Jump to content

Recommended Posts

Posted

I'm buying a new computer. To

play at 4k So how much system ram is recommended?

I'm looking at 96 Gb.

Will that give me a perfect performance?

Maybe I need more like 128 Gb or even more?

What about vram. I'm looking at buying an nvidia 5080. How much vram is recommended for 4k

Posted (edited)

64GB of RAM is recommended for DCS, especially if you tend to do complex missions and/or get into MultiPlayer (40GB+ of RAM usage is not uncommon). 
Of course, 96GB will be even better - that's kind of overkill today but at least you won't bother again about RAM for years. 
There is no need for 128GB, that's way overkill.

The thing is, not all RAM is equal (speed and latency varies), and advice on which memory kit for your to get will depend on the system...
Which processor and motherboard is going to be used for that new system?❓


As for the VRAM on the graphics card, models with 16GB are recommended.
DCS has many large textures being triggered and swapped at the same time, it can easily consume over 12GB of VRAM, and even more so when in MultiPlayer. 

If it's for a 4K screen, either RTX 4070Ti Super or RX 7900 XT will be good. RTX 4080 Super or RX 7900 XTX even better.

You may wonder about something extreme like an RTX 4090, but spending on that one only makes real sense if you're going to play DCS in VR.

Edited by LucShep

CGTC - Caucasus retexture  |  A-10A cockpit retexture  |  Shadows Reduced Impact  |  DCS 2.5.6 - a lighter alternative 

DCS terrain modules_July23_27pc_ns.pngDCS aircraft modules_July23_27pc_ns.png 

Spoiler

Win10 Pro x64  |  Intel i7 12700K (OC@ 5.1/5.0p + 4.0e)  |  64GB DDR4 (OC@ 3700 CL17 Crucial Ballistix)  |  RTX 3090 24GB EVGA FTW3 Ultra  |  2TB NVMe (MP600 Pro XT) + 500GB SSD (WD Blue) + 3TB HDD (Toshiba P300) + 1TB HDD (WD Blue)  |  Corsair RMX 850W  |  Asus Z690 TUF+ D4  |  TR PA120SE  |  Fractal Meshify-C  |  UAD Volt1 + Sennheiser HD-599SE  |  7x USB 3.0 Hub |  50'' 4K Philips PUS7608 UHD TV + Head Tracking  |  HP Reverb G1 Pro (VR)  |  TM Warthog + Logitech X56 

 

Posted
1 hour ago, LucShep said:

There is no need for 128GB, that's way overkill.

There is no amount of memory that would be useless in DCS currently. Since process manager could show ~32GB in use all other RAM fully utilized as standby mem for fast re-access once opened files, and even bare CoreMods folder now has a weight around 135GB without any pepsi maps

  • Like 1
  • 4 weeks later...
Posted

Take care when you select the RAM for your new PC! I've upgraded my PC after more than a year from 64 GB (2 x 32 GB) to 128 GB (4 x 32 GB), using the identical DRAM5-5200 Modules.

As my system didn't boot with XMP enabled (even though this configuration is in the QVL for my Mainboard), I sadly had do deactivate XMP and lower the DRAM frequency to 4800 MHz.

It's obviously a known behaviour that you may won't benefit of the full DRAM speed, if you populate all four DRAM slots. So try to get 2 x 48 GB or 2 x 64 GB modules if possible to get the full speed of your DRAM Modules.

[Modules] A-10C, A-10C II, AH-64D, F-4E, F-14A/B, F-16C, F/A-18C, FC3, Ka-50, P-51D, UH-1H, CA, SC
[Maps] PG, NTTR, Normandy, Sinai, Syria, TC

[OS] Windows 11 Pro
[PC] MSI Pro Z790-A, i9-13900K, 128 GB DDR5, RTX 4090 24 GB GDDR6X, 2 x SSD 990 PRO 2 TB (M.2), Corsair 5000D Airflow, HX1500i, H150i RGB Elite, Acer X28, TM HOTAS Warthog (Grip@WarBRD Base), MS SW FFB2, Thrustmaster TFRP, TrackIR 5 & TrackClip Pro
[Checklists] A-10C, F-16C, F/A-18C, AH-64D, Ka-50, UH-1H

Posted
On 11/2/2024 at 9:13 PM, rotan1999 said:

 

Will that give me a perfect performance?

 

No amount of hardware you throw at it will give you perfect performance in DCS. 

Specs: Win10, i5-13600KF, 32GB DDR4 RAM 3200XMP, 1 TB M2 NVMe SSD, KFA2 RTX3090, VR G2 Headset, Warthog Throttle+Saitek Pedals+MSFFB2  Joystick. 

Posted (edited)
2 hours ago, AstonMartinDBS said:

Take care when you select the RAM for your new PC! I've upgraded my PC after more than a year from 64 GB (2 x 32 GB) to 128 GB (4 x 32 GB), using the identical DRAM5-5200 Modules.

As my system didn't boot with XMP enabled (even though this configuration is in the QVL for my Mainboard), I sadly had do deactivate XMP and lower the DRAM frequency to 4800 MHz.

It's obviously a known behaviour that you may won't benefit of the full DRAM speed, if you populate all four DRAM slots. So try to get 2 x 48 GB or 2 x 64 GB modules if possible to get the full speed of your DRAM Modules.

In this case you're likely better off taking out two sticks and run 64Gb at higher speed and better timings instead of cobbling together 128Gb and running them and snail-speed JEDEC speed and timings. 

Running 4 sticks these days can still be a big ask depending on your motherboard and memory controller on your CPU. Couple that with potentially mismatched (even though they may be same "model") memory sticks, makes it that much less likely to have success. 

My recommendation for DCS at this point, get the fastest two sticks you can get, either a 64Gb or 96Gb kit. 

 

Edit: looks like you were running pretty slow 5200 DDR5 (which I would still question "why" on a 13th gen Intel...), so you're not losing as much as I initially thought, but your timings are still significantly slower at JEDEC vs XMP. 

Edited by EightyDuce

Windows 11 23H2| ASUS X670E-F STRIX | AMD 9800X3D@ 5.6Ghz | G.Skill 64Gb DDR5 6200 28-36-36-38  | RTX 4090 undervolted | MSI MPG A1000G PSU | VKB MCG Ultimate + VKB T-Rudders + WH Throttle |  HP Reverb G2  Quest 3 + VD

Posted
4 hours ago, AstonMartinDBS said:

Take care when you select the RAM for your new PC! I've upgraded my PC after more than a year from 64 GB (2 x 32 GB) to 128 GB (4 x 32 GB), using the identical DRAM5-5200 Modules.

This was your mistake. You already had very slow RAM and you should really have replaced it, instead of adding more slow memory.

I would suggest 2x48 GB DDR5-6000 if you can live with a little less than 128 GB, but otherwise 4x32 GB DDR5-6000. 

Posted (edited)
3 hours ago, EightyDuce said:

looks like you were running pretty slow 5200 DDR5 (which I would still question "why" on a 13th gen Intel...),

As I configured my PC in 05/2023 I checked the Memory QVL of my MB. I equipped it with 2 x 32 GB initially and set the priority to the upgradability to 128 GB. This was the reason, why I've chosen those DDR5-5200 modules.

image.png

 

1 hour ago, Aapje said:

This was your mistake.

I didn't know about the restrictions when populating all four DIMM slots until I got the POST error message after upgrading to 128 GB a few days ago... the QVL promises a "Supported Speed" of "5200" with 4 DIMMs populated.

 

1 hour ago, Aapje said:

You already had very slow RAM and you should really have replaced it, instead of adding more slow memory.

"Very slow" is exaggerated. Well, the modules are not the fastest, but I see absolutely no performance loss after decreasing the DRAM frequency from 5200 to 4800. Everything runs smooth as usual. And I think an upgrade to 6000+ MHz modules isn't worth the performance benefit.

Edited by AstonMartinDBS
  • Like 1

[Modules] A-10C, A-10C II, AH-64D, F-4E, F-14A/B, F-16C, F/A-18C, FC3, Ka-50, P-51D, UH-1H, CA, SC
[Maps] PG, NTTR, Normandy, Sinai, Syria, TC

[OS] Windows 11 Pro
[PC] MSI Pro Z790-A, i9-13900K, 128 GB DDR5, RTX 4090 24 GB GDDR6X, 2 x SSD 990 PRO 2 TB (M.2), Corsair 5000D Airflow, HX1500i, H150i RGB Elite, Acer X28, TM HOTAS Warthog (Grip@WarBRD Base), MS SW FFB2, Thrustmaster TFRP, TrackIR 5 & TrackClip Pro
[Checklists] A-10C, F-16C, F/A-18C, AH-64D, Ka-50, UH-1H

Posted
14 minutes ago, AstonMartinDBS said:

"Very slow" is exaggerated. Well, the modules are not the fastest, but I see absolutely no performance loss after decreasing the DRAM frequency from 5200 to 4800. Everything runs smooth as usual. And I think an upgrade to 6000+ MHz modules isn't worth the performance benefit.

The transfer speed is only one factor. The timings are generally at least as important for gaming. So your test doesn't paint the full picture, since going from 5200 CL40 to 6000 CL32 is going to be way more of a difference than the difference in speeds would suggest. In fact, timings are relative to the transfer speed, so having both worse CL and worse transfer speeds, means that the actual timings are a lot worse than they seem when comparing the CL-numbers.

For example, 5200 CL40 has a latency that is a bit over 15 nanoseconds, while 6000 CL32 is a bit over 10 nanoseconds. So while the difference in CL is about 20%, the latency difference is nearly 50% due to this multiplier effect.

Posted
33 minutes ago, Aapje said:

The transfer speed is only one factor. The timings are generally at least as important for gaming. So your test doesn't paint the full picture

You're right, but my timings @4800 are the same as @XMP-5200, so the performance loss should be around 7,7 % compared to XMP-5200:

image.pngimage.png

For me it is important, that I see no performance loss in DCS or FS2024. Only benchmarks will show the difference.

[Modules] A-10C, A-10C II, AH-64D, F-4E, F-14A/B, F-16C, F/A-18C, FC3, Ka-50, P-51D, UH-1H, CA, SC
[Maps] PG, NTTR, Normandy, Sinai, Syria, TC

[OS] Windows 11 Pro
[PC] MSI Pro Z790-A, i9-13900K, 128 GB DDR5, RTX 4090 24 GB GDDR6X, 2 x SSD 990 PRO 2 TB (M.2), Corsair 5000D Airflow, HX1500i, H150i RGB Elite, Acer X28, TM HOTAS Warthog (Grip@WarBRD Base), MS SW FFB2, Thrustmaster TFRP, TrackIR 5 & TrackClip Pro
[Checklists] A-10C, F-16C, F/A-18C, AH-64D, Ka-50, UH-1H

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...