Ram upgrade : add 2*8go to 2*16go

[fab.13]

Hello

is it possible to add a 2*8go dual kit on my motherboard ( asus z490 Intel 10th ) which already  have a 2 * 16go dual kit ?

Both kit are identical ( fréquence, timing, g.skill ) the only différence is the ram quantity

The goal is to get 48go as DCS ram need are increasing…I didn’y find the answer in the asus manual

thanks for sharing your expérience

 

[LucShep]

No harm in trying.  

Intel 10th gen (with Z490) was the last CPU line-up that, in my experience, permited that sort of RAM goofing around.
Just make sure that they are indeed the same make and model, same exact XMP settings (speed and timings, etc) before going for that mix-remix of RAM capacity.

Place the four sticks. Start the PC. Then get into the BIOS, and load the XMP.
Afterwards, still in the BIOS, search for the DRAM VOLTAGE setting and increase it by +0.05v (stock value from XMP should be 1.35v, so tweaked voltage value should be 1.40v).
This very slight increase for the RAM voltage is usually recommended when running four sticks (wether of same capacity or not), and it's completely harmless.

SAVE and EXIT the BIOS.
Boot the PC to Windows, do some tests (DCS or whatever you feel is right to stress test it), and come here again later to tell us the story of how it went....  

Edited October 13, 2022 by LucShep

[fab.13]

Thanks

 

Edited October 14, 2022 by fab.13

[fab.13]

So i have tried,
Motherborad, XMP 2 is ok about the 48go, settings are the same , windows run fine.

But when i run DCS, it crash after  5 / 10 minutes of flight    ( repeatable on different mission that i'm used to fly )

I suspect an overheat,

because the 4 slots are very near from each other,
No clean air can be moving around the DDR4 modules

I dont understand why ASUS slot are so tricky !  ( see attached picture ..)

Is my opinion about the crash could be right ?

I think only one 64go dual kit will be better solution to increase ram on my machine and about fresh air...
 

Edited October 14, 2022 by fab.13

[LucShep]

6 hours ago, fab.13 said:

So i have tried,
Motherborad, XMP 2 is ok about the 48go, settings are the same , windows run fine.

But when i run DCS, it crash after  5 / 10 minutes of flight    ( repeatable on different mission that i'm used to fly )

I suspect an overheat,

because the 4 slots are very near from each other,
No clean air can be moving around the DDR4 modules

I dont understand why ASUS slot are so tricky !  ( see attached picture ..)

Is my opinion about the crash could be right ?

I think only one 64go dual kit will be better solution to increase ram on my machine and about fresh air...
 

 

No, that's not heat.

I'm seeing "TridentZ" and "Royal TridentZ" mixed there. That's not really an exact model match, as you suggested in your previous post.
While same manufacturer and (apparently according to you) the same speed and timings, the components and XMP settings (sub-timings and etc) can vary wildly between the two models, and that could very well be why it's crashing when pushed.
The problem is that you're trying to run unmatched dual kits of RAM with a certain XMP setting of one in the other.

I still think that "32 + 16" mix of different model RAM that you got there can work fine, but it's not a "plug n' play" one click button solution.

You can try the following in the BIOS, for the memory settings:

• Instead of using XMP settings, use "Auto" settings for the RAM.
   
• Once that's done, manually insert the speed (3200 was it?)* 
  * I have no idea what specific memory models you have there, so you have to figure out what's their speed and timings
   
• After that, manually insert the main timings (usually in memory advanced settings), which are tCL, tRCD, tRP and tRAS (16-16-16-32 was it?)* 
  Leave all the other sub-timings on Auto.
  * again, no idea what specific memory models you have there, so you have to figure out what's their speed and timings
   
• Increase the DRAM VOLTAGE to 1.40v (a small +0.05v increase to what you'd see with XMP).
  This +0.05v increase for the RAM is strongly recommended for stability with four sticks, and it's completely harmless. No overheating, nothing, no problem whatsoever.

If all that fails, then you can try reducing the RAM speed, one step below (for example, if it's 3200 reduce to 3000).
It'll almost certainly work fine.

And yes, using a 64GB dual kit (2x 32GB), or adding a second dual kit of 2x 16GB (as long as the same exact model that you have), are better solutions.

Edited October 15, 2022 by LucShep

[fab.13]

Thanks

I could not get the exact same model on store, but timing are the same : dual 3200cl14 14 14 34 on the paper and cpu-z ( trident-z for the booth kit )

So i m going to set auto + settings and give a try as you describe

 

my question : dram overheat not exist as a potential issue ?

Edited October 15, 2022 by fab.13

[BitMaster]

DRAM overheat can occur when certain criterea meet together.

4 modules + little fan power + a GPU tha dumps a lot of heat + WORST, the GPU infrared heat radiation that you can't fight properly with air alone.

Buildzoid has a nice video about that issue. 

 

If your RAM or board reports DRAM temps, mine for example does, you can measure the temps at different loads. Mine can get pretty warm if I stress them but they don't fail. 

As a rule of thumb, don't let them go much over 60°C.

 

Your case is a perfect show-case for others asking RAM questions: Adding RAM is not as easy as many think and should be your 2nd choice only.

Edited October 15, 2022 by BitMaster

[fab.13]

Hello

So i tried.

- manual settings in BIOS : Windows failed to boot and machine restart
- I reverted to XMP II in BIOS, and play with the first 32 go kit only :
         DIM sensor show 58° Celsius max on a long VR DCS mission at 90hz ( my machine is i5 10600k + 1080ti + Rift cv1 )
- I decide to remove one side panel of my rig : 48° max
- I added the second 16go dual kit ( nearly same specs Trident-Z but the size ) : 48°, no crash
- I put back the missing side panel : 58°, no crash !

=> So there is no crash more ever with the both kit ( 48 go )

I tried many mission
I connected to multiplayer syria server where global memory need could be 36 go ( switching slots.. ) : everything is Ok ( and quick to switch plane )
I don't understand, maybe i met a pure DCS bug on the few missions i tried first to test memory ( there are some script not of mine )

I'm going to do more test to confirm that it is now stable,
maybe i'll remove one panel side in summer because the machine was really quiet

PS:

I used a very good tool to inspect sensor : "SIV - System Information Viewer"  http://rh-software.com/
gskill 3200cl14D have temps sensor , not all gskill kit have.
Thanks to both of you, i have been learning more about DDR4 for now ( 60° target limit about DDR4... )

[fab.13]

Hello,

Follow of my history

I succeded DCS to crash again after a mission about 45 minutes.

We played in VR,
and the temperature has been very high i think ( GPU of the 1080ti was near 95% because of 90hz )
SIV show that some module has been over 61° ( see screen copy )

Whatever DDR setting i have, i learned that going over 60° is not good for the modules.

Maybe I should run a MEMTEST to confirm may memory configuration is not the problem
It shoud have crashed before than 45 minutes of flight.

The next test will be to open a side panel and try to reproduce a crash and get temperature scan running.

And a solution is about a better fresh air on my rig ( or only one Dual kit ) :
- adding fan to extract more air
- side panel with a grid to enter fresh air...

[LucShep]

Have you tried to reduce the RAM speed, a step below (for example, as it's 3200 reduce to 3000)? That can help immensely with stability, regardless.

I thought initially that you had a problem of mismatched timings and/or of voltage, but it can be a temperature (overheat) problem.
Even if it was the two different dual-kits conflicting, I agree that 45 mins of flight in MP server (with high RAM consumption) should have shown problems long before that (IMO).

Fast low latency mem such as B-Die (like you have there) can be sensitive to temperature. But, have to say, it does look a little odd to me to see ~60ºC (Max) with just XMP loaded. (....is DRAM voltage at 1.35v or 1.40v, not more? ....maybe restrictive case with very little airflow? ...something's definitely playing up there)

If you look around (google something like "B-die temp stability", or related subject) you'll see a few people running them fine at mid 50ºC, but most can not run it stable at all if above 45ºC.  That said, this temp problem usually applies to people overclocking and running tweaked (tighter) timings... you're not really pushing that memory, as you're just using XMP settings (i.e, "stock") with fairly loose timings.

Right now, see if you can mount a 120mm or 140mm fan pointing air directly to that RAM (maybe using some contraption with zip-ties, like this fella here) and see if it helps.
Maybe not urgent, but I'd recommend to invest in good airflow for that system before upgrading anything else. 

Edited October 21, 2022 by LucShep

[BitMaster]

It highly depends on the board's circuits and the CPU's integrated memory controller, lesser a RAM concern itself.

I have 2 kits of 4-module B-Die and they run great, IN THE RIGHT BOARD.

If you are unlucky, not much will make them stable but reduce MHz, 3000 is a safe value in my experience.

 

For example:  My 3600-16-16-165-36 32GB/4x8 kit ran up to ~4000MHz in my 7700k Z270 combo, but it only runs w/o BSOD at 3000MHz with my 8700k Z370 combo, no matter the Volts. If I take 2 modules out of the Z370 they again run like hell, just not all 4 in that particular board. Temps were never an issue.

With my 5900X rig I cut RAM corners so to say and hoped to compensate it with luck & skill. I bought 3200-14-14-14-34 and oc'ed them to 3600-14-14-14-34 @ 1.45 ( results in 1.48v in HWinfo etc. ).

They can get pretty warm when stressed, around 60ish°C, idle temps are mid 40's°C. Till today they passed all tests I could throw at it, most important, it is stable and does not crash, never did since I own it.

To top it and because I was bored last night I downloaded AMD Ryzen Master, hit the Curve Optimizer and let it run for 90min to tune my voltage curves, raised the PPT TDC  and EDC to north of 250w and ampere and let it run linpack extreme and prime9 with all AVX etc ON, stable... it boosted to 4.6ish GHZ all core under stress, consumed ~200w, peak was 230w, temps on CPU were in the high 80's for loads that fit in Cache but it passed it. I didnt fear the CPU to crash, more the overclocked RAM when the CPU IMC is stressed by OC and the heat from it but it worked out. Was it skill, no, was it LUCK, yes. Unless you dial in the wrong values it's mostly the parts, they either can do it or not and you only find out if you test it.

 

Imhoi t's more like the board or the IMC is playing foul in your bcase