MSI built a desktop motherboard that supports CAMM2 memory

Will be nice to see water blocks that can cover the cpu and memory. Should be pretty easy with this format.

Oh you mean the ram thats 2x as much as normal dimms ... fu msi

maxxcool7421 said:

Oh you mean the ram thats 2x as much as normal dimms ... fu msi

Click to expand...

This is just the standard chicken & egg problem in tech; parts following new standards won't get cheaper until they ramp up production, but obviously you need someone who is willing to pay the higher prices at first to get that production ramped up. If you don't like the idea of paying more for the new RAM style MSI offers plenty of motherboards still using DIMM RAM.

maxxcool7421 said:

Oh you mean the ram thats 2x as much as normal dimms ... fu msi

Click to expand...

If it becomes popular, it will be as cheap as the old format

Dreadcthulhu said:

Its really exciting on laptops - you know how many laptops these days have soldiered memory, since it is faster, uses less power, and takes up less space than SO-DIMMS? Well CAMM units are nearly as fast & low power as soldiered memory, also fit into slim confines, while still being easy to replace. For desktop use in their current form they are less exciting, but still should allow for higher memory bandwidth, and make it easier to fit in big CPU coolers.

Click to expand...

These are good points. For laptops and ultra-mobile there great benefit to be had. For desktops though? Not so much. What MSI is trying to do here remains a mystery.

bviktor said:

It would've helped to explain why I should care about CAMM2 which I only first hear about now.

Click to expand...

https://www.techspot.com/news/96668-camm-memory-could-successor-dimm.html

Dell revealed CAMM over a year ago. Most OEMs and JEDEC liked CAMM so they made some slight tweaks and ratified it as 2 official standards (CAMM2, LPCAMM2). (SO)DIMMs are clunky, 90's tech. DIMMs are bulky and suffer speed limitations due to the difficulty in running equal-length traces for them. CAMM addresses these problems and also has the added benefit of always saturating the memory bus (no more single/dual channel confusion). Hopefully it will be implemented quickly and laptop OEMs decide that a CAMM connector is more cost-effective than LPDDR5/6. CAMM2 could also enable desktop motherboards to mount memory on the backside of the motherboard. In fact it may be that DDR6 marks the death of the DIMM.

The only question is, will CAMM2 prevail or will the new trend of on-package DRAM as seen by Apple and now Intel become the standard?

https://www.techspot.com/news/100904-lunar-lake-mx-leak-suggests-intel-serious-about.html

I don't like idea of CAMM2 in desktops. Example if you want upgrade from 32GB to 64GB you need to buy entire new 64GB module instead additional 32GB traditional DDRs.

bviktor said:

It would've helped to explain why I should care about CAMM2 which I only first hear about now.

Click to expand...

It also is way more efficient than regular dimm modulars. In laptops they showed how the trace to the cpu is way less than traditional ram and it can compete against the efficiency of arm processors without the ram being built inside the chipset and therefore not replaceable.

r00lz said:

I don't like idea of CAMM2 in desktops. Example if you want upgrade from 32GB to 64GB you need to buy entire new 64GB module instead additional 32GB traditional DDRs.

Click to expand...

Not strictly true.

One of the possibilities of CAMM is you can mount it on the rear of the motherboard, right behind the CPU to keep the traces as short as possible. In this configuration, you could have two CAMM slots.

Even then, how many people do you know who buy 1 dual-channel kit, and then decide to upgrade in a short enough time span that they are able to get their hands on a second, identical-model kit? Very few, I bet. Seems the more usual trajectory for most folks is to either: A) buy 1x dual-channel kit, then a year or two later decide to upgrade by buying a larger dual channel kit or 2x new dual channel kits; or B) start off with 2x dual channel kits, or a 1x matched kit of 4x sticks.

mbrowne5061 said:

Even then, how many people do you know who buy 1 dual-channel kit, and then decide to upgrade in a short enough time span that they are able to get their hands on a second, identical-model kit?

Click to expand...

Anyone who is building on a budget, can only afford one set at first and then adds the second a few months or a year later. Happens all the time.

mbrowne5061 said:

Very few, I bet.

Click to expand...

You would lose that bet.

mbrowne5061 said:

Seems the more usual trajectory for most folks is to either: A) buy 1x dual-channel kit, then a year or two later decide to upgrade by buying a larger dual channel kit or 2x new dual channel kits; or B) start off with 2x dual channel kits, or a 1x matched kit of 4x sticks.

Click to expand...

And that does happen, but not with everyone and certainly not always.

ZedRM said:

Anyone who is building on a budget, can only afford one set at first and then adds the second a few months or a year later. Happens all the time.

You would lose that bet.

And that does happen, but not with everyone and certainly not always.

Click to expand...

I find it increasingly common for people to stick with one kit, especially with how cheap RAM has gotten. It’s also best practice to use as few sticks as possible per channel to alleviate the memory controller.

Either way CAMM2 is likely going to put an end to using multiple modules in one system unless you’re dealing with servers or HEDT.

opckieran said:

I find it increasingly common for people to stick with one kit, especially with how cheap RAM has gotten. It’s also best practice to use as few sticks as possible per channel to alleviate the memory controller.

Click to expand...

Not in the area I live in. People here are constantly installing a second set. Most are upgrading from 8GB or 16GB. Some add a 16GB set but some get a 32GB kit. 40GB & 48GB is getting to be common configs as people are adding 32GB kits to their existing 8GB or 16GB set.

opckieran said:

Either way CAMM2 is likely going to put an end to using multiple modules in one system unless you’re dealing with servers or HEDT.

Click to expand...

Not really. Most people like having options and flexibility. The CAMM standard will likely be good for laptops, but desktops are not going to be well received.

ZedRM said:

Not really. Most people like having options and flexibility. The CAMM standard will likely be good for laptops, but desktops are not going to be well received.

Click to expand...

Again, that’s assuming DIMMs even EXIST after DDR5. Desktops are no different in terms of the challenges DIMMs must overcome compared to CAMM2, but since there’s more space on a desktop board, those challenges aren’t nearly as exacerbated as they are on a laptop board. Even if there are DDR6 DIMMs, the form factor will be dead by 2030 (or whenever DDR7 is introduced…)

opckieran said:

Again, that’s assuming DIMMs even EXIST after DDR5.

Click to expand...

Yes, I am presuming that. Why? History is an excellent teacher. What does history teach us?

opckieran said:

Desktops are no different in terms of the challenges DIMMs must overcome compared to CAMM2

Click to expand...

That is completely false. Desktop PC's have lots of space for parts and components, by design. Your suggestion does not hold water.

opckieran said:

Even if there are DDR6 DIMMs, the form factor will be dead by 2030 (or whenever DDR7 is introduced…)

Click to expand...

That's you making illogical assumptions. History says otherwise. The DIMM package for memory works perfectly as is and has for decades. The form-factor of the DIMM doesn't need to change drastically, only adapted for the new technology as progress is made.

ZedRM said:

Yes, I am presuming that. Why? History is an excellent teacher. What does history teach us?


That is completely false. Desktop PC's have lots of space for parts and components, by design. Your suggestion does not hold water.


That's you making illogical assumptions. History says otherwise. The DIMM package for memory works perfectly as is and has for decades. The form-factor of the DIMM doesn't need to change drastically, only adapted for the new technology as progress is made.

Click to expand...

Actually… what history shows us is that technology EVOLVES. That’s why computer monitors don’t weigh 50 pounds anymore, for example. That’s why we don’t use slots for CPUs anymore, for example. And why nowadays we use smartphones to do literally 99% of what we did on laptops 10 years ago plus all the smartphone-exclusive things we really can’t do on laptops without bending over backwards.


But if you want to be delusional and ignore how technology has evolved throughout all of history as well as the more recent concerns of the DIMM form factor posed by the hundreds if not thousands of actual engineers across a dozen manufacturers and standards agencies who are far more engrossed in and actually qualified to comment on the rapidly approaching limitations of the old DIMM form factor than a guy stuck in the past, on a tech forum, then be my guest. Just face it. DIMMs are DONE. They’ve hit their wall. They had a good run. But they’re done. And desktops are not far behind, mark my words. (If you’d actually read my above comment in the first place you’d have seen my caveat for desktops not suffering as much as laptops but still likely to face similar issues down the road).

opckieran said:

They’ve hit their wall.

Click to expand...

Kinda like talking to you. Brick wall comes to mind.

And no, DIMMs have NOT reached the end of their usefulness. Not even close. Quit claiming they have. Such a statement is just clueless.

ZedRM said:

Kinda like talking to you. Brick wall comes to mind.

And no, DIMMs have NOT reached the end of their usefulness. Not even close. Quit claiming they have. Such a statement is just clueless.

Click to expand...

What are your thoughts on increasing operation frequencies complicating the task of electrically length-matching memory bus traces in a multilayer CCA?

As memory modules become more sophisticate - on-die error correction, power conversion, increased burst sizes, increase bank sizes - the pressure is to increase pin counts on the DIMMs themselves, which in turn reduces the contact area at mechanical interfaces and decreases distance between traces below ideal minimums, both of which reduce signal quality. What strategies would you propose to either consolidate/reduce pins and/or improve signal integrity in an electrically noisy environment?

mbrowne5061 said:

What are your thoughts on increasing operation frequencies complicating the task of electrically length-matching memory bus traces in a multilayer CCA?

Click to expand...

My thoughts are simple: CAMM/CAMM2 will not solve any problems. It will only create problems and will make user access and upgrade-ability a serious problem.

mbrowne5061 said:

As memory modules become more sophisticate - on-die error correction, power conversion, increased burst sizes, increase bank sizes - the pressure is to increase pin counts on the DIMMs themselves, which in turn reduces the contact area at mechanical interfaces and decreases distance between traces below ideal minimums, both of which reduce signal quality. What strategies would you propose to either consolidate/reduce pins and/or improve signal integrity in an electrically noisy environment?

Click to expand...

So what you're saying with this statement is that you don't understand the subject matter involved? That's what this statement says very clearly.

ZedRM said:

My thoughts are simple: CAMM/CAMM2 will not solve any problems. It will only create problems and will make user access and upgrade-ability a serious problem.

Click to expand...

Which problems will it create? Why will these problems be a net-loss for the systems that use CAMM over DIMM?

ZedRM said:

So what you're saying with this statement is that you don't understand the subject matter involved? That's what this statement says very clearly.

Click to expand...

Then explain it. In detail. Or go troll somewhere else.

mbrowne5061 said:

Which problems will it create?

Click to expand...

Try reviewing my previous statements.

mbrowne5061 said:

Or go troll somewhere else.

Click to expand...

I was involved in this conversation before you joined in. YOU are trolling me.

mbrowne5061 said:

Then explain it. In detail.

Click to expand...

No.