Another thread about RAM, only this time --> RAM usage under Win 7, pagefile, utilization, etc.

No, your pals generally had positive reinforcement for doing the correct thing.

 

You guys miss something here. Not so much ram usage as it so fast movement for ram clearing etc happens rappidly. The problem is using the HDD or SSD. When you do this you flood the pci bus. Any system with this bus flooded with data or wating for data etc will slow the system down immensly.

If you double dip into the bus by utilizing the page file you slow this down further and by alowing an unlimited page file and dynamically setting it you can slow this access even further. This is the reason MS treis to utilize it as little as possible.

Under most normal usage we see little to no page file usage access. It still though is a legacy required need for the OS unde generic usage. This is why I never recommend to users to disable it as ti truely is an OS requirement. For those in the KNOW that can live without it and REALLY know what they are doing, then removing it is fine. Most proclaimed experts are not that knowledgable even though they think they are.

It should NEVER be recomended to another user to remove the page file. It has to be an individual informed decision. Even the most highly educated expert in all of us out there can even be wrong. This is almost always true especially in this, as well we need to be wrong on occasion for once we are never wrong we will stop learning. So in this instance where you could be wrong for yourself do not assume you are being correct for someone else. Let the individual experiment and decide for themselves only IF THEY WANT TOO and as always YMMV..................................

 

Don't know about the rest, but I certainly did not get your point...

 

I know what TANWare is saying.

First rule of 'tweaking': do no harm.

If one person can get worse results by anything you suggest (eg. disable pagefile), then you shouldn't suggest it. You can suggest they experiment with it, but as pointed out... when the O/S requires the pagefile to function properly under many/varied conditions, then suggesting to disable this essential file is definitely in the realm of 'doing harm'.

As I mentioned in this thread earlier, I disabled the pagefile with 8GB RAM and in 20 minutes I was getting unacceptable errors in 'light' usage (light for me).

See:
http://forum.notebookreview.com/7115082-post35.html


I even tried to replicate your issues while copying large files - with all my pagefiles (almost 32GB worth across 4 partitions) enabled - and I did not see any sign of problems except the expected slowdown of the system (which is why I suggested you look elsewhere for your issues, like a clean install of your O/S).

Also, TANWare's suggestion that it is simply the bus that is getting saturated does warrant a closer look. I don't have a good idea right now on how to test that though.

 

I understood that, but:

1. No other way to know what's better, than to try each possibility, especially when, as in this case, everything is 100% reversible. So I would always advise people to try out and figure out what works best for them, especially when there is no risk of permanent damage as in the current situation.

2. TANWare's explanation does not explain why there seems to be a difference between 32 bit and 64 bit windows. Whatever the reason is, I cannot accept this behavior being normal! I tried to explain it as detailed as I could already, but imagine what it is not to be able to continue browsing or generally working on a PC with up-to-date configuration when you are installing a program or copying a folder! This is ridiculous...!

 

disabling or enabling something would have different effect or else it is a useless excercise.

it is a matter of choice.

for some, OOM is the most scary thing on the planet(many of my users prefer single sentence NPE over java/c# stack trace). for others, predictable responsive is more important. I am not the one saying one is more important than the other.

However, don't misinform if you want people to make informed decision. page file is not a requirement of the OS. it is a feature of the OS. Microsoft's particular implementation of autogrowing is an additional feature. linux doesn't have it as far as i know. and all live-cd/usb boot would not have a page file.

for 99% of today's Windows user, enable or disable it makes no noticeable difference. So if I am an IT support guy handling 200 users, I would be out of my mind for recommending it to people using a machine that I am responsible to support. Because the worst case for enabling is only they just sit there and wait and complain which may end up asking the boss to buy something faster, not my issue.

@tilleroftheearth
if you are really eager to try, find a machine with 2/4G RAM fresh install(single partition) with Windows managed page file then repeat what gracy123 did.

 

Gracy123,

1) True; experimentation on our own systems will give each of us 100% reliable results.

2) As I've tried to say a few times... the behavior you witnessed on your system is not normal (with over 200 Win7 installations I've played with - most of them 64bit).

You may want to try a new RST driver (experimentation! ) to see if things are smoother for you:

See:
http://forum.notebookreview.com/7136575-post8542.html


I have been running it the last couple of days with no issues and even some (slight) improvements.

Also, the huge differences I was noticing with ThrottleStop (and 'MAX' C-States enabled) is much less noticeable with this new IRST driver (and I've now stopped playing around with TS).

From afar, it sounds like NCQ is not enabled on your system - and that could be a driver issue (or not!). The fact that 2GB extra RAM effectively hid that problem also points to me that there is software running (Sony stuff?) that is slowing your system down even if it is responding normally now (with the RAM and pagefile disabled).

 

First, that would not compare with Gracy123's 'cleaned-up' installation (she does not have a clean install).

Second, the closest I got to your suggestion was here:

See:
http://forum.notebookreview.com/7114375-post29.html


I did not try copying any large files - didn't have to. The system performed like it was on a 20MB HDD from 1980 or so.

I'm happy that the issue seems to have been solved for Gracy123; I just feel that it is being masked though (have not seen those symptoms as described on any recent system).

 

Oh thanks for pointing to the above, just installed it, the RST monitor still has 10.1.0.1008 (I guess the installation does not update it?), but the driver version in Device Manager has effectively changed to 10.1.2.1004.

I'm quite skeptic though, experimenting with RST drivers was one of the first ideas I had to resolve the problem - tried numerous versions (even older than the one I had at the time) - no noticeable difference in any direction :-/ But I like being up-to-date so thanks once again for bringing my attention to this one

There is pretty much no (useless) Sony software running. I could give you guys a list of all running applications if you want to have a look at them, but except the really necessary Sony stuff, I only still have Vaio Gate, Vaio Update and Vaio Care still installed. Apart from Vaio Update which I can't say I really use, the other 2 are really useful and according to my research - not taking up any resources.

 

That is what I meant 'predictable responsiveness'.

BTW, your 7k500 is quite possible to be amplifying the problem.

 

I doubt that. I cloned my entire system to the original drive - Toshiba 5400 - no difference. (You can see how many hours I spent looking for the reason )

 

1.) you agree each has to decide for themselves on the tweakability and usefullness of the tweak itself to the user. That is a good thing, people need to be aware though the fact by doing this they are running the OS in a non standard configuration and the possible effects.

2.) It was not meant as a post to determine difference between 32 vs 64 bit. Other than you are tying up the bus. On that note one possible explination is that when using the page file, just as in ram memory, it is addressing it as 64 bit not 32 bit. even though this is virtual address space it is still address space subject to the 64 bit address space criteria................

Edit;

3.) The 7K500 is an excelent drive and should not contribute to any HDD issue. Of the HDD's the 7K500 should be one of the least problematic................

4.) if on the drives you have write caching enabled when the device is not free it will cache to ram or even virtual memory. This can cause a cascading degradation of performance if ram is filled and it can't write to the device because it is trying to write to the virtual address space. Again just another possibility.........

 

When I said that, I was thinking about Scorpio black in my mind which is shown by test that can handle random access better.

If you are still interesting in testing, one thing I can think of is to set a very large page file say 32G and see if there is any difference between this and the system managed setting.

 

To be honest, I am not sure about that... I spent quite a lot of time reading benchmarks and forums about 7200 RPM HDDs, never read a single post about Scorpio being faster/better performance-wise than 7K500 ...
7K500 is considered overall the fastest 7200RPM hdd. I'm not saying you are wrong, just that I haven't seen a confirmation about it, rather the opposite.

But the fact that the 5400 Toshiba didn't make things worse either makes me think it is not the HDD itself... Maybe the controller, but I know no way to "benchmark" it or diagnose it...

 

P.s. If anyone with Scorpio would want to run and upload some benchmarking results (screenshots), I'd be willing to run the same tests and compare... just out of curiosity.

I'm sure every parameter is comparable using the right benchmarking software and test...

 

That actually is the thing we are interested to find out whether it is this cascading self feedback loop further tax an already over taxed HDD. And disable page file is one way to stop the loop. setting a fixed size page file is another. Both seems to have effect on her machine.

 

Actually only if the size is small, in other words if the size is manually decreased. Manually setting it to say 4 or 6GB did not make a difference as far as I remember...

 

This is defiantely x64 causing the issue. I run 8GB and no page file but when I copied a 33GB picture folder in my data drive HDD I could see all of the memory go to stand-by and free go to 0. if I had the page file enabled I am quite sure it would have been highly utilized.............

 

That again is something we want to find out. Whether the copy process distinguish between real memory vs pseudo memory. Using up all the free ram is fine, crowding out others is not. What she has experienced is a crowding out effect.

 

Ok, so on the D: drive I enabled a system managed page file and repeated the test. While I got the hard faults just as she did I did not have a slow system. Yes it seemed a bit slower in opening minimized windows of IE etc it was just a slight slow down. I should note it did not seem to have all the hard fault spikes without the page file.

My D: drive is a XT 500 GB if it matters at all. So yes the page file seemed to be hit and the extra bus saturation made the system ever so slightly sluggish but no major effect. I can tell you there was no sluggishness without the page file. Without the page file you would never even know the D: drive was saturated while opening, restoring or swithching between the programs etc.

 

that was my experience too which I have mentioned before. I can notice the difference but not to the point of annoying. However, I know every well that I would not hit OOM and I don't care if it does, so I turn the page file off.

 

Been a few weeks since I disabled the pagefile.
Well I just had my first OOM recently (screenshor taken shortly after, when Firefox was automatically closed to free up RAM):



It was caused by actively using Photoshop (had over 15 pictures opened at once and working on)

Having in mind that this is not something I do every day, I'm still keeping my pagefile disabled and enjoying better performance this way

 

Gracy123,

Having mentioned you use PS, I knew this would happen. Hope you didn't lose too much work?

I don't understand your preference to keeping the pagefile disabled though?

Oh well!

I can't afford to lose any work/time with my photo-editing jobs: setting the pagefile on all partitions has given me a very responsive and extremely stable platform that is just as fast (except for booting/shutdown) as when no pagefile is set at all. But I do have 8GB RAM though.

Thanks for the update (and your honesty, too!).

 

Actually I did not loose anything - nothing crashes - just a message that the memory is low occurs and that I should save my work immediately and close some programs. As I failed to do so and continued working with PS, Firefox was closed automatically. On reopen it restores all windows and tabs. So not really a big deal and really the first time this happens since I disabled it.

I determined my "typical" RAM usage when not using PS:

~ 2000 MB - In Use
~ 150-500 MB - Modified
~ 1500-2000 MB Standby
the rest ~ 1000-1500 MB - Free

So I definitely think 6GB are perfect for my needs, 8GB would be just luxury. I might upgrade further one day though, just not now.

Why I stand by the disabled pagefile - it really makes noticeable positive difference to have it this way! I guess it is individual more or less, but my system definitely, without any doubt, is more responsive this way...

However I am really thankful to you for all efforts and time spent to try to help me and all valuable ideas you gave!! Wish there were more people like you around!

 

Thanks for the kind words.

Nice to see Win7's 'graceful' way of handling it (with FF's help of course).

Those numbers are very close to my 'idle' numbers too except for ~3GB free (with 8GB RAM), so I guess you've essentially optimized matching the hardware to your needs.

Good call on anything more being a 'luxury'.

 

I had similar experiences and opted for the same solution. OOM, sure I have seen it a few times for the 3+ years but each time, there was first a warning and if I choose, I would just close out some IE tasks and it would be fine.

No way I would allow the page files to drag my feet for the rest 99.9% of the time.

 

SO. Gracy123, you had enabled paging file, it was 400 MB and your system was slower than with disabled one, am I right? This thread is huge for reading all that tell all in short please. What you had (If I misunderstood), what you did and what you have now? And what is your usual work.

I have 5400 rpm HDD and thinking of changing paging file. Have 6GB of RAM but use RAMDisk and eBoostr (minus 1.5GB together). Paging file is 3GB (yes, I know it is huge). When I was working on XP I was setting it to 400-500 MB. But this is Windows 7 so I hoped that Windows works with paging file better allowing RAM work if it is free (which as I think was happening with you in the beginning of thread).

 

Well to sum up (but there are a lot of details and examples you need to read the whole thread for...):

Speedwise:

System managed pagefile < Limited size pagefile < Disabled pagefile
(slow < faster < the fastest (most responsive))

Disabling pagefile however was only possible after upgrading to 6GB RAM - 4GB is simply not enough.

My usage pattern - a lot of multitasking (5-15 browser windows, each with 2-10 tabs) + Picasa + Photoshop + Skype + other small programs and tools.

 

Sorry, I'm jumping late, but here are some questions for Gracey123 and others who tested this:

a) How did you test the copy? Did everyone use Windows Explorer? What happens if you use the Command Prompt "copy" or "robocopy" commands? Any difference in performance? What about something like Teracopy for the large files?

b) Just to double check everyone else, this is Win 7 and not Vista correct? Is SP1 is installed?

I remember when Vista was released, Windows took a big performance hit on file copy - See Inside Vista SP1 File Copy Improvements - Mark's Blog - Site Home - TechNet Blogs for more details. Is it possible this fix didn't carry over in Windows 7 x64? Probably not.

c) Do these hardware specs apply to anyone?

- Slow file copy - Page 2

d) Anyone have Exchange to test ESEUTIL? See - Slow Large File Copy Issues - Ask the Performance Team - Site Home - TechNet Blogs


It could be that Windows is merely using up all RAM to cache these large files. The contention for HDD resource to read the files and cache the data, in addition to paging virtual memory could explain the issues Gracy first encountered.

If that indeed is the problem then a copy utility which does NOT cache the file data would be a possible solution to this problem.

 

That is exactly the problem and disable page file stops it from further banging the slowest part of the system(i.e. HDD contention).

Though I am not sure using a copy utility is the solution as it is natural for people to just use the explorer. Beside, there is no down side of disablng page file(at least in her case).

With today's RAM price, just give your computer what it wants and stop using page file as RAM(except for situations where programs refuse to run without page file).

 

Hi

Straight to your questions:

Yes, I use Windows explorer only. I never tried using any other methods - I just want to work - easy, fast and efficiently For the time I go to command prompt and type commands every time I want to copy a directory, I could just sit and observe the non-responsiveness of the computer - it will take just as long

Yes, Win 7 Pro x64. At the time of the problems, SP1 was not released and did not officially exist. I haven't done any testing since SP1 is installed.

Nope - VAIO here Intel HM55 MB.

As chimpanzee already stated above, this is exactly what the problem seems to be - copying big amounts of data uses too much RAM, forcing programs cache to be moved elsewhere - the pagefile. At the same time the HDD is overloaded with the copying process and it all spins around in a circle, making the system unusable until the copying process finishes.

Disabling the pagefile (possible only when enough RAM is present and 4GB was definitely not enough in my case) solved the issue, although there was a very noticeable difference coming from the additional RAM only as well, when I upgraded to 6GB.

 

I never use system managed pagefile. Last question (perhaps) Is limited size of 400-500 MB significantly better than 3000MB for example or another one bigger than 1GB?

 

Ask as much as you want, no worries

In my experience and according to my testing results and evaluation - yes - the smaller the pagefile - the better the responsiveness (again - as long as enough RAM is present so that you don't hit OOM). "Significantly" is of course very subjective and of course also depending on the other components and especially the HDD. But to me (with a fast 7200 RPM HDD!) - yes, I would say there was a quite noticeable difference... It should be less noticeable with SSD and even more noticeable with 5400.

Try it out for a few days and see how it works for you

 

Even if a SSD it should be quite noticeable. RAM is far faster than any HDD. I remember Seagate or Samsung (I forget which) did a 24RAID0 drive array of SSD's and they hit ~2GB/s... 1066mhz RAM is ~17GB/s.

 

Thanks for posting back. As mentioned earlier in this thread, disabling use of the swap file should only be done by someone who knows their system's resources (like memory), and understands the risks involved. I don't recall reading anyone pointing out the copy performance problem was due to caching a large disk copy and thrashing of apps within the virtual memory manager. Hopeful this this performance problem is clear for other readers who may come across this thread. Thanks, again.

In regards to the chimpanzee's post, I was wondering if the cache problem was present in Windows Explorer, but not found in other "default" and "third-party" Win 7 utilities used to copy files.

 

Based on the link you mentioned, the explorer opts to do it that way(in fact it gave you a history of how it evolves). I am not sure what other default or third party thing you mean as anything that is not 'explorer' are special purpose.

For the people who want the page file enabled by default(general users on the street like mom and pop), they are also the people who only know explorer and would be scared by the black command window. Though they are also the people who would just sit there and wait(and may be curse and think about switching to Mac)

BTW, I did mention that it was the crowding out effect some where in the piles of the post and why I suggested to disable page file in the first place.

 

You could make a data collector monitor of paging file usage while copying. Just open system monitor, create data collector group, choose paging file to monitor, start it and then start copy. After that stop monitor and look how are the numbers. When I played Crysis 2 My paging file usage was almost 3% of 3GB

 

complex games may be a legitimate reason to enable page file. These games have to do lots of complex calculation and memoization them in memory is faster than recalc. And instead of using their own file based memoization backing, they just ask for memory from the OS(and they are smart enough to know how much to use).

In such a situation, a big page file means they can memorized more calculation and can have an effect on frame rate.

 

While this is without a doubt true, I can't get tired of writing that trying is the only way to find out and there are pretty much no risks involved. Worst case scenario - the system will hit OOM (Out Of Memory) and a pop up message will appear that you need to save your work and close some applications as the system is short on RAM (has happened to me couple of times when the system was under a lot of load). If this happens during normal/typical workload - it means you have insufficient RAM installed and your system performance can benefit from more (with and without a pagefile!).

Disabling the pagefile requires a restart, whereas enabling it back doesn't.

That's all about the risks involved.

This said, I absolutely don't understand people like a few in this forum, who spend hours trying to convince that turning the pagefile off is wrong based on theory, but are afraid to just disable it and see for themselves...

I am sure there are situations where it better be on, but again - only one way to find out...!
It worked great for me and would I have known if I didn't try it out?

 

Well I never turned it off because if your system will crach or BSOD or else, the paging file is the place where debug information will be saved.

 

True. And how often does your system "crach or BSOD or else"?
And how much are you able to do about it yourself from the debug information stored?

If it happens often or actually at all, you have a far more serious problem, than performance throttling down due to the pagefile ))
It's like saying "I don't drive faster than 120 km/h as otherwise if any of my wheels detaches, I won't be able to see where it goes" ....

But if you are a developer or change hardware or even software on daily basis it makes sense...

 

Full core dump is only useful for Microsoft, not end users as without the debugging information, they are useless.

If you want to know what driver caused the core dump, you can see it on screen(by not letting it to restart on crash which make sense for laptop).

If you are managing servers, not monitoring actual memory usage and depends on 'unlimited growth' of page file means you have failed the job.

 

I took this advice months ago and slapped 8G in my DV7T. I haven't had so much as a hint of a problem running with no page file.

 

For your system and the apps it uses, possibly. But for others, that may not always be the case.

For example, more and more apps are using things like SQL Server Express and other embedded database technologies for local storage. I can personally testify databases running into a large amount of "out of memory" errors can trash parts of that database, causing data loss. Not a great situation, unless you're vigilant about backups.

Again, I'm not advocating for or against, as I can see both sides. I just want to make sure everyone is educated, understands what is happening as well as any risks about their systems or the data in those systems.

 

"copy" and "robocopy" are available w/ out installing any other software, and I don't really care if this is Explorer or special purpose to do the copy. What I was looking for was if anyone tried other methods available within Windows to see if they encountered the same thrashing problem.

From what I can gather from other posts is it was only tested using Explorer. Thanks everyone for your feedback.

 

And you are right doing so. But once again - only one way to find out...
Every single system tuning takes certain risks.

There is a saying where I come from - "Don't go in the woods if you are afraid of bears"...

 

Agreed. "Knowing is half the battle."

 

Are you sure? I mean doesn't this information help somehow to recover a system at least in some scenarios? I guess I need to start a new thread about this to ask.

Yes. Because of eBoostr I got many different crashes for last 15 days. Every next boot i saw a window which gave me info including path of files where I can find detailed information about crash. Did debug information provided that? I don't know...

Just for a note, I am not proving or urging about smth because I do not have my general opinion. I only have my personal one about little limited PF for being safety which I share with that people who already (or still) use big paging files.

Very strange saying. What, you live in country where bears are everywhere? What, is it Russia?
Just kidding. But not many EU countries have bears in woods (at least I thought so). Is it somewhere like Slovenia?
We are saying "Being afraid of wolves, can't go into forest".

I will continue checking my usage. For now pagefile.sys is used on 4% = 120MB.

 

I don't know if you've seen this before, but even with page filing 'disabled', hdd space is used up; on my computer when I have 8GB ram installed, the system 'reserves' 8GB of HDD space (10.4GB free HDD space); and if I remove one of the 4GB sticks then the disc 'reserve' goes down by 4GB (14.4GB free HDD space)...

I'm pretty sure this is related to page filing, I have page filing set to 'no page file' under Virtual Memory; but I'm not sure if there is some other option I should set.



BTW, on my laptop I disable page filing because the 1.8" HDD I have is just so SLOOOOW. and I have not run into issues other than what I mentioned above; so it is staying disabled.

 

if you disable page file then reboot. it would either disappear or you can manually delete it. if it keeps on coming back, that means your page file is active.