), or a link to some source that you base your statement on?
|
|
|
|
|
I can't find it now, but I read it in a blog on the Microsoft homepage some years ago.
But as I understood it: The pagefile is built in at a so low level and is so optimized, so that when they added the possibility to disable it they didn't actually remove or bypass any code in the kernel.
All code that has to do with the pagefile is still used whether the file exists or not. And he claimed that it performance wise is better to have a small pagefile instead of none, because of how this is done.
|
|
|
|
|
|
A pagefile wouldn't solve the problem of some process going berserk - it would just take a little longer.
I love automatic garbage collection. For a number of years, I worked with C/C++ and its explicit malloc/free: For every single piece of C/C++ software I have worked on, a walkthrough of the memory use has revealed leaks. You fix them all before the release, but for the next release, new leaks has in mysterious ways crept in. Always. Then you become too eager, and free() memory that is still referenced by a pointer that you overlooked... If I should point out one single benefit from switching from C/C++ to C#, the automatic garbage collection is the obvious winner.
If you truly need more virtual space than what is physically available, then you must of course have a pagefile to support it. For my home machine, I have never seen more than slightly over half of my physical RAM being in use for anything, even when I deliberately tried to start many "large" applications at the same time. I guess part of the reason is that none of my applications have excessively large data structures - executable code doesn't require pagefile space (except possibly while being debugged), and code pages that you don't reference are not brought into RAM at all, no matter how large the executable is.
My funniest out-of-memory story happened back in the DOS days, not to me but to one of my collagues: This dBase-application refused to start; we didn't have a clue about why. After a long search for software causes, the PC was wheeled down to the hardware guys to verify that the hardware was OK. Down there, they plugged in probes and meters, and the application started with no problems. The PC was returned to my colleague, application woudln't start. Back to the lab: The application started.
After several trips back and forth, it was realized that this application was so demanding that you had to keep the cover off the PC to give it enough space. Many old tower cabinets had covers that you had to slide backwards: This was deliberate, to make sure that you disconnected all cables (or: the power cable) before digging into the electronics. The HW guys of course had to plug the power cable back in to do the testing.
But they did not plug in the mouse: The dBase application input was keyboard-only. So at boot-up, the mouse driver looked around for a mouse, found none, and terminated, freeing up a couple hundred bytes. When the PC was returnd to be installed in the office, the mouse was plugged in (we had started using Windows 2.11), the mouse driver found hardware, and clinged to its bytes of RAM. This was enough to make the total RAM requirements exceed the 512 kbyte available. (But 640 kbyte would have been sufficient for anybody...)
|
|
|
|
|
Member 7989122 wrote: A pagefile wouldn't solve the problem of some process going berserk - it would just take a little longer.
No, but what it does do is to replace catastrophic random failure with graceful degradation. If you system gradually starts running slower than normal looking at task manager and seeing 25 of 16GB ram used will make the problem clear, and then you can go to the per app tab and see what the pig is. Failure to allocate ram will cause whatever program attempts to make the first allocation past the system being maxed out fail. In cases where the primary offenders ram footprint is slowly growing due to heap fragmentation will frequently be some other random application. The case I mentioned the first time around took so long to track down because it was completely random which of the half dozen applications we were running would terminate. As it was it only was resolved as fast as it was because I noticed there wasn't a pagefile. The person who disabled it had no idea it would cause problems and was busy chasing the same red herrings that my coworker and I were in trying to figure out why the new machines were unstable.
Your DOS story brings me back. I was only a kid at the time; but also the resident PC expert at home. Dealing with DOS games with different tight memory limits resulted in the PC having a half dozenish boot menu options that controlled what type of additional memory was enabled (XMS, EMS, both, or neither) along with if all the normal TSRs were loaded or just a subset to free more <640k ram. I knew what was what since I set it all up, my younger brother had a cheatsheet for when a game he wanted to play needed a non-default setting, everyone else in the house either just used the default or picked the direct to win3.1 option.
Did you ever see history portrayed as an old man with a wise brow and pulseless heart, weighing all things in the balance of reason?
Is not rather the genius of history like an eternal, imploring maiden, full of fire, with a burning heart and flaming soul, humanly warm and humanly beautiful?
--Zachris Topelius
Training a telescope on one’s own belly button will only reveal lint. You like that? You go right on staring at it. I prefer looking at galaxies.
-- Sarah Hoyt
|
|
|
|
|
There was another memory leak story in The Old New Thing[^] blog: This web server that gradually had its memory filled up due to a leak. The server had to be up 24/7 without interruption, so while they were searching for a solution, they replaced the one server with a cluster and a load balancer. When any of the machines were reaching a critical limit, it was taken out of the cluster and rebooted, and then put back in. Once the leak was found, they went back to a single server setup.
(I read the story in the book that presents a long series of the blog posts. Some of them are really entertaining. They are all available on line, but this is several years back and I won't spend the time to search up this one post. Both the book and the blog is certainly recommended!)
|
|
|
|
|
Wouldn't that be the link to the blog of Mark Russinovich that you also posted?
He writes "Windows can write pages on the modified list (which represent pages that aren’t being accessed actively but have not been saved to disk) out to the paging file, thus making that memory available for more useful purposes (processes or file cache). So while there may be some workloads that perform better with no paging file, in general having one will mean more usable memory being available to the system (never mind that Windows won’t be able to write kernel crash dumps without a paging file sized large enough to hold them)."
This comes after at throrough discussion of how to determine the commit charge of your biggest workload. The size of available RAM plus the size of the pagefile should be at least the size of the commit charge. Fair enough: If you really need the space, then a page file is infinitely better thatn not having that virtual space. He refers to more useful purposes - assuming that there are more useful purposes.
Also, the way you describe it, it sounds like some work is done whether there is a pagefile or not. If MS had bothered to skip over the code when not needed, it could have been even faster with no pagefile. So there is a slowdown, compared to the ideal. When there is a pagefile present, the slowdown is required; slowdown isn't reduced.
So I suspect that you, over the years since you read it, your memory of what he said has become a little blurred. He did not make it as an unconditional blanket statement, but explained in which situations a pagefile is good to have.
The situation initially described in this thread was that your machine has RAM in such amounts that all "useful purposes" finds space in RAM. You have no need to send pages out to the pagefile to make more memory available. Why would it then speed up things, when you never invoke the function?
Now this Russionvich blog is ten years old. I wouldn't be surprised if ten years is enough to make speed optimizations so that executing paging code is skipped when there is no paging to do. I've got the latest Windows Internals at home - Russinovich is no longer stated as the (primary) author of the seventh edition (he was for the previous six), but it goes even deeper into internal workings than his old blog. I will see if I can find any explanation there for how the availability of a feature that is not used can speed up a program - if that really is the case.
|
|
|
|
|
|
Interesting. Thanks! 
|
|
|
|
|
The latest version of Win10 Pro (1909) includes Windows Sandbox[^] to test dodgy software (or stuff written by QA questioners where the last thing it's ever going to do is what the author intended).
I mentioned it to Herself, and her response was "So ... perverts can watch what they like without leaving any traces?"
Hmm. Hadn't thought of that one.
"I have no idea what I did, but I'm taking full credit for it." - ThisOldTony
AntiTwitter: @DalekDave is now a follower!
|
|
|
|
|
|
|
Pah, I already used ever since last 2 weeks! :P
|
|
|
|
|
No we know why you've been missing and what you've been up to... 
#SupportHeForShe
Government can give you nothing but what it takes from somebody else. A government big enough to give you everything you want is big enough to take everything you've got, including your freedom.-Ezra Taft Benson
You must accept 1 of 2 basic premises: Either we are alone in the universe or we are not alone. Either way, the implications are staggering!-Wernher von Braun
|
|
|
|
|
|
I've got a feeling the garbage bits are still around on the drive. FBI sniffers will find it.
#SupportHeForShe
Government can give you nothing but what it takes from somebody else. A government big enough to give you everything you want is big enough to take everything you've got, including your freedom.-Ezra Taft Benson
You must accept 1 of 2 basic premises: Either we are alone in the universe or we are not alone. Either way, the implications are staggering!-Wernher von Braun
|
|
|
|
|
OriginalGriff wrote: I mentioned it to Herself, and her response was "So ... perverts can watch what they like without leaving any traces?"
Hmm. Hadn't thought of that one.
Why else would she be thinking that unless she's saying she knows you're "leaving a trace"...?
|
|
|
|
|
Interestingly the 'Virus & Threat protection' was off after the restart of 1909...
Does it mean that I need no protection with all those sandbags?
"The only place where Success comes before Work is in the dictionary." Vidal Sassoon, 1928 - 2012
|
|
|
|
|
After you displace sandbags protection becomes redundant, it's better to focus on mortar and machineguns positions.
GCS d--(d+) s-/++ a C++++ U+++ P- L+@ E-- W++ N+ o+ K- w+++ O? M-- V? PS+ PE- Y+ PGP t+ 5? X R+++ tv-- b+(+++) DI+++ D++ G e++ h--- r+++ y+++* Weapons extension: ma- k++ F+2 X
|
|
|
|
|
As a pervert, I can watch whatever I want.
I don't owe anyone an explanation of my viewing habits 
How perverted am I?
Just last week I read some stuff about front-end frameworks! 
So sue me 
|
|
|
|
|
Sander Rossel wrote: How perverted am I?
Just last week I read some stuff about front-end frameworks!
There is the exit; please use it.
Your hat and coat have already been deposited in the street.
Freedom is the freedom to say that two plus two make four. If that is granted, all else follows.
-- 6079 Smith W.
|
|
|
|
|
Front-end frameworks from Victoria's Secret?
|
|
|
|
|
Adaptive AutoSAR is much worse than that!
GCS d--(d+) s-/++ a C++++ U+++ P- L+@ E-- W++ N+ o+ K- w+++ O? M-- V? PS+ PE- Y+ PGP t+ 5? X R+++ tv-- b+(+++) DI+++ D++ G e++ h--- r+++ y+++* Weapons extension: ma- k++ F+2 X
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
![Elephant | [mastadon]](https://codeproject.global.ssl.fastly.net/script/Forums/Images/elephant.gif)
was wrong with our new systems because the affected applications just terminated without displaying useful error messages. 
