Framework 16 with CAMMs (Compression Attached Memory Module)

This would be huge for Framework 16 to support in the future, and I hope the Framework team designed the FW16 with this in mind. CAMMs will support LPDDR5 and higher speeds than it’s currently possible with SODIMMs. This open standard was bought to the JEDEC by Dell and then accepted by the JEDEC, so it’s happening.
PCWorld CAMM Video (Shows Form Factor)


JEDEC CAMM isn’t finalized and on the market yet.


Will the 16 support CAMM?

Currently? No.
The Framework-16 that is up for pre-order uses traditional sodimm RAM.

Also, I suggest searching

1 Like

For what it’s worth, I’d like to hear Framework R&D team’s thoughts on CAMM before fully endorsing it.

It’s clear something must come after the SODIMM form factor. But is CAMM the right option? Dell thinks so. But I wonder what Framework (with an eye on right to repair, etc) thinks.

I wonder if Framework is talking to the JEDEC working group to ensure any concerns they have are addressed. Be it directly or indirectly.

1 Like

If CAMM becomes more mainstream, I guess there wouldn’t be any reason for Framework to NOT use it. Much like how architecture changes dictate the need for different memory types (ex: AMD CPU requiring DDR5), I can see them switching out to CAMM once it becomes more ubiquitous, AND the architecture supports or require it.

I say “or” because we know 13th gen Intel supports DDR5 but they chose DDR4 to keep the upgrade path for the previous gen owners the same for memory (so they don’t have to buy DDR5 memory to upgrade). AMD 7040 series CPU doesn’t support DDR4, so here we are with DDR5.

So I’m sure it depends on where the architecture is heading towards rather than Framework requiring it or making it an option (since more options = more SKUs, which is not that feasible with a startup).


Maybe next gen, or the one after. Who knows.

Shouldn’t expect a product that’s months from release to support an un-released memory standard, unless the OEM are closely involved in that standard.

Cool though. SODIMM’s old. Always something else new & exciting being developed. Like that Direct Storage stuff that’s now starting to become a thing; expecting to see PCIE gen5 storage in later FW mainboards for sure to better allow that.


CAMM is pretty new, but it’s frankly kind of odd why has nobody thought of this prior – even DIMMs are getting to the point where signal integrity is a real thing, let alone SODIMMs.

The only “problem” I see is that CAMMs are single-layered, while you can double-stack SODIMM slots. But that doesn’t prevent anyone from making it slightly taller and sneaking other components under it, such as the side-connected m.2 slots we all crave for. Or sound cards. Whatever.
You can even get CAMM to SODIMM adapters, which while as redneck as it sounds, could be a stop-gap solution.

Look at that massive space taken up by the SODIMM slots on the motherboard – it even have massive gap in the middle for the traces to pass through. How inefficient (it’s just the way board layout works, there’s no getting around). With CAMM, the area taken up is unlikely to change, but I can see m.2s (or other system resources) being able to find home under that.

JEDEC is planning on releasing a version (of the standard) sometime this year, so I’m not too worried about that. It’s not there yet, but I can almost smell it, and I am certainly excited for that.

But I still think that FW 16 isn’t as radical as it could be – it still has 1 layer of RAM, and while it’s double stacking SSDs, you aren’t able to put in two full fat sticks, and one of them is a 2230. Making the machine 3mm taller would likely solve the problem, but here we are.
Imperfect. But getting there.

The biggest deal imo is the ability to have swappable lpddr which does mostly remove most of the “hurr durr soldering the memory is absolutely nessecary for efficiency” line of argument.


I hope CAMM is ready by next-gen and Framework adopts it as well!


I hope it is done by the time I need a new one XD. Socketable lpddr would be so freaking nice.

1 Like

Then I hope you have a relatively recent gen platform, because we know how it usually goes. With DDR5 just recently on current gen, I can see CAMMs being further out - at minimum 2 generations maybe?

I have the amd 13 pre-ordered and I hope that’ll last me at the very least 3 generations.

Maybe, though I can see it becoming a thing pretty quick once the jdec standard gets finished. Since it’s the same chips on different boards with a new connector the r&d required to get it working looks pretty manageable, especially since the engeneers are allready used to some very adverse conditions from working with sodimms (the minimum trace lenghts the form-factor dictates are just absolutely huge for modern low power/high speed memory).

1 Like