Jump to content

Recommended Posts

Posted (edited)

Hi,

at the moment i play a lot of vr and i heard that more than 32gb could be a good thing, so i want to upgrade.

 

now i have 32 gig of kingston fury beast with 2x16gb.

 

Iam not sure i read so much different stuff, should i buy 2x16gb again and work with 4 ram blocks or is it realy better (and stable) that i buy 2 new blocks to 32 gig. Its a bit pity that i cant use the 2x16gb anymore.

 

I cant decide because there is so much information about that out there, so i would love to know what you guys here think.

I have no problem with pay a bit more and have no problems.

 

my system

mainboard: TUF GAMING B550-PLUS

rtx4070

cpu:  AMD Ryzen 7 5700X, 8x 3.40GH

ram as you see

Edited by felstrikeraut
  • felstrikeraut changed the title to Upgrade Ram from 32 to 64gb
Posted

2x32 is generally recommended. Can be faster, and also prevents memory timing issues that crash DCS due to memory overclocking, XMP/EXPO profiles.

AMD 7800x3D, 4080Super, 64Gb DDR5 RAM, 4Tb NVMe M.2, Quest 2

Posted

2x32 is better, especially considering AMD is more sensitive to RAM than Intel. But 4x16 should work if you can get the same kit you already have.

I went from 32 to 64 some time ago and I can tell you it's a good improvement. Not night and day "I can't believe this" better, but an improvement nonetheless.

 

Posted (edited)

It may work better with 2 but it may work just as well with 4. 

There are so many variables that you have to try each configuration to actually know.

 

What you should know is, regardless of 32GB or 16GB modules, there are no 32Gbit DDR4 dies for Client Computers modules. There are 16Gbit dies but they have a latency penalty but alow speeds in the 4xxxMHz range....which is of no actual use or benefitial for your AMD 5700X/B550 combo...your world ends at 3600MHz, as does mine, see my sig.

The best way, for your config, is 4 modules with Samsung B-die with 16 dies on each module for 16GB size. B-Die only comes as 8-Gbit but is the best for lowest latency in your desired speed range 3200-3800 MHz. B-Die is not the best for way up the 4xxx range but it's the king in 3600ish MHz range and yet unsurpassed.

The thing you have to look for when looking at DDR4 kits and spot the B-Die based kits is the latency on 2nd tier , they all do 3xxx ( 3200-3800MHz ) 16-18-18-38 or 18-18-18-42 or worse but only B-die has 14-14-14-36 or 16-16-16-36 or 38, only B-dies afaik have secondary values of 16 or even 14, x-16-16-x.   A 3600MHz kit for you would be 3600-16-16-16-38  or maybe 36, 1T, maybe 2T. That is 99,99% Samsung B-die and no Micron or Hynix or other Samsung Die. You want "B"-die, THE "B" die

 

Still, even with the best kit with best knowledge, there are variables and hurdles, Bios versions, Volts, etc...  just want to mention that again, there is no guaranty. 

Edited by BitMaster

Gigabyte Aorus X570S Master - Ryzen 5900X - Gskill 64GB 3200/CL14@3600/CL14 - Sapphire  Nitro+ 7800XT - 4x Samsung 980Pro 1TB - 1x Samsung 870 Evo 1TB - 1x SanDisc 120GB SSD - Heatkiller IV - MoRa3-360LT@9x120mm Noctua F12 - Corsair AXi-1200 - TiR5-Pro - Warthog Hotas - Saitek Combat Pedals - Asus XG27ACG QHD 180Hz - Corsair K70 RGB Pro - Win11 Pro/Linux - Phanteks Evolv-X 

  • 1 month later...
Posted (edited)

I have 2x16 of brand A, and bought 2x32 of brand B. Both brands are 6000Mhz DDR5

Since I have 4 slots, can I use all of these ? Or will 2x32 perform better.

Also, does the placement order matter ?

16 -32 -32 16 

vs

16 -16 - 32 -32

Edited by winghunter

DCS Web Editor - New 3D Mission Editor for DCS that runs in your browser

DCS Web Viewer free browser based mission planner

dcs web editor signature2.png
4090 RTX, 13700KF, water cooled, Quest 3

 

Posted
vor 3 Stunden schrieb winghunter:

I have 2x16 of brand A, and bought 2x32 of brand B. Both brands are 6000Mhz DDR5

Since I have 4 slots, can I use all of these ? Or will 2x32 perform better.

Also, does the placement order matter ?

16 -32 -32 16 

vs

16 -16 - 32 -32

Not the best starting point but it may work.

You must not mix the kits !  There are 2 Memory channels, A and B. Each kit must populate their corresponding slots ONLY. Usually it's 2nd and 4th slot for kit #1 and 1st and 3rd slot for kit #2. 

What could be tricky is XMP. YOu will likely have to tune the faster settings by hand, use the XMP values as a guideline and stick to the looser timings if the secondary values differ.

**The weakest link in the chain** is the ruling thing here.

 

You may also want/need to up Volts for DDR by 0,05-0,10v

  • Like 1

Gigabyte Aorus X570S Master - Ryzen 5900X - Gskill 64GB 3200/CL14@3600/CL14 - Sapphire  Nitro+ 7800XT - 4x Samsung 980Pro 1TB - 1x Samsung 870 Evo 1TB - 1x SanDisc 120GB SSD - Heatkiller IV - MoRa3-360LT@9x120mm Noctua F12 - Corsair AXi-1200 - TiR5-Pro - Warthog Hotas - Saitek Combat Pedals - Asus XG27ACG QHD 180Hz - Corsair K70 RGB Pro - Win11 Pro/Linux - Phanteks Evolv-X 

Posted

As mentioned, the basic rules apply irrespective of MOBO maker or CPU type.

Use two slots rather than four.

Use RAM sticks of the same make, model, production date.

Check your Bios to ensure the RAM sticks are running at the correct speed and settings (ratio etc).

 

  • Like 1

AMD 3800x, Asrock 570X Taichi, 32GB Corsair Platinum, MSI 1080Ti, Corsair MP600 Gen 4 1TB NVMe. Samsung 970 EVO Plus 1TB NVMe, Samsung 860 1TB SSD, Custom Watercooling, AOC 32" 4K Screen.

Posted (edited)
15 hours ago, pegasus1 said:

As mentioned, the basic rules apply irrespective of MOBO maker or CPU type.

Use two slots rather than four.

Use RAM sticks of the same make, model, production date.

Check your Bios to ensure the RAM sticks are running at the correct speed and settings (ratio etc).

 

I received the same advice about using two sticks of the same make and production date vice mixing  up 4 even if they are the same brand bought at different times.

Edited by Tophatter14
  • Like 1
Posted
3 hours ago, Tophatter14 said:

I received the same advice about using two sticks of the same make and production date vice mixing  up 4 even if they are the same brand bought at different times.

Its not a hard and fast rule, a PC can work fine using difference RAM stick but with every variation you increase the chances of instability.

The best scenario for stability is to use RAM that is on the MOBO approved list and then set it to XMP/Expo in the Bios, but thats not to say you cant get 100% stability by using RAM that falls outside of this, it just increases the chances of things going wrong.

  • Like 2

AMD 3800x, Asrock 570X Taichi, 32GB Corsair Platinum, MSI 1080Ti, Corsair MP600 Gen 4 1TB NVMe. Samsung 970 EVO Plus 1TB NVMe, Samsung 860 1TB SSD, Custom Watercooling, AOC 32" 4K Screen.

Posted
13 hours ago, Tophatter14 said:

I received the same advice about using two sticks of the same make and production date vice mixing  up 4 even if they are the same brand bought at different times.

Just because it is the same brand doesn't mean that it is the same stick or even the same manufacturer. The manufacturers of the memory chips keep improving them, so later chips can be of a different quality, which then clashes with the poorer quality earlier chips.

Also, occasionally the companies that assemble the sticks can get a better deal by getting the memory chips from another manufacturer and they switch over, despite keeping the branding the same.

The safest is always to get a completely new kit of preferably 2 sticks. When sticks are packaged together, they are always from the same production batch and unless there is a manufacturing fault, they should be identical.

  • Like 1
  • Recently Browsing   0 members

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