Is there any benefit getting more than 32GB of RAM for DCS?

[klabo_71st]

What I should set at paging files with 64 GB of RAM ?

[Hiob]

2 hours ago, klabo_71st said:

What I should set at paging files with 64 GB of RAM ?

You can leave it alone. Use the Windows standard settings. It won't be used anyway.

If it does make you feel better to fix it, set min and max to 16384 MB.

[klabo_71st]

Thank you. Already have nice improvement in busy missions, specially when rain or snow. Unplayable with my setting with 32GB of RAM.

Set my Paging files to 96000 max192000  

[BitMaster]

You should always have a little more RAM than you will ever need. It's been like that since forever

Edited May 19, 2022 by BitMaster

[klabo_71st]

[Bossco82]

Hi guys, I'm running 32gb of G Skill 3200C14 and I have a 3090. However I have recently got into multiplayer and the general rule is you need 64gb.

Can any multi player users tell me their max dram usage in multi. I haven't maxed out yet or seen a performance drop, its hovering between 22-27gb on Syria. I have started looking on eBay for a 64gb kit though. Prices are all over the place for nice kits, UK anyway.

[Max Thunder]

My max usage today was 45GB. Tried everything to max out the RAM. Switched between airplanes, locations an so on.

[Bossco82]

Thanks Max Thunder for the response. I have 64gb now and have seen between 36-42gb of usage in DCS multiplayer, so well worth the upgrade.

[Gambit21]

On 4/20/2022 at 8:36 AM, Hector45 said:

Don't expect the XMP profile to work with two separate kits, I learned that the hard way myself. Hopefully you get lucky.

 

Worked fine for me.

[Headwarp]

My 64GB kit arrived today and I parked on the airfield that hits me the hardest.  My FPS still tanked to the 30's here and there but I loaded in faster, and so far it does seem like it's a smoother experience overall.   I haven't personally seen afterburner report more than 32GB usage but, I have seen it use at least that much.   This was on a completely full growling sidewinder with MSAA 2x enabled. I was at 45fps 99% of my flight at least.  

I'm starting to debate how much I'm being affected by having only 12GB of VRAM at this point.. but that's another thread.  

Going to play around with settings, msaa on/off and see how that goes. 

 

*Edit* i turned off MSAA and bumped my SS% in the index to 150% and while the problematic airfields still had me lower than I'd like... in just about any other case on the ground I was at 75fps, I'd see a spike in the 60's occasionally.   But once I had any altitude I was hitting 80FPS with the headset set to 80hz.    While I did experience typical issues that low fps in VR might cause, the gameplay was smoother.  

I feel like I can confidently say going to 64GB has improved my experience in DCSW MP.   

Edited July 8, 2022 by Headwarp

[CobraKhan]

With DDR4 is there any performance difference between 2x32GB and 4x16's ?  I have an i9-9900K processor, and I am running everything on a single Samsung 970 EVO 1TB drive that is now just over 1/2 full.  I am pretty much maxing out my existing 32 GB of RAM.  And with all of the other Window's functions running off of the single disk while I try to play DCS, well it likely is not helping the situation!  In a big multiplayer mission, performance issues are so bad that I often cannot even participate anymore so I need to make some changes ...  just don't know if 64GB of RAM is my solution or not?

[Headwarp]

1 hour ago, CobraKhan said:

With DDR4 is there any performance difference between 2x32GB and 4x16's ?  I have an i9-9900K processor, and I am running everything on a single Samsung 970 EVO 1TB drive that is now just over 1/2 full.  I am pretty much maxing out my existing 32 GB of RAM.  And with all of the other Window's functions running off of the single disk while I try to play DCS, well it likely is not helping the situation!  In a big multiplayer mission, performance issues are so bad that I often cannot even participate anymore so I need to make some changes ...  just don't know if 64GB of RAM is my solution or not?

If you live in the US, can afford the experiment, and live near a Best Buy they accept returns on just about anything up to 15 days.   If they don't have it in stock locally, order it online.  Return it to the store.   Eat the shipping cost if you find it doesn't give you an improvement, you won't have to pay return shipping if you can drop it off at a store front location. 

I'm on day 2 of having my kit and so far it seems like an improvement.   I can't promise you'll experience the same.  I'm still questioning whether it's all in my head even though I feel like having the headroom has offered a smoother experience.   I was having a pretty good time on a full GS earlier spawning in at Kutaisi,  60fps on the ground 75-80 while flying at 150%SS in my Index.  an FPS drop here and there but none of the stuttering.  

I list all my specs including display in my signature not for bragging rights but because there's a lot to consider when talking about performance bottlenecks and ymmv.   Okay.. my peripherals are kind of bragging rights but still. 

Edited July 8, 2022 by Headwarp

[BitMaster]

vor 14 Stunden schrieb CobraKhan:

With DDR4 is there any performance difference between 2x32GB and 4x16's ?  I have an i9-9900K processor, and I am running everything on a single Samsung 970 EVO 1TB drive that is now just over 1/2 full.  I am pretty much maxing out my existing 32 GB of RAM.  And with all of the other Window's functions running off of the single disk while I try to play DCS, well it likely is not helping the situation!  In a big multiplayer mission, performance issues are so bad that I often cannot even participate anymore so I need to make some changes ...  just don't know if 64GB of RAM is my solution or not?

It's not possible to really say yes or no as it depends on the IC's used to make your modules. 

The 32GB module is likely made of 16x 16Gbit IC's = 32GB/module = 2 Ranks/module = 4 Ranks total

The 16GB module can be 16x 8Gbit -or- 8x 16Gbit. Samsung B-die only came as an 8Gbit die, Micron i.e. has 16Gbit IC's, etc... they have the pros and cons.

 

What you were asking I assume is the amount of Ranks vs. bandwidth.  In any case, you have at least 4 Ranks, maybe even 8. The more ranks, the more stressfull for the IMC.

It all gives and takes, you gotta find the sweetspot.  16Gbit IC's have a lesser timing but can achieve higher frequencies, so in order to make them shine you have to fly north of 4000MHz

to overcome the latency penalty. On the other hand, a nice B-die may not like speeds at or above 4000MHz but will offer you great timings. 

 

Take your pick

[bartoshka]

So, I'm rebuilding my PC. I could get definitely faster 48GB or slower 64GB. It's just physics, nothing specific to particular hardware (aiming for Hynix A-die anyway), generally if you want to hit the absolute top speed you can't go more than 32GB. So, what would you choose? 64 or a bit faster 48?

Edited September 14, 2023 by bartoshka

[kksnowbear]

Without going into a lot of agonizing detail, I'd do 64 no question (and I actually did, it happens).  Mind you that assumes the difference in speed is within a certain range.  You don't specify speeds or CAS level (which is important and can 'level the field by itself).

Admittedly I haven't tested it, but it does seem there are a lot of (at least anecdotal) experiences reported that the sim will use more than 32G.  So I think having more RAM is likely a more noticeable difference 'in game' than if the memory were only slightly faster (which is something you're not even likely to notice, TBH).

Generally, unless there's a very big difference in speeds, then I think - assuming the game uses it (which DCS apparently does) - that having twice the RAM is going to be of more benefit overall than something that's slightly faster.

Just my $0.02 FWIW

[bartoshka]

That's my thinking. In general I'm trying to go for the top. I got the Trident Z5 6400c32 2x32. While not the best 64GB kit available, they're pretty close since AFAIK I got the A-die. I'd expect single digit performance gain with 48 vs 64 gigs, which I would argue can be noticeable.. Check this out as example: https://youtu.be/ghJb0ttZMV4 .

Truth be told I'm already giving up performance in virtually all other games (not going 32GB way, which is undoubtedly the fastest). I just wonder whether 48 vs 64 can really make a difference. From what I've read it seems like 48 can be balancing near the edge, while 64 would providde some margin. Will probably stay with this choice.. Although it keeps bogging me.

EDIT:
As for the CPU I'm talking agnostically to focus on the memory only, the rest of the system being off topic really. You can always find a bottleneck

Edited September 14, 2023 by bartoshka

[kksnowbear]

EDIT:

OK, so digging deeper...turns out the links are somehow goofy and they don't take you to the RAM that he says he's actually testing.  So there's that.

But, even when I finally scanned through the video and found where the speeds/latencies are shown, here's what we come up with:

32G 8000 CL36 (9ns)
48G 8000 CL36 (9ns)
64G 7200 CL32 (8.88ns)

So these are supposedly what he's actually testing.  (Would've been nice had he actually mentioned accurate values for CAS somewhere so one could search the transcript, or linked the actual modules he was testing). Anyhow...given the values he listed in the video, there's an obvious difference...the 64G kit is a slower kit.  Even with better latency, it's still not as fast as either of the other two kits.

I do realize the state of affairs is such that you're pretty much topped out as far as kits with certain speed/latency...but, if the test is supposed to be apples and oranges (as it should be), then he should be testing kits of the same speed and latency.  Otherwise, it's just not a valid conclusion.

Laying aside the differences above, let's look at the actual results in the video:

I don't think he actually discussed his methodology, but particularly if the numbers he cites are not an aggregated result of (many) multiple runs, they're not terribly accurate.  I'm not sure how he conducted his tests, but the differences in those charts could easily be margin of error.

He's running tests with a 4090 at 1080p; frame rates are in the 200-300+ range (for obvious reasons).

The differences in the game charts are single digits of multi-hundred values - in other words, usually around 2-5%.  And while this can actually be measured, no one is ever going to be able to reliably and accurately notice 5 frames out of 300+.

Toward the end he finally goes up to 1440 resolution and sees at one point on the order of 16 frames out of ~255, which amounts to ~6%.  Again, though it can be measured, no one can reliably and accurately "see" the difference in 247 and 263 FPS. (In fact, as I mentioned earlier, this is fairly easy to prove, but that's another discussion)...

Anyhow, I've edited my original reply after looking more closely at the video.  It was kind of a waste of time TBH because the conclusion is exactly the same: Forget about some minuscule difference in a value that, although possibly measurable, isn't noticeable in any practical context, and go with what can make a much bigger difference (i.e., having 64G RAM vs 32 for running DCS).

As always, my own $0.02

Edited September 15, 2023 by kksnowbear

[BitMaster]

I would not invest too much time, ideas, hopes and finally money into the fastest RAM kit you can obtain. There are many reasons for this unless you have deep pockets and generally don't mind spending hours over hours testing between 1.46v and 1.48v...maybe that CL will hold with 0.01v more..maybe not... etc..

 

CPU and board play a big role on how far you can push things anyway ( faster with MHZ and lower with latency ), the more modules ( 2 vs. 4 ), the faster and lower the latency the more grey hair !.

My tip, Capacity over Speed, always. Nothing is worse than too little memory in your system, nothing will fix that but more memory. The penalty in gameplay experience between 3200 and 3600 is smaller than a system with too little RAM vs. one with just enough RAM. Look at servers and Workstations, Capacity over Speed. Reliability over "the last FPS tweaked out"

 

If you are lucky, you can tune your RAM to some nice values. But it needs luck, patience and some skills which you can learn. Just dont blame anybody if you spend a whole night booting and testing...LoL

[dugite57]

Here's what happened to me.  Due to stock availability I ended up with 4 x 16gb sticks EXPO 6000/36

With bios on auto the 4 sticks run at 3600hz, 2 sticks run at 5300hz (or is it 5200?).

hit EXPO and it runs at 6000hz, But it wont do a cold boot.

I observed stuttersn at 3600hz but not at 6000hz,

I pulled out 2 sticks and ran the memory (now 32gb instead of 64gb), EXPO and 6000hz. PC now cold boots without issue.  DCS runs smooth without stutter, This is with HUEY and Channmel Map. I know it's 1944 but it's great low level eye candy.  Same in the Last Stand Huey campaign. 

So, for me on 7800X3D the game runs without stutter with 32gb @ EXPO 6000hz,  If I run 4 sticks for 64 gb it defaults to 3600hz and stutters.

So I'm happy with less but faster RAM. I'll stick with this while the memory evolves and then do a 2 x 32gb upgrade.  If I had my time again and was building new and wanted 64gb I'd not let time presure and bad advice from PC shop lure me into 4 sticks. My advice is to stick with 2 sticks of whatever capacity you desire,

Two sticks good, four sticks bad. 

[kksnowbear]

4 hours ago, dugite57 said:

Two sticks good, four sticks bad. 

While I understand everyone's experience is different, I strongly disagree with the conclusion being drawn.

For one, if this were true, there probably wouldn't be any four-slot boards - yet there are lots.

There also definitely wouldn't be lots and lots of people running four sticks - yet there are.

In the time I've built and maintained computers professionally (40 years or so), there have been challenges at times getting four sticks to run - but it's not impossible. There are actual technical reasons it can be problematic, but just as well, there are real technical solutions.

As stated above, too little RAM can be a dismal failure where faster RAM is often unremarkable.

My advice, work with the right professional.

[Mr_sukebe]

Agreed with the above post.  That issue sounds more like an issue with either tuning the use of the motherboard, or a compatibility issue between the RAM and motherboard.

[Eugel]

4 hours ago, dugite57 said:

So, for me on 7800X3D the game runs without stutter with 32gb @ EXPO 6000hz,  If I run 4 sticks for 64 gb it defaults to 3600hz and stutters.

 

Two sticks good, four sticks bad. 

This sounds to me like some problem with the mainboard or bios settings and not normal at all. So I wouldn´t draw the conclusion that four sticks are worse than two.

I don´t have any experience with DDR5, but I upgraded my 2 x16 GB of DDR4 Ram to 4x16 GB and didn´t get any changes in speed, they still run with their full speed and I get less stutters in game.

Where in your bios does it show this 6000 Mhz/3600 Mhz ? You know that the I/O frequency of DDR Ram is half of what the designation of the modules is ?
DDR5-6000 would normally run at 3000 Mhz, and because it´s "DDR" Ram (double data rate), that number is doubled. But the clock rate is still only 3000 Mhz.
So, if you set them to 3600 Mhz, you could actually be overclocking them.

[kksnowbear]

Right...these are among the "technical solutions" I was referring to, of course.  Not by any stretch the extent of the methods, either (after even a little legwork, I find it seems likely that several other factors are at work).

I didn't see the need to go into detail, TBH, because (to me) there appears a very limited understanding of the subject at hand, and - in my professional assessment - this is not only causing the problem, but is likely to make arriving at a solution exceptionally difficult if not impossible for that poster, unaided.

To prove the point, the end result is a conclusion which is almost certainly inaccurate (i.e. two sticks good, four sticks bad).

Again, my advice - particularly in that case - work with a competent professional.

Edited September 22, 2023 by kksnowbear

[BitMaster]

Usually, these symptoms originate from motherboards with sub-optimal memory traces on the PCB. There is not much that you can do but change the mobo. It will likely not work with 4 modules at higher speeds. Signal integrity and timing issues are the main culprit afaik, but that is beyond our reach to fix. 

Bottom line out of my experience. Your board won't run 4 sticks at higher speeds and good latency, maybe not even with high latency. I have been there with a few boards myself and it's a waste of time to try to get it working. 

 

my 2 cents

[kksnowbear]

1 hour ago, BitMaster said:

Usually, these symptoms originate from motherboards with sub-optimal memory traces on the PCB. There is not much that you can do but change the mobo. It will likely not work with 4 modules at higher speeds. Signal integrity and timing issues are the main culprit afaik, but that is beyond our reach to fix. 

Bottom line out of my experience. Your board won't run 4 sticks at higher speeds and good latency, maybe not even with high latency. I have been there with a few boards myself and it's a waste of time to try to get it working. 

 

my 2 cents

Naturally, the "technical solutions" I referred to includes hardware choice, albeit that is unfortunately in retrospect at this point.

If this was the result of working with a competent professional then they should be willing to either demonstrate it can work, or issue a refund (provided it's within a reasonable time frame and no misuse or damage has occurred).

Edited September 22, 2023 by kksnowbear