Tuesday, April 26th 2022

Dell's DDR5 CAMM Appears in More Detail, Comes in Several Shapes, Won't be Proprietary

Last week the first details of Dell's CAMM (Compression Attached Memory Module) made an early appearance courtesy of a product leak, but now official details have appeared and the good news is that Dell is saying it won't be a proprietary solution. The Compression Connector looks unlike anything used by consumer computers today and Dell is said to be hoping that it'll be the next industry standard for memory modules, according to PCWorld. The interposer mentioned in the previous news article is also mentioned and allows for a pair of DDR5 SO-DIMMs to be used, albeit with a much taller Z-height.

Dell is apparently planning on getting its CAMM approved by the JEDEC, which is the standards organisation when it comes to memory. However, even if the CAMM format is accepted as a JEDEC standard, Dell holds patents and is likely to charge some kind of royalty fees to interested parties. That said, if it becomes a JEDEC standard, Dell has to follow RAND or Reasonable and Non-Discretionary terms, so the royalty fees would have to be reasonable for JEDEC to agree on making CAMM a standard. The main benefit of Dell's CAMM is that the memory traces end up being shorter and more direct, since the CAMM has a single-sided interface, whereas SO-DIMMs are interfaced on both sides, just like standard DIMMs. This would allow for higher speed memory interfaces, without the need of using something like signal re-drivers or re-timers.
Dell is apparently getting ready for DDR6 already and told PCWorld that by the time DDR6 arrives, SO-DIMMs are no longer going to be fit for purpose. Another advantage of the CAMM, is that higher memory speeds can be used in combination with greater densities of RAM, as a single CAMM can host 128 GB of DDR5. The type of connector used is known as DGFF and Dell is already using it in some of its products today, as a bridge connector for GPU boards among other things. Dell claims that the DGFF connector is capable of handling frequencies of up to 20 GHz, or four times the speed of DDR5 memory at 4800 MHz. The physical CAMM will come in different shapes and sizes, but the common part is the compression connector, which is as the name implies, compressed in place with the help of a pair of screws. Additional screws are also used to hold the CAMM securely in place, with the modules shown, using between four to six screws in total.

Dell also claims that the CAMM connector can act as a heatsink and help with heat dissipation, although the company didn't go into any details of exactly how this is meant to work. There will be what Dell calls a bolster plate that sits above the CAMM, most likely to protect it from damage from the screws, but it seems like it could also be extended as part of a heatsink if needed. The bottom bolster is what the top bolster screws into, rather than the compression connector itself. It should be noted that the CAMM doesn't have any kind of pins, so precision installation is going to be key, but judging from the pictures provided there are some kind of guides to make this easier. Each of the 14 interface rows appears to have 44 contacts per, for a total of up to 616 interface contact points. The contact points seem to vary in shape and size depending on their function. A CAMM can be single or double-sided, depending on memory density and Dell has developed CAMM's ranging from 16 to 128 GB. Dell will start shipping computers with CAMMs installed later this quarter.
Sources: PCWorld, VideoCardz
Add your own comment

37 Comments on Dell's DDR5 CAMM Appears in More Detail, Comes in Several Shapes, Won't be Proprietary

#26
dyonoctis
Chrispy_Thinner?

Who cares?! The height of a SODIMM slot isn't the thing that prevents laptops from getting thinner, it's the thickness of heatpipes and their associated cooling fins. Make those too thin and you get machines that thermally throttle with woeful little high-pitched fans that can barely handle a 15W TDP
*cough* Apple *cough* Dell XPS *cough*.

No thanks, laptops are already thin enough and the benefit of having an almost unnoticeably thicker chassis is that the speakers can be a little bit less godawful and the battery can be very slightly larger.

Who the hell is asking for thinner laptops anyway? They're already so thin that they're horrible to type on and way too fragile....
And also, faster. current SoDimms are not really know for groundbreaking speed and tight timing
Chrispy_This is what I'm saying, that's a lie presented to you by manufacturers for why they solder the RAM onto the board. They do it because it's a cost-saving for them, and they can force/persuade buyers into CPU/Screen/SSD upgrades they weren't necessarily interested it by segmenting the larger RAM capacities to those upgraded SKUs. Sadly it's an incredibly anti-consumer practice that has been obscured by easily disproven claims that it's for reducing the height.

I have one of the thinnest 15" laptops on the market, an Envy X360 and it has two SODIMM slots:




Even with an EMI shield over the top, and it being directly under the keyboard deck, The SODIMM slots are still not the limiting factor in the laptop's thickness:
  • The speakers and the battery are both each as tall as the keyboard, motherboard, SODIMMS and EMI shield combined
  • The keyboard deck actually flexes a little on the right because there's no RAM behind like on the left side to fill the void between the fans and the battery.
  • The rear exhausts for the two fans are laughably thin slits because the laptop is no taller than a USB-A and HDMI port at it's thickest point. Even with two fans, this thing is very audible at just 25W cTDP and it spends most of its time at 15W because that's all the fans can handle quietly.
This thing is entirely limited by the thickness of the battery at the front and the already-far-too-slim cooling system at the back. There's loads of empty space in the middle for chunkier components and absolutely no advantage to replacing the SODIMMs with something thinner; All that would do is increase the keyboard flex. I guess in an ideal world there would be an extra M.2 slot here to fill the empty void, or a larger fan and greater fin surface area to decrease the noise when running at its rated 25W TDP.

Looking at it the other way, if there was a hypothetical laptop that was much thinner than this it would have to forgo either cooling or battery. The ancient Asus Zenbook UX305 models used 5-Watt passively cooled Core M CPUs and therefore the battery fit in the thicker end of the tapered chassis where cooling and heatpipes would go in a modern machine. They were basically as thin as the touchpad at the thin end and behind the touchpad there was nothing at all just useless space that was too thin for anything at all but had to be there as a palm rest and to make the laptop deck as long as the screen was tall. As portable as those things were, they were underwhelming in real-world use, fragile, and we retired our whole fleet of those earlier than expected due to lack of upgradeability and low, fully-soldered specs that prevented us from buying higher spec when new.
Yes, but you only have two slots. The new format allows a laptop to have an expendability of 4 slots.
Posted on Reply
#27
TheLostSwede
News Editor
QuitessaSo it's basically LGA for Memory then
Not quite, even though it's somewhat similar.
This is the current bridge connectors that Dell is already using in some of their products, but in multiple sets to connect graphics cards.
Below that is a card with three sets of pads, which is what the CAMMs will have. Note that the graphics card was made in 2018, so this is not something new for Dell.
It's not nearly as complex or as easily damaged as LGA sockets.

Posted on Reply
#28
Valantar
Chrispy_This is what I'm saying, that's a lie presented to you by manufacturers for why they solder the RAM onto the board. They do it because it's a cost-saving for them, and they can force/persuade buyers into CPU/Screen/SSD upgrades they weren't necessarily interested it by segmenting the larger RAM capacities to those upgraded SKUs. Sadly it's an incredibly anti-consumer practice that has been obscured by easily disproven claims that it's for reducing the height.

I have one of the thinnest 15" laptops on the market, an Envy X360 and it has two SODIMM slots:




Even with an EMI shield over the top, and it being directly under the keyboard deck, The SODIMM slots are still not the limiting factor in the laptop's thickness:
  • The speakers and the battery are both each as tall as the keyboard, motherboard, SODIMMS and EMI shield combined
  • The keyboard deck actually flexes a little on the right because there's no RAM behind like on the left side to fill the void between the fans and the battery.
  • The rear exhausts for the two fans are laughably thin slits because the laptop is no taller than a USB-A and HDMI port at it's thickest point. Even with two fans, this thing is very audible at just 25W cTDP and it spends most of its time at 15W because that's all the fans can handle quietly.
This thing is entirely limited by the thickness of the battery at the front and the already-far-too-slim cooling system at the back. There's loads of empty space in the middle for chunkier components and absolutely no advantage to replacing the SODIMMs with something thinner; All that would do is increase the keyboard flex. I guess in an ideal world there would be an extra M.2 slot here to fill the empty void, or a larger fan and greater fin surface area to decrease the noise when running at its rated 25W TDP.

Looking at it the other way, if there was a hypothetical laptop that was much thinner than this it would have to forgo either cooling or battery. The ancient Asus Zenbook UX305 models used 5-Watt passively cooled Core M CPUs and therefore the battery fit in the thicker end of the tapered chassis where cooling and heatpipes would go in a modern machine. They were basically as thin as the touchpad at the thin end and behind the touchpad there was nothing at all just useless space that was too thin for anything at all but had to be there as a palm rest and to make the laptop deck as long as the screen was tall. As portable as those things were, they were underwhelming in real-world use, fragile, and we retired our whole fleet of those earlier than expected due to lack of upgradeability and low, fully-soldered specs that prevented us from buying higher spec when new.
Okay, so, here's the thing: something being difficult does not make it impossible, and connected to this; something being possible does not make it easy. And something being possible does not make it well suited to the task. I've never said that no thin-and-light design can fit SO-DIMMs; I've said that most can't, especially in the 13" class. And no, this isn't a marketing lie, it's an easily proven fact. Heck, your own laptop demonstrates it - look at the massive area taken up by those DIMMs! And that's in a 15" chassis - even if yours does seem to have a ton of barely utilized board space on the left, there's no way that would fit into a 13" chassis unless you also make it thicker, using smaller footprint, taller components stacked more densely.

And that's part of the thing: that being a 15" device inherently makes it easier to make it thin, through allowing for the use of wider, lower profile components (chokes, batteries, fans, etc.). That yours seems to use its SO-DIMMs as a structural component just supports this - apparently there's zero clearance there! So, for example, a thinner memory module could then allow for a better supported bottom plate design with more rigidity. Or it could allow for a larger battery than that relatively paltry 51Wh unit. Or it could allow for the second fan to be the same size as the first.

(Also, it's rather odd for you to seemingly use the structural support offered by the DIMMs as some kind of pro here - that's clearly a poor casing design, neither more nor less, and using a more space efficient memory type wouldn't affect that in either direction. That the laptop does overall seem to have some very odd priorities made in its design - such as that tiny secondary fan - speaks more to a poor overall design than this somehow being related to its use of SO-DIMMs. Also, it's often cheaper to stick a pair of SO-DIMM slots on a board than to design trace routing for a soldered memory layout. Off-the-shelf, high volume components are very cheap.)

(You're also misrepresenting some things a bit here: your battery is stacked with the touchpad, so it's only partially full thickness, and the speakers live in the tapered portion of the chassis, mostly being far thinner than the middle. The chassis also seems to have an overall wedge shape, being slightly thinner at the front, but that might just be an optical illusion from the tapered front edges. But regardless of this, both batteries and speakers can be shaped quite freely to fit whatever space is available. SO-DIMMs can't.)

Heck, you have 13" devices with SO-DIMMs too, like the Framework laptop, but in that you can again see just how ludicrous the board space requirement for dual SO-DIMMs is for such a small device. (And it's greatly helped by using a 3:2 display, giving it just that tad extra area to work with.) That doesn't mean it can't be done, but it comes with severe drawbacks, and it has a strong deterministic effect on the possible design variations. A more spatially efficient design will change those effects, giving more design flexibility.

After all, as that CAMM-to-SO-DIMM carrier board shows us, the CAMMs are barely wider than a SO-DIMM, yet they are inherently dual channel - and they don't need to be as deep (long?) as the full sized designs. Unless you're designing for a memory capacity of 64GB or more (which isn't necessary for the vast majority of consumer products), you can design for the smallest CAMM size, which is barely larger than a single SO-DIMM in X and Y, while being lower in Z.

Just to be clear (which IMO I have been, but apparently repetition is still needed): I have in no way argued for the creation of thinner laptops. I've argued that this can allow for more densely packed devices with more efficient internal layouts without sacrificing repairability and upgradeability. SO-DIMMs due to their size and Z-height have a strong determining effect on where they can be placed and how they fit, and even if they hang off a motherboard edge they require a lot of vertical clearance to fit as they're nowhere near in line with the motherboard. Doing the same with this CAMM would be far more spatially efficient, and allows for more flexible implementations overall. You can still make the design thicker if you want to, and if you do, you have more space for structural support, larger batteries, larger (or more) speakers, better keyboards, etc. Or you can go thin without needing to resort to soldered memory. This can allow for less compromised designs. And that's the important part. Sure, it can also be used for ludicrously thin, bad designs - but if that's the way an OEM wants to go, they can already do that with soldered RAM. There are no new disadvantages to this design outside of possibly cost and the time needed for standardization and adoption. But there are also plenty of advantages - OEMs just need to make use of them.

Here's another example: my work laptop, a Dell Latitude 7390 2-in-1. It's a pretty bad laptop in many regards, with poor cooling and a CPU configured to a mere 12W. It also has the smallest, thinnest fan I've ever seen (and it still vibrates and touches the casing at times!). Most of this is likely because Dell chose to give this laptop a pretty great keyboard (for its thickness), plus (like with yours) a thick, reinforced display for pen input.

Still, I'd like you to let me know how they could have fit upgradeable memory in here with SO-DIMMs. Because judging by the size of these CAMMs, they could have fit the smallest size of CAMM in here, under the RF shield where its soldered memory lives. (It's also worth noting that the Latitude has a bigger battery than your 15" device!). So: tradeoffs have been made. This is still not a very thin device, mostly thanks to the very thick display. But it's fine in terms of size. If CAMM could allow for designs like this (or ideally better executed ones) with upgradeable memory, that would be amazing.
TheLostSwedeNot quite, even though it's somewhat similar.
This is the current bridge connectors that Dell is already using in some of their products, but in multiple sets to connect graphics cards.
Below that is a card with three sets of pads, which is what the CAMMs will have. Note that the graphics card was made in 2018, so this is not something new for Dell.
It's not nearly as complex or as easily damaged as LGA sockets.

I really, really wish something like this could be adopted as an industry standard. That would be so great, on so many levels.
Posted on Reply
#29
TheLostSwede
News Editor
@Valantar another thing that would be interesting with the CAMM, is if it was possible to use with LPDDR types of memory, as these are always soldered today.

Also, my old 12.5" Thinkpad X250 has a SO-DIMM slot and it has two batteries as well...
Personally I still think we need to move to a better interface, if this is it or not, I don't know. However, at some point in time, progress is needed.
Posted on Reply
#30
wahdangun
i think everyone miss the point, this thing beside being smaller and space efficient, its also dual channel and can support much more speed, and if this thing can prevent soldered ram, then I'm all for it, no more single channel ram.
Posted on Reply
#31
Valantar
TheLostSwede@Valantar another thing that would be interesting with the CAMM, is if it was possible to use with LPDDR types of memory, as these are always soldered today.

Also, my old 12.5" Thinkpad X250 has a SO-DIMM slot and it has two batteries as well...
Personally I still think we need to move to a better interface, if this is it or not, I don't know. However, at some point in time, progress is needed.
Yeah, that thought struck me too - if this can make for modular LPDDR, that would be fantastic. And yeah, we need new interfaces - across the board, really. Keeping compatibility is great (for PCIe and the like), but when you run into a wall, you run into a wall. And it's not like DIMMs are compatible across generations anyhow. Moving to a new interface more suited to the needs of upcoming designs and hardware standards is a necessity at this point - and argubably should have happened with DDR4, seeing how that was when soldered RAM seemed to take off too.
Posted on Reply
#32
Chrispy_
wahdanguni think everyone miss the point, this thing beside being smaller and space efficient, its also dual channel and can support much more speed, and if this thing can prevent soldered ram, then I'm all for it, no more single channel ram.
Yeah, the speed of SODIMMs is an issue that does need correcting. Timings on SODIMMs are awful compared to the same ICs on desktop DIMMS.

I have no problem with SODIMMs being universally superseded by something faster, but if CAMMs fragment the DDR6 market into conflicting standards, that's a terrible thing.
Posted on Reply
#33
trsttte
ValantarYeah, that thought struck me too - if this can make for modular LPDDR, that would be fantastic. And yeah, we need new interfaces - across the board, really. Keeping compatibility is great (for PCIe and the like), but when you run into a wall, you run into a wall. And it's not like DIMMs are compatible across generations anyhow. Moving to a new interface more suited to the needs of upcoming designs and hardware standards is a necessity at this point - and argubably should have happened with DDR4, seeing how that was when soldered RAM seemed to take off too.
The reason for soldering wasn't only space, also power savings (in the case of lpddr4) and the illusion of the supposed cold boot attack freezing memory modules to retain data. Depending on the situation it can also be cheaper to manufacture since soldered memory is placed by a machine with the rest of the components, modules need to be inserted manually. DDR4 moving to a new interface wouldn't necessarily change any of that (granted it could certainly help in avoiding it).
Chrispy_I have no problem with SODIMMs being universally superseded by something faster, but if CAMMs fragment the DDR6 market into conflicting standards, that's a terrible thing.
It's more likely to fragment DDR5, by DDR6 they must sort it out
Posted on Reply
#34
Valantar
trsttteThe reason for soldering wasn't only space, also power savings (in the case of lpddr4) and the illusion of the supposed cold boot attack freezing memory modules to retain data. Depending on the situation it can also be cheaper to manufacture since soldered memory is placed by a machine with the rest of the components, modules need to be inserted manually. DDR4 moving to a new interface wouldn't necessarily change any of that (granted it could certainly help in avoiding it).
Yup, all of those, but also AFAIK LPDDR4 and 4x can't handle the trace lengths and in-line connectors needed for SO-DIMMs, so soldering is the only option. My hope is that this can allow for the CAMM to be closer to the CPU (no need to space things out for matched trace lengths to two sockets), which might get within the requirements for LPDDR, plus the improved connector will no doubt have less resistance and other detrimental effects on the signal. Of course that freezing DIMMs thing is ... well, rather theoretical given the level of physical access you need for that to work - plus a screwed-in CAMM would likely make that very challenging to pull off in practice as the removal/installation procedure would take quite a lot longer, and you'd need to keep it frozen that whole time. But that's an extreme edge case either way. IMO the cost is a wash - one is cheaper in some situations, another in others, but the difference isn't likely to be huge either way. CAMM will no doubt be more expensive in the beginning, especially due to the rather unusual connector, but nothing that economies of scale can't fix - and trace routing to a single socket is likely slightly easier than to two SO-DIMM sockets.
Posted on Reply
#35
wahdangun
Chrispy_Yeah, the speed of SODIMMs is an issue that does need correcting. Timings on SODIMMs are awful compared to the same ICs on desktop DIMMS.

I have no problem with SODIMMs being universally superseded by something faster, but if CAMMs fragment the DDR6 market into conflicting standards, that's a terrible thing.
it will be JEDEC standard so if some manufacture other than dell want to used it they can.
Posted on Reply
#36
Valantar
wahdangunit will be JEDEC standard so if some manufacture other than dell want to used it they can.
Saying "it will be" is rather premature. Dell reportedly wants it to be one, and there are good reasons for this or something similar to become a JEDEC standard, but JEDEC is an industry standards body, and no single actor can just come up with a new standard and then say "hey guys, this is the new standard". They can suggest it, and it will then go through the regular feedback and iteration process that such standards go through, likely involving feedback from all relevant/interested JEDEC members.
Posted on Reply
#37
noel_fs
thats surprising unironically
Posted on Reply
Add your own comment
Jul 15th, 2024 21:42 EDT change timezone

New Forum Posts

Popular Reviews

Controversial News Posts