Samsung just showed off LPCAMM which combines some of the advantages of LPDDR5x memory with some of the tech Dell had been pushing with its CAMM memory modules.
Advantages:
Disadvantages:
0 voters
JEDEC is working to adopt CAMM with LPDDR5x support. I’m not sure if this an implementation of that, but if it is a highly support it.
It is as high bandwidth as soldered ram from last year. I suspect that first gen LPDDR5x CAMM modules have been in development for a while so it makes sense that they are a bit behind.
I’m not extremely well versed on the topic, but I thought that Dell’s CAMM tech already targeted the speed and power advantages of LPDDR, and that they were pursuing making it a JEDEC standard. If that’s the case, then what advantages does this Samsung LPCAMM offer?
After skimming the article, it reads like the Dell/almost-JEDEC implementation isn’t quite as good as it was originally planned to be - as if the signal paths just aren’t ideal. That’s disappointing if true.
LPCAMM is not proprietary. JEDEC adopts it. Or at least JEDEC CAMM is adopted by JEDEC.
CAMM is a proposal put forth my a member (Dell) which has yet to be ratified. As such it is still proprietary
LPCAMM is a new completely different proposal by a member (Samsung) which also has yet to be ratified.
My apologies. The article I was referencing seems to be incorrect when it said
While the CAMM format has barely gotten off of the ground itself – JEDEC has yet to even approve the standard – Samsung is essentially opting to run with the idea to do something different with it by focusing on LPDDR memory.
Final edit: reading through your reference there isn’t any mention of ratification. Just that the proposal now includes ddr5 modules. I didn’t find anything after a quick search about ratification either. And now there is a competing standard with clear upsides, CAMM might be abandoned for LPCAMM.
CAMM and LPCAMM should have the same pinout.
Nevermind:
While the CAMM format has barely gotten off of the ground itself – JEDEC has yet to even approve the standard – Samsung is essentially opting to run with the idea to do something different with it by focusing on LPDDR memory. Notably, however, the resulting LPCAMM form factor is incompatible with CAMMs – both physically and electrically – so despite the similar names and use of compression connectors, the two are not interchangeable. But both pursue the same ideas for their respective memory types.
Seems to be way to much fragmentation happening right now. I’d prefer for Framework to switch only if they can be confident in a standards long life.
It’s interesting because the similar loooking Adata module that Anandtech mentions in that article wasn’t labelled as a LPCAMM, but as a regular CAMM module. Since the JEDEC 1.0 spec isn’t finalised, I wonder if LPCAMMs are an evolution of the 0.5 CAMM spec that arose due to connector changes to accomodate LPDDR5… Anyways, we’ll know for sure when the final JEDEC spec comes out later this year.
I really do hope CAMM (or some other standard related to it) gets off the ground and replaces SODIMMs though! The enthusiastic reception CAMM got when it passed its 0.5 spec is a pretty good indicator that JEDEC and industry partners are as excited about CAMM as I am 
In fact, Schnell said, the acceptance went over quite well with the 20 companies or so in the task group voting for it.
“We have unanimous approval of the 0.5 spec,” Schnell told PCWorld. Schnell said JEDEC is targeting the second half of the 2023 to finalize the 1.0 spec, with CAMM-based systems out by next year.
IMO, it’s isn’t about if JEDEC will adopt some form of CAMM, it’s when. Not only because it has gotten a good reception from JEDEC and industry partners, but because SODIMM is gonna hit its performance ceilings really quickly as we progress through the DDR5 generation… SODIMM DDR6 memory might not even be possible!
Edit: I think it’s kind of odd that the only article I could find referencing the difference between LPCAMM and CAMM was the Anandtech article. The Samsung newsroom press release didn’t mention anything about it, so I wonder if maybe the author misinterpreted something or Samsung clarified to them off-record. I’d like to see Samsung themselves publicly address this before I draw any conclusions.
I agree, but I also wouldn’t expect Framework to even consider something like this until at least DDR6. They stuck with DDR4 for the 12th and 13th gen Intel laptops for greater backwards compatibility (ease of upgrading), so I’d expect them to continue using DDR5 SODIMMs for the same reason even if DDR5 CAMM suddenly becomes feasible and widespread.
By the time DDR6 starts to become not only a available but required for new CPUs, I imagine the merits and standardization of SODIMM vs. CAMM will be clearer.
Before the JDEC standard is ratified and there are at least a couple manufacturers making compatible modules there isn’t really a point in using it jet, but once that is covered there is pretty much no downside.
The current sodimm form factor has so many drawbacks, which are completely worth it (to me at least, turns out not to most manufacturers) to gain socket-able memory as long as there is no alternative but once there is it should definitely be put to rest. I actually wonder if something like camm showing up earlier may have slowed down the soldered memory trend a little.
I agree! If all the DRAM manufacturers like Micron, SK Hynix, and Samsung get on board with this I can see it getting big really quickly, I bet laptop manufacturers are itching to ditch SODIMMs.
As for the timeline I’m pretty optimistic, but it will definitely hinge on whether the JEDEC standard is released on schedule i.e. by the end of 2023. My optimistic prediction is that we’ll see the standard finalised in 2023 and a lot of early looks at consumer CAMMs at CES 2024.
Oh, and I’ve seen Dell use SODIMM interposers to maintain compatibility even with mainboards that use CAMMs, so that might be an option for Framework to adopt CAMM but still maintain backwards compatability with DDR5 SODIMMs… assuming the height works out (I’m not sure on that one).
I really hope one of the CAMM standards becomes common enough at some point. Because that is the only downside that I can see. Once it becomes a proper standard, and available in decent quantities, it wouldn’t have to be much more expensive that SODIMM.
It might even replace SODIMM altogether, as SODIMM is becoming less common over time anyway as more laptops are moving to soldered memory. Maybe a good CAMM “market share” can even prevent some laptop designs from moving to soldered memory.
But it’ll take some time. Maybe it’ll happen when DDR6 is around. Maybe it could replace SODIMMS entirely at that point?
A quote from the article:
" As for standardization, according to Samsung they are working with partners to get a JEDEC standard for LPCAMMs. All the while, the JEDEC announced back in March that they were also working on extending the CAMM standard to cover LPDDR memory, using the same connector for both DDR5 and LPDDR5. At this time we’ve heard no further information from the group, and absent a ratified CAMM standard, it’s unclear if JEDEC is still moving forward with their LPDDR proposal. Ultimately, we may yet see rival CAMM-style standards for LPDDR memory, depending on how standardization shakes out."
It does look like the “proprietary” negative point is actively being worked on being adressed.
I’m extremely hopeful this tech will find adoption in consumer products, the energy efficiency of LPDDR is one of the (many) reasons the macbooks are ahead of the curve when it comes to battery life, anything that can help on that front while not sacrificing upgradability is a win in my book!
Additionally, CAMMs seem to be the solution for the limitations of SODIMMs we will be hitting soon.
Let’s just hope the standartization happens soon, and one one single connector, whichever one it ends up being.
Looks like both reg DDR5 and LPDDR5 were approved by Jedec however they will have different mounts to stop accidental mismatches.