Do you need 32gb RAM, and if so why?

User Support Hub Hardware & Peripherals
,
21

i would say its doubtfull more ram would help much in your instance (or frankly with the game in general unless you had really slow ram and also used multiple apps together, like streaming and discord at the same time)

ram is a temporary state and its dynamic - meaning the more you have the more the system will ‘try’ to use - the reason yours isnt using much is your ram is so fast it simply doesnt need to
with the high speed ram you have i would expect the mb thru-put or some other part to bottleneck Long before ram became any real issue on your system - all it has to do is ‘swap’ data fast enough to keep up with say the gpu and/or cpu in this case

EDIT: (just noticed you stated an hdd for cache? not a great idea, cahe is there to ‘speed’ up delivery so is just as important as any part of the game)
if any part of the game is on hdd, that would be your first upgrade imo, get nvme if possible but at least an internal ssd drive for everything is best

1 Like
22

Exactly this. If you’ve got it running at CL16 3800mhz stable that’s very good timings.

23

I upgraded from 32 gb to 64 gb (DDR4 2133) just for the hell of it (and it was on sale) and maybe magical thinking …

should I overclock to 3600 ? …

24

RAM was not cheap, but not too expensive, so I upgraded to 32GB a while ago.
I can’t say I’ve seen any real difference.

I’m happy with my frame rates (and I was happy before) and I have no stutters (and I had some before).
I experience no CTDs except for a reproducible one in a very specific location in Austria which I haven’t been able to troubleshoot. But I’m no techie and couldn’t honestly say it was RAM that cured my stutters.

System specs in my profile.

25

For me, RAM was the compromise when I had the PC designed last year. I just upgraded this afternoon from 16 to 32 and I wouldn’t say it’s an instant frame rate upper or anything (but didn’t expect so either).

I testloaded the Fenix A320 on MK Studios Rome, both pretty draining in terms of performance and it sure became stutterfree. I didn’t experience a lot of stutters before though, as I cap Vsync at 20% outage on a 165 hertz monitor with G-sync enabled too.

What I did see however, and made it worth it, was the memory usage being 18 gb. Obviously that was previously send to the cache file and it’s now nice to see that being handled internally.

GPU utilisation went up a lot to about 98% so I suppose that too made quite the difference.

I read a lot here on the forum prior to my upgrade and I too would recommend the upgrade. Mine was 180 euros which isn’t cheap but not terrible either I suppose.

26

I did put them to work dual channel. It’s a self-built system. I don’t think there is any reason not to put them in different channels (and the mobo manual downright says to do it), and I don’t believe such a thing as dual channel incompatible memory exists (and even if it did, it would not in 16gb two stick kit).

Heh, I wonder about that myself. The game and it’s addons and all the system stuff (e.g. page file) are on 4th gen NVMe SSD. I just put the rolling cache on HDD to preserve my poor little 1400 TBW SSD a bit. :smiley: In practice, I’m not really seeing noticable difference between running with the HDD cache and running without the cache, though perhaps SSD cache would speed it up will have to try. My impression of the rolling cache is that it’s more there to help preserve Asobo’s servers, but perhaps I’m wrong.

27

This video will explain it better than I can. It’s likely your ram is single rank, 8 GB sticks often are, 16GB sticks are always dual rank. You can probably google your specific ram and find out if its single or dual rank.

I know that for dual rank kits there’s no benefit to 4 sticks, so 2 is better for future upgradeability potential and stability improvement with less bus traffic. But if your sticks are single rank there might actually be a benefit to 4, although it wont be better than 2 x 16 GB

Watch the video and see if it makes sense, I’m not familiar with 8GB sticks so I’m not sure.

As regards the rolling cache I haven’t used it for months, i get better results with it disabled, although I have a fast fiber optic connection.

Here’s a direct game comparison 2 x 8 vs 4 x 8

2 Likes
28

Ah OK. A different concept then (and one I was frankly not familiar about). The vendor does claim they are single rank, Kingston’s website is surprisingly useless (though “8GB 1G x 64-Bit x 2 pcs.” seems to indicate single rank if I understood correctly). That would then indicate it would be reasonable to not expect a loss in performance if upgrading with 16gb more of the same.

1 Like
29

msfs2020
msfs20201920×1080 219 KB

Here’s my usage on Windows 10. A fraction of what Linux/Mangohud was reporting. I wasn’t expecting that.

For note, on Windows 10 I have MSFS 2020 and its cache on a 1TB NVME while on Linux they’re on a 3 disk 8TB Zraid. It wouldn’t surprise me that much if the disks and cache made up the difference in memory usage.

30

Yes, I think if its single rank and you can get 2 more identical sticks you might get a slight bump in performance even if you can’t maintain the same timings. I don’t fully understand it myself but I think this is because two ranks per channel is optimal. So 16GB sticks are already dual rank so with just 2 sticks you have all channels used. Putting in 4 sticks of dual rank ram means more traffic on the dual channels so timings may suffer, but with single rank ram it will take 4 sticks to fill both dual channels, so you aren’t saturating the bus.

So yes, its reasonable to think you might get a slight performance boost.

1 Like
31

Rank. That is a new term for me.

The term rank was created by JEDEC, the memory industry’s standards group, to distinguish between the number of memory banks on a module as opposed to the number of memory banks on a component, or memory chip. The concept of memory rank applies to all memory module form factors, though in general it tends to matter primarily on server platforms, due to the larger amounts of memory they manage.

A memory rank is a block or area of data that is created using some, or all, of the memory chips on a module. A rank is a data block that is 64 bits wide. On systems that support Error Correction Code (ECC) an additional 8 bits are added, which makes the data block 72 bits wide. Depending on how a memory module is engineered, it may have one, two, or four blocks of 64-bit wide data areas (or 72-bit wide in the case of ECC modules.) This is referred to as single-rank, dual-rank, and quad-rank. Crucial denotes this on the module label as 1Rx4, or 2Rx4, 2Rx8, or similar.

The x4 and x8 refer to the number of banks on the memory component or chip. It is this number, not the number of individual memory chips on a PCB, that determines the rank of the finished module. In other words, if a module has chips on both sides of the PCB, which makes it dual-sided, it can still be single-ranked, dual-ranked, or quad-ranked, depending on how those chips are engineered.

What is memory rank | Crucial.com.

Kind of getting too far into the weeds for me.

I have 64GB of 3600.

It is for the PC (Windows 11 and all Apps running under it) and not just FS2020.

FS2020 is running 4K Ultra and smooth (i9-990K & GTX 1660 TI).
My UserCfg.opt

UserCfg dot Opt Options 2
UserCfg dot Opt Options 21674×1446 67.7 KB

1 Like
32

Ha yep, we’ve definitely gone down the rabbit hole in this thread for sure. Hopefully it’s been useful/interesting to some.

1 Like
33

It was, thanks for the input all.

3 Likes