1 hour ago, Noel said:

Great, I was just using ImDisk and could never get Windows not to create a small page file during initialization so was caught in a catch 22.  Have you determined use of a RamDisk has any net positive effect for MSFS, and if so what if you will?  And...did you use Primo when you had 32Gb installed?  I guess that is the question w/ regard to MSFS, at least for this issue, of 32 v 64.

I think more benefit can be obtained by using the other product from Romex Software: PrimoCache - there should be a trial version of this available as well. It's been a while since I've tested with FS2020 but I did measure improved load times when using PrimoCache.

The big catch to using these tools however is that they significantly reduce your available RAM. You don't want your system to start using the page file at all if you can help it. Even with the pagefile on a Ram disk you'll encounter stutters in any application when your system starts paging.

I have 4 x 32 gig modules sitting in the cart of an online retailer waiting for stock to arrive at the moment. 😁

2 hours ago, JSmith2112 said:

Even with the pagefile on a Ram disk you'll encounter stutters in any application when your system starts paging.

I have 4 x 32 gig modules sitting in the cart of an online retailer waiting for stock to arrive at the moment.

What kind of evidence do we have that paging is substantially reduced when you have more  RAM installed?  I have 32Gb, rarely see more than 15Gb in use running MSFS plus 4 or 5 other apps, though have seen as high as 21Gb briefly.  My guess is that the #1 priority for Windows Memory Management is to NEVER run out of Virtual Memory (physical + page file) so leaves a lot of physical RAM out of service at any moment in time as a giant buffer.   It's too bad you can't completely stop paging and see what happens but Windows and I guess programs made to run in Win 11 won't let you.  After all, if it doesn't work then can revert to pagefile.sys.

6 hours ago, Noel said:

Seems like it would be more useful to see read/writes to the page file in events per unit of time, rather than a % of the total available page file.  Mine's at 13,336mb right now that is Window's managed so this current capture at climb/cruise in the A320NX showed 0.5% which is about 67mb in use.  But how many read/write operations per second?

 

It doesn't make sense to me that *any* actual paging would occur when you are maintaining over half of your 32GB of physical memory available.

Add to that the fact that the page file usage remains constant, and I would guess that the page file is just being populated with tabular housekeeping data like page allocations and block addresses to be used in case paging were to be triggered, and not actually performing paging of program data to/from the file system.

You should be able to use Mark Rossinovich's ProcMon utility to capture some metrics on I/O throughput to/from the page file.  https://learn.microsoft.com/en-us/sysinternals/downloads/procmon

 

5 hours ago, Noel said:

It's too bad you can't completely stop paging and see what happens but Windows and I guess programs made to run in Win 11 won't let you.  After all, if it doesn't work then can revert to pagefile.sys.

There may be a way around this... creating a RAM disk and setting the page file on it.

Because disabling the page file completely often makes the system unstable even with 64 GB or more.

For what it's worth, I jumped on this bandwagon for the fun of it and am now on 64Gb rather than 32Gb. Same brand and same timings so those are constants.

My experience is:

There is now less of a stutter when swapping between internal and external views. Not sure how I'd measure it, but it's appreciable.

Much less appreciable, but also apparent, is a slight increase in what people call "smoothness" : not the average FPS necessarily, but maybe low 0.1% / 1% are fractionally better?

Edited October 11, 20241 yr by tfm

9 hours ago, Noel said:

What kind of evidence do we have that paging is substantially reduced when you have more  RAM installed?  I have 32Gb, rarely see more than 15Gb in use running MSFS plus 4 or 5 other apps, though have seen as high as 21Gb briefly.  My guess is that the #1 priority for Windows Memory Management is to NEVER run out of Virtual Memory (physical + page file) so leaves a lot of physical RAM out of service at any moment in time as a giant buffer.   It's too bad you can't completely stop paging and see what happens but Windows and I guess programs made to run in Win 11 won't let you.  After all, if it doesn't work then can revert to pagefile.sys.

You can actually disable it. Pagefile.sys would still be there. However if for some reason a program even if the chance of it happining is very small expects it  the machine will BSOD and reset.

Edited October 11, 20241 yr by 40track

Not really recommended to mess with the pagefile, mainly because even if you've got plenty of RAM, it's still being used for minor swaps and pages and software expects it to be there. I know some Adobe software flat out refuses to work if you disable it. So could be fun to disable it for testing purposes, in MSFS at least but for daily operations I'd definitely leave it alone.

FWIW, I'm using HWiNFO and when I load PMDG 738 into KLAS just now current 'Virtual Memory Committed' is over 37Gb (in my 32Gb RAM system).  'Committed' seems to be a function of what everything currently running is 'requesting' be guaranteed by the OS to be available should it be needed.  I tried using a smaller custom page file size of 4Gb and even that will not allow MSFS to fully load to the departure gate which makes perfect sense given the amount of VM being requested.  Mostly curious if 64Gb ends up substantially reducing paging to disk as it's easy to believe paging to disk might play a role in some kinds of stuttering.   One would certainly hope the OS would look at that and start paging to RAM, not disk, if paging is unavoidable.

Edited October 11, 20241 yr by Noel

14 hours ago, Bob Scott said:

It doesn't make sense to me that *any* actual paging would occur when you are maintaining over half of your 32GB of physical memory available.

The logic underneath the sense you have seems to suppose Win MM will default to using physical RAM over pagefile.sys when there is that much of a glut of seemingly available RAM.  That certainly was my assumption when it came down to considering 64Gb upgrade, hence wasn't sure it would do anything meaningful.  As you can see in this thread there are people who upgraded and saw little to no difference, and others who say they see a difference in stuttering. 

14 hours ago, Bob Scott said:

You should be able to use Mark Rossinovich's ProcMon utility

I installed ProcMon and used ChatGPT to advise me how to set this up.  It had me create filters for 2 operations, FileWrite and FileRead and assign the path for that to pagefile.sys.  I opened this while MSFS was running at KSFO, proceeded to open a few videos I had stored, and also open 5 web pages and ran video there on YT.  There was not one hit, which of course makes me wonder if I have ProcMon set up correctly for monitoring read/writes to pagefile.sys, or it's indeed not paging, and yet HWiNFO says X% of page file "in use".  Next time I will create a RAMdisk of 16Gb to mask half my RAM so that MSFS is forced to page and see what happens.

 

I already have 64GB ram. Got it on some kind of whim, a while ago.

2 hours ago, Noel said:

I installed ProcMon and used ChatGPT to advise me how to set this up.  It had me create filters for 2 operations, FileWrite and FileRead and assign the path for that to pagefile.sys.  I opened this while MSFS was running at KSFO, proceeded to open a few videos I had stored, and also open 5 web pages and ran video there on YT.  There was not one hit, which of course makes me wonder if I have ProcMon set up correctly for monitoring read/writes to pagefile.sys, or it's indeed not paging, and yet HWiNFO says X% of page file "in use".  Next time I will create a RAMdisk of 16Gb to mask half my RAM so that MSFS is forced to page and see what happens.

This is above my head. For now I'll just monitor page file use in HWiNFO. 

Currently at LFPG, Fenix A320, FSHud with 20 active 30 parked, REX Atmos, Your Controls, mobiflight connector, map enhancement mod.

Physical memory used 23.5GB page file 0.3%

VRAM 22.8GB of 24GB

...

I don't understand the other values e.g. virtual memory commited (55GB). I'm using default page file settings:

Edited October 11, 20241 yr by Cpt_Piett

11 hours ago, tfm said:

My experience is:

There is now less of a stutter when swapping between internal and external views. Not sure how I'd measure it, but it's appreciable.

That is exactly my experience going from 32 to 64 - no stuttering when switching views in a complex aircraft (especially after completing a long flight). My GPU also seems to be running a couple of degrees cooler, but that may be a coincidence.

36 minutes ago, Cpt_Piett said:

I don't understand the other values e.g. virtual memory commited (55GB).

'Virtual Memory' =  total capacity of page file plus physical RAM. 

My understanding is that 'committed' in this context is "a function of what everything currently running is 'requesting' be guaranteed by the OS to be available should it be needed."   A program like MSFS and many more may fluctuate in their demand for memory space.   When I launched PMDG to a gate at KSFO I think it said 37Gb committed, and that's all it means--it doesn't mean it will necessarily all be used but again this is what the OS will assure is available by adjusting what it can, which is of course the page file size.  In the above example it went from 37Gb committed down to 31Gb as we climbed to our cruise altitude.

So in your case at that moment somehow the OS decided you needed 32GB RAM + 33Gb pagefile.sys.  Whereas for me, the OS created a pagefile.sys with 13,336mb capacity.  Presumably as long as there is disk space it can change this on the fly.  I've not yet seen more than 37Gb committed after everything is loaded and running, and my 32Gb of RAM + 13.3Gb of pagefile.sys is 43.3Gb, and our committed max was still below that.

Off topic, but it seems to me I'm getting better fps in 2D and VR after Windows 24H2 replaced 23H2. I can't be sure.

4 hours ago, Fielder said:

Off topic, but it seems to me I'm getting better fps in 2D and VR after Windows 24H2 replaced 23H2. I can't be sure.

I think there’s performance improvements for all Ryzen CPUs in 24H2. Is 24H2 the official build now? I’m in Insider dev channel and I’ll probably be stuck on that until Windows 12…

https://www.forbes.com/sites/antonyleather/2024/08/27/windows-11-24h2-surprise-11-game-performance-boost-on-ryzen-cpus/

Edited October 12, 20241 yr by Cpt_Piett