SoylentNews is people
SoylentNews
CAMM = Compression Attached Memory Module
CAMM to Usurp SO-DIMM Laptop Memory Form Factor Says JEDEC Member
So, farewell, SO-DIMM. After a quarter century of service in laptop, all-in-ones and other compact designs, it looks like the end of the road for SO-DIMM is in sight. JEDEC committee member and Dell Senior Distinguished Engineer, Tom Schnell, told PC World that the new 'CAMM Common Spec' will be the next RAM standard for laptops. There already seems to have been a lot of progress in the background, with the v0.5 spec already approved by 20 or so companies in the task group, and JEDEC expected to finalize the v1.0 spec in the second half of this year.
[...] The new information from PC World editor Gordon Ung's chat with Tom Schnell helps us weigh up some of the pros and cons of CAMM, and point to some ways it has progressed from Dell's pre-JEDEC-approved spec. Apparently, as well as improved density (more RAM capacity in a smaller space), CAMM is amenable to "scaling to ever higher clock speeds." Specifically, the new information indicates that the DDR5-6400 'brick wall' for SO-DIMMs will be shrugged off by CAMMs.
When CAMM reaches devices, there are a couple of tech advances which could help spur its adoption. We mentioned the faster DDR5 speeds above, but it is thought that CAMM could really take off when DDR6 arrives. Another appealing variation might be for adding LPDDR(6) memory to laptops. Traditionally LPDDR memory is soldered, so the new spring contact fitting modules might mean much better upgradability for the thinnest and lightest devices which tend to use LPDDR memory.
Previously: Dell Defends its Controversial New Laptop Memory (CAMM)
(Score: 3, Interesting) by turgid on Wednesday January 18, @10:16PM (3 children)
What sort of compression? Does it to entropy encoding on the data? That will end it tears.
I refuse to engage in a battle of wits with an unarmed opponent [wikipedia.org].
(Score: 4, Interesting) by RS3 on Thursday January 19, @12:26AM (1 child)
You remember seeing the old clothes washing machines with the two compression rollers?
But seriously, it's not data compression, but rather mechanical connector compression, much like CPUs that have the flat bottom with pads, rather than pins. The "socket" (if you can call it that) has little flexible and springy fingers that contact the underside of the module.
I was trying to figure out what problem this solves. Gleaning from various this and that articles, it seems to increase connector density, which helps shorten the wire paths between RAM and CPU. (Shorter paths means faster possible speeds).
Also from looking at the diagrams, DIMM and SODIMM connectors have a "U" shaped connector, which makes parallel contact with both sides of the module. That's electrically bad in that it creates two different length but parallel paths for the signal, which means signal pulses moving through the longer path will meet up with and add to the same pulse that moved through the shorter path, but slightly later, therefore widening the electrical pulses. Wider pulses means lower maximum frequency, or more data errors if you try to push the frequency.
The CAMM socket has only one path, so the pulses stay narrower and cleaner, which means faster speeds are possible.
(Score: 4, Informative) by RS3 on Thursday January 19, @12:29AM
I meant to link to this pic which helps visualize the parallel versus single paths:
https://cdn.mos.cms.futurecdn.net/5htHVduZiMb7hL3w2UBBu7-1200-80.jpg.webp [futurecdn.net]
(Score: 4, Insightful) by bzipitidoo on Thursday January 19, @12:01PM
This, this is why "data compression" is bad terminology. Totally misrepresents what is really going on, giving the impression that storage needs are being reduced by somehow increasing the density of the data. Entropy, not density, is what's being increased. I say it should be called "redundancy removal", or "redundancy reduction". Could also try "irrelevancy removal" for what we currently call "lossy compression".