[SOLVED] Corsair dimm compatibility

[dfdf43]

Hi

I've got 32GB ram in my system at the moment, which 2 x 16GB dimms, corsair part number CM4X16GC3200C16K4.
I've got 2 empty slots on the mother board , which supports up to 64GB so I wanted to just buy 2 more of the same dimms but I cant find them anywhere, at least not at a reasonable price, however there is another corsiar part number which appears to have the same specs: CMK32GX4M2E3200C16.

Motherboard is B450M DS3H.

Does anyone know if these dimms can be mixed - or how to determine that?

many thanks

 

[drea.drechsler]

Try to make sure they are of matching construction as much as possible: that is, same mfr./type of DRAM's on the DIMM's (Hynix CJR, or Micron E-die or Samsung B-die for instance). And that they are arranged the same: single rank or double rank, same number of chips on each side of the PWB.

Your current DIMM's should be easy to find out construction details: get a utility called Typhoon Burner and it will tell you a lot. More easily HWInfo also gives information in the Main screens, on the Memory blow-out. Finding out about the new memory is harder, hopefully the shop will let you return and exchange if they prove to be vastly different and incompatible for this.

Otherwise, I think you could use them. Try first with matching DIMM's not occupying the same channel sockets. So one kit's DIMM's should occupy A1, B1, the next kit's DIMM's occupy A2, B2. If the memory controller initializes and trains dual-channel operation might still work for you.

Shouldn't need saying: reset CMOS before even attempting to boot the first time after installing any new memory or changing their socketing arrangement. That will force the CPU IMC to train memory again and initialize correctly for what it finds.

It will be sub-optimal. You probably won't reach high clocks...maybe 2666 but maybe even no better than 2400. Setting XMP will highly likely just end up with boot loops (reset CMOS to get out of that). But with 4 DIMM's in that board I wouldn't expect high clocks even if they're all matched. This is a necessary compromise for running huge memory arrays with 4 DIMM's. It should be done only because it's infinitely better than Windows hitting the swapfile when an application needs the memory.

When done even if it boots don't assume it's good. Test first...run Memory Diagnostics in Windows, get MemTest and/or MemTest86.

 

[dfdf43]

thanks for the reply.

So I have now installed the RAM and it appears to work fine.
Just one note that might be of interest, it works fine with XMP disabled. If I enable XMP then it acutally boots fine and looks like all is working but if I leave memtest running for say 24 hours so that it completes 4 or 5 passes it will log an error occasionally i.e. most passes do not log any errors but every 2 or 3 passes will log an error.
However with XMP disabled no errors get logged so I'll just live without XMP

thanks

 

[drea.drechsler]

...
However with XMP disabled no errors get logged so I'll just live without XMP

thanks

Even an occasional error can be catastrophic if it quietly corrupts your Windows installation...or ruins work that you do. That's why testing is important.

But since it's so infrequent it probably is very close and might could be fixed with a few tweaks to memory timings or raising DIMM voltage a bit higher (if it's 1.35V try 1.38V or even 1.40V). But it also means more long tests to make sure no errors pop up.

 

[dfdf43]

thanks
I've run it got 24hours XMP disabled as well and no errors logged there so should be fine.
I'm using the machine at the moment but this weekend I should be able to just leave memtest running friday evening to monday morning, so assuming still no errors logged then (I'm just leaving XMP disabled now) then I'll consider it a victory

 

[drea.drechsler]

thanks
I've run it got 24hours XMP disabled as well and no errors logged there so should be fine.
I'm using the machine at the moment but this weekend I should be able to just leave memtest running friday evening to monday morning, so assuming still no errors logged then (I'm just leaving XMP disabled now) then I'll consider it a victory

If your work was memory constrained and constantly hitting the swapfile it would most definitely be a victory even at lowest DDR4 clocks.