It depends on how many programs are running while you are running MSFS.
In my case, normally these tools are running:
MobiFlight - for WinWing FCU
SpadNext - for Saitek devices
FSHUD - for ATC
FSLTL - for AI Traffic
FSUIPC7
APLv2
Navigraph Charts
Flying with the Fenix I reached 87% of 32GB RAM with all the problems that come with it.
I removed the 2x16GB kit and replaced it with a new 2x32GB KIT that is slightly faster.
Now the occupied RAM does not exceed 45% but above all I noticed that the stutters have disappeared and the FPS have also increased.
Therefore in my case it was an absolutely correct choice.
I picked up a 64GB kit in the recent Amazon Prime Deals day. It was the cheapest it’s ever been and I can sell my existing 32GB kit so not too costly an upgrade.
I did my usual hardcore FPS test, taxiing at inibuilds Heathrow in the Fenix with FSLTL traffic. As expected, there’s no change in the FPS numbers, average or 99th percentile.
Seemed to be less stutters when panning around but that could just be placebo effect.
You’re confusing what it can use (committed) to what it’s actually using. You’ve stated several times your setup with 128 GB is stutter free, which implies it’s using it.
Feel free to show a screenshot of 128 GB of physical RAM being used, but I suspect we’ll be waiting a while.
As a general side note on this same subject, I’d strongly suggest getting a 2x24 GB non-binary kit if your motherboard supports it. These are single-rank as opposed to dual-rank 2x32 or 2x48 kits, which is not only easier to stabilize due to being easier on the memory controller, but the frequency can be pushed further if desired.
I’ll reiterate what was said earlier. In a CPU bound title, you should always prioritize RAM frequency over capacity above a certain point. Both improved latency and bandwidth can have a significant uplift on the 1% lows, particularly for MSFS.
As I wrote, the OS does not let you see the ram it’s using as an object code cache. All OSes do this now, in a completely transparent fashion, because that ram is marked as unused and can be claimed by programs at any time.
IIRC this is literally the same memory as the disk cache for native executables loaded from disk. It is displayed in aggregate as “cached memory” in Task Manager, but there’s nothing executable-specific in it.
This is really interesting, didn’t know this. I guess this could explain why people are seeing performance increases with 32 vs 64 GB, even though the sim didn’t even fill to near 32 in the first place. Maybe it’s the additional caching capability that is making a difference?
Hard to get around the little Dbag kiddies that are practiced thieves on ebay. Usually if I sell something electronic even to someone with a 150+ feedback that I’ll make a specific ‘dot’ with marker somewhere on them and take a pic. They’ve really shafted sellers over time(unless your some Chinese goliath). Being I don’t need a national audience for ram, would probably do something like FBMP.
Now I’m curious if 64 makes a difference, just because. It’s cheap enough these days should of checked on prime day.
What apps are you going to use along with the sim? Just load them all up with the sim and take a look at either Task Manager or something like HWiNFO64 to see how close you are to 32gig.
This will paint a key picture if you will actually need 48 or 64gig.
You definitely don’trecall correctly. This is a memory allocation scheme that even most low-level system developers never need to look at. Definitely not “cached memory” It’s pieces of executable code that can be reloaded w/o hitting the storage system - even faster than copying from RAM disk, because all you’re doing is changing a memory pointer instead of re-allocating memory and then copying to it.
Task manager has NO access to this memory whatsoever, only the memory allocator does. Look up the linux kernel slab allocator as one example.
To the operating system, it looks like you allocated some new memory and loaded the data into it - but that never happens. Just a pointer change and marking the memory as allocated.
[quote=“BearsAreCool510, post:111, topic:656398”]
Oof, what were you doing to hit 60? Any special plane, scenery, or setting?
[/quote]As an experiment, allocated 1tb as a rolling cache, on its’ own 4tb drive. Took forever, did squat, deleted it.
As others have discovered, if you have a fast internet connection, the rolling cache is just unnecessary overhead. More’s the pity.
A lot of younger guys seem to think static precautions are a myth. The reason being zapped modern RAM often functions fine for a while but given time or pushing it a bit harder will start to give grief.
One of the younger guys at work point blank refuses to take even basic static precautions and then 6 months later is saying the brand xxx RAM we bought for this laptop is useless, we only had it 6 months and it failed.
More likely they’re lying and you’ve been scammed. Sorry for your loss. I have 2 sticks of 16gb that were all that they had in stock during the chip crisis - they’re now sitting on the shelf “just in case” I need to replace one of the 8 sticks of 32gb ram (same manufacturer and specs) or decide to do a 3rd build with it. Or a friend or family needs it.
I’m a lot more comfortable giving it away to someone I know than trying to sell it on fleebay.Same as I’m a lot more comfortable buying locally than on Amazon (Amazon’s prices are higher anyway, and more and more people are buying there and assuming they’re getting a good deal).
Hello everyone, I followed this thread and, taking advantage of the discounts on Amazon, I bought 2 more Corsair 3600 sticks, and now I have a total of 64 GB of RAM. The PC has an AMD Ryzen 9 and Nvidia RTX 4070 Ti Super 16 GB. I haven’t gained any extra FPS (I keep the cap at 60 with ultra settings, and it drops to around 40 during taxiing in large, crowded airports).
What I can really confirm is that I no longer experience any stuttering under any conditions.
In the past, in certain scenarios, especially during the final approach, there was always some stuttering; now there’s none, anywhere, in any condition.
Thank you for sharing the changes you noticed regarding DDR 4x16GB RAM and stuttering in MSFS.
I had the opportunity to test DDR5 64GB (2x32) of RAM while playing MSFS and I didn’t notice any difference in performance. For example, there are still stutters that depend on the position of the sun.
(Ryzen 7 8700X3D, RTX 4090 Gaming OC, MSFS Ultra, TAA,2K)
When we’re outside and rotate the camera with the sun directly behind us - there’s a single big stutter.
However, I have to admit that I’m tempted to get 64 GB since RAM prices have dropped recently.
It is displayed in aggregate as “cached memory” in Task Manager, but there’s nothing executable-specific in it.