Next Crop of MSI Project Zero Motherboards to Implement CAMM2 DDR5 Memory

[TheLostSwede]

Sorry. You're right. They already did something stupid with the standard.

It's a technical limitation of using different types of memory.

Do you want more stupid?

 

[AGlezB]

It's a technical limitation of using different types of memory.

Do you want more stupid?

View attachment 348594

Why is my nose bleeding and my left eye twiching?

 

[seventy]

And again two-brain-cells commenters are hating on an actually good idea.
In general, PC performance is limited by the amount of heat you can comfortably dissipate, assuming you're not limited by power.
If you ever overclocked sams b-die, you would have known that that kind of stuff is exactly what you want. If I could easy attach a cpu cooler to my memory, I would have actually done it.
Slow access to memory is number one case of frame stutter in games. That's the reason X3D cpu's are so good for gaming, they alleviate the slow speed of main mem.

 

[TheLostSwede]

@AGlezB only the right most one can be stacked from what I understand, but it's also likely to be the least common type.

 

[Dristun]

Sensing a lot of "old good, new bad" vibes. Let's see if this brings better speeds, oc, memory cooling and/or something else and then judge. Hey, maybe one day they'll gather the confidence to come together and replace ATX too. Little by little!

 

[TheLostSwede]

Sensing a lot of "old good, new bad" vibes. Let's see if this brings better speeds, oc, memory cooling and/or something else and then judge. Hey, maybe one day they'll gather the confidence to come together and replace ATX too. Little by little!

But it doesn't change the flaws of the ATX form factor though, which is the issue. Also, Asus, Gigabyte and MSI have competing, incompatible designs for this, which is just dumb.

 

[Dristun]

But it doesn't change the flaws of the ATX form factor though, which is the issue. Also, Asus, Gigabyte and MSI have competing, incompatible designs for this, which is just dumb.

Admittedly, my view on the matter is silly, but I do genuinely believe that if they at least keep tinkering away and trying dumb stuff then maybe eventually someone hits the jackpot and a new design takes off. If nobody keeps trying anything then we will be stuck with what we have. It's definitely better than nothing, in my opinion.

 

[LabRat 891]

But it doesn't change the flaws of the ATX form factor though, which is the issue. Also, Asus, Gigabyte and MSI have competing, incompatible designs for this, which is just dumb.

It's only been around for what? 29-30 years?

I'm a big proponent of 'don't mess with standards that work.' but... I have to agree. At this point though, it'd be a hellova shakeup to the industry to try and change that.
Years ago, Intel tried w/ BTX, (AFAIK) pretty much only Dell was fully on-boarded.

 

[TheLostSwede]

Admittedly, my view on the matter is silly, but I do genuinely believe that if they at least keep tinkering away and trying dumb stuff then maybe eventually someone hits the jackpot and a new design takes off. If nobody keeps trying anything then we will be stuck with what we have. It's definitely better than nothing, in my opinion.

As long as the major motherboard players can't agree on a common standard, it's never going to take off though.
And I can tell you that they don't even discuss this stuff between themselves, as that would be to give in to the competition according to the Taiwanese mentality on this kind of stuff.
Trust me, I have tried to talk to these companies.

It's only been around for what? 29-30 years?

I'm a big proponent of 'don't mess with standards that work.' but... I have to agree. At this point though, it'd be a hellova shakeup to the industry to try and change that.
Years ago, Intel tried w/ BTX, (AFAIK) pretty much only Dell was fully on-boarded.

Shuttle as well, plus some others. The issue wasn't the BTX form factor, but rather the size of the coolers that killed it, oh and the fact that it was for OEM/ODMs only, not DIY.

Threw this up as well, for those that wants some more CAMM2/LPCAMM2 action.

LPDDR6 LPCAMM2 Pictured and Detailed Courtesy of JEDEC

Yesterday we reported on DDR6 memory hitting new heights of performance and it looks like LPDDR6 will follow suit, at least based on details in a JEDEC presentation. LPDDR6 will just like LPDDR5 be available as solder down memory, but it will also be available in a new LPCAMM2 module. The bus...

www.techpowerup.com

 

[ir_cow]

Yet it takes up the same real-estate...

 

[dgianstefani]

Yet it takes up the same real-estate...

Same horizontal real estate, much less vertical, I'm not sure if this enough of an improvement to entirely replace DIMM, but it's seems at least slightly better, since better signal integrity etc. Wonder if we'll see this used with those DDR5 redrivers etc.

 

[n-ster]

Desktop hyper-standardization was great until now, but it does feel clunky and inneficient nowadays. CAMM2 may be the best way to keep upgradeability without tempting manufacturers into SoC designs, but I have to agree, having 2 module slots to prevent e-waste and have upgreadability seems like it would be nice, would the cons really outweigh the pros? Now if we could also get the industry to adopt ATX12VO v2.1 and use an appropriate form-factor for high-performance or high-capacity SSDs, that would be the dream

 

[Papusan]

Nice. We already have mobile processors with phone cores as desktop processors. Now we will get laptop memory modules ported over to desktops as well. Whats next? More restictive firmware with no options in same as for locked down Jokebooks/tablets? Jup, the modern tech trend continue. More and more disgusting for each new day.

 

[Wirko]

Well, not cabling, but if you want to slap a custom loop on it, now there's the opportunity for the monoblock to extend on the RAM itself. Needlessly watercooling the standard DIMMs is quite hard to do in an effective way. Most of the cooling solutions have a narrow crossflow for heat unlike the old Koolance "baggy" coolers which had the water "touch" the chips itself via an elastic bag-like thing that separated the water.

All DDR5 CAMM variants appear to be double-sided. (CXXX has 4 chips on either side, AXXX has 8, and BXXX and DXXX have 16.) Unless there's a hole in the mobo, it's going to be really hard to cool the bottom chips.

 

[rv8000]

Same horizontal real estate, much less vertical, I'm not sure if this enough of an improvement to entirely replace DIMM, but it's seems at least slightly better, since better signal integrity etc. Wonder if we'll see this used with those DDR5 redrivers etc.

Based on the MSI image in the article the majority of that seems questionable. The CAM module itself looks wider than 4 DIMM slots with respect to horizontal real esate.

The location of ICs and designation of which/what area of the module is which “channel” seems like trace length to each individual ic on each channel will have a higher variation; that can’t be good for signal integrity.

 

[TechLurker]

It would be less of a novelty if the CAMMs could be stacked on; maybe 1 on the bottom and 3 on top, or redesigning the layout to allow two on the bottom, closer to the CPU, and 2 on top in the same positions of the bottom CAMMs, akin to how some Quad-Channel DDR are placed on both sides of the CPU socket.

Then partner with Noctua or Cooler Master to get out a larger down-blowing cooler that can cool the VRMs, VRM heatsinks, and upper CAMMs in addition to cooling the CPU, while also having a larger heatsink. Depending on the size of the theoretical mobo and the CAMM types chosen, it might even be possible to make a 240mm long down-blowing cooler that accommodates 2 120mm fans.

 

[Vayra86]

Well at least MSI also realizes the net gain from this Project is going to be exactly Zero.

Gotta applaud the realism

 

[Caring1]

I recall an adapter that allowed the use of laptop Dimms in a PC Dimm slot, perhaps they could have made it so the sodimms could be removed from the board and added to or swapped.

 

[Kohl Baas]

Watercooling RAM is nonsensical for regular PC use. It doesn’t get that hot. And for insane OCs there is LN2.

Have I forgot to say "needlessly"? I think not...

All DDR5 CAMM variants appear to be double-sided. (CXXX has 4 chips on either side, AXXX has 8, and BXXX and DXXX have 16.) Unless there's a hole in the mobo, it's going to be really hard to cool the bottom chips.

Too bad...

 

[Random_User]

All DDR5 CAMM variants appear to be double-sided. (CXXX has 4 chips on either side, AXXX has 8, and BXXX and DXXX have 16.) Unless there's a hole in the mobo, it's going to be really hard to cool the bottom chips.

Exactly. I'm all for advancement in ancient motherboard and RAM technologies. But the cooling aspect is what worries me. And not only the OC side, but the overall airflow thing. It's not a big problem with tower coolers, like NH-D15, where the middle fan grabs and pulls the hot air from the RAM. But what about this?
Despite how ironic and strange this sounds, the CAMM2 modules have more luck to being cooled in laptops, both due and despite the space constrains.