Article: CAMM, Dell's new SODIMM standard

scottypippin

scottypippin

Weaksauce
Joined
Feb 28, 2022
Messages
106
I think I might be reviving an old topic, but it is relevant now that we have actual physical specimens of the thing.

https://www.notebookcheck.net/CAMM-memory-preview-The-Dell-SODIMM-revolution.658666.0.html

Its easy to be skeptical of this - it looks like a modern dell "creates more problems than it fixes" solution to me, but that could be unfair given they are not going the proprietary route. My main takeaways (which the article also comments on):
  • Bad thing: Takes up less vertical but way more horizontal space
    • It's possible this could eventually not matter if memory modules decreased in physical size, but....not sure that has happened. Laptops are becoming more Z dimension compact, but not as much X and Y so maybe this won't be an issue.
  • Good thing: There is room (as Dell claims) to put something like a wireless card underneath the CAMM module. That'd be nice as boards become less and batteries become more of the internals of consumer devices.
    • However (I'm speaking from some ignorance here) given DDR5's increase in temperature, would it be feasible to cook your wifi card with new modules? Again, asking from ignorance here...
  • Weird thing: 6 screws?
Like the article says, interface speed improvements are needed:
More specifically, RAM speeds greater than 6400 MT/s are not possible with the current standard. This ceiling is becoming a problem on very high-end laptops and workstations while most consumer-level laptops have yet to reach the limit. Naturally, more laptops will hit the ceiling in the years to come as RAM speeds become faster and faster.
Click to expand...
Anyone know of any other manufacturers taking a stab at a new SODIMM connection standard?
 
Dell has some great engineers but the problem is they are looking for solutions that didn't have problems. They like to over engineer crap that makes everything worse. I guess the main thing is to make stuff as proprietary as possible.
 
I'm going to be completely honest.

In the PC industry, anyone who does  anyhting that is not an open standard, can fuck right off.

Proprietary standards have absolutely no place in this industry, and customers need to vote with their wallets and absolutely crush them whenever they rear their ugly heads.

The same with firmware lockouts (or lock-ins)

Absolutely everything needs to be universally compatible.

Thats what makes the PC industry tick.
 
This is both cool and aggravating, at least Dell is trying to make it a standard. It does tackle a problem that DDR5 has introduced, the SODIMM format does have a performance ceiling that DDR5 touches upon. I don't like that Dell is the one championing this though, but I guess somebody has to. Samsung brought us the M.2 interface standards so maybe Dell brings us this, who knows?
 
Zarathustra[H] said:
I'm going to be completely honest.

In the PC industry, anyone who does  anyhting that is not an open standard, can fuck right off.

Proprietary standards have absolutely no place in this industry, and customers need to vote with their wallets and absolutely crush them whenever they rear their ugly heads.

The same with firmware lockouts (or lock-ins)

Absolutely everything needs to be universally compatible.

Thats what makes the PC industry tick.
Click to expand...
I'd rather say that if it's going to be proprietary, it should be a truly unique-to-that-brand design that gives it a clear advantage. Folks might not like that Apple has its own silicon with RAM built into the SoC, but that has led to Macs that are sometimes better (or at the least, clearly different) than their x86 Windows counterparts.

My beef is when companies create "standards" that are really attempts to control an industry. You see this most in software. Remember how Google/BlackBerry/Palm/etc. insisted that all must worship Flash, or how there was pressure on Apple to support PlaysForSure on the iPod (and thus chain the entire digital music biz to Microsoft)? Yeah, it's not a standard if one company not only created the technology, but is solely responsible for maintaining it.
 
The more obvious solution that most companies have and are moving towards is simply memory that isn't removable. It "solves" basically all the problems other than not being upgrade-able or replaceable. Which at least still gives the users the benefits of fast(er) memory.

Power users hate it. But the honest truth is laptops are for the most part disposable commodities. Power users at the top end are replacing their machines with frequent regularity and casual users at the other end won't care. Basically it's only power users that want to buy their own RAM to undercut the OEM who will most care about soldered in RAM. And frankly OEM's don't care about that.
 
Yeah, sounds like it's going to become a lot more expensive and/or a lot less practical to upgrade the RAM in Dell laptops.
 
GotNoRice said:
Yeah, sounds like it's going to become a lot more expensive and/or a lot less practical to upgrade the RAM in Dell laptops.
Click to expand...
Well the high end Lenovo and HP moble workstations are moving over to soldering the ram down, so I would argue the Dell solution is still cheaper and more practical.
 
Zarathustra[H] said:
I'm going to be completely honest.

In the PC industry, anyone who does  anyhting that is not an open standard, can fuck right off.

Proprietary standards have absolutely no place in this industry, and customers need to vote with their wallets and absolutely crush them whenever they rear their ugly heads.

The same with firmware lockouts (or lock-ins)

Absolutely everything needs to be universally compatible.

Thats what makes the PC industry tick.
Click to expand...
I'm with you on this, and it doesn't even look like it will save much space on the z-axis.
This is proprietary garbage that the rest of the industry needs to quash.
 
Oh man, this smells like the dell-only gpu format that was labeled as an "open" standard which promptly, no one adopted. They killed off mxm to make a few more bucks. HP still uses mxm in their highest end zbook.
As an owner of many precision laptops, my current one will be my last.
 
commissioneranthony said:
Oh man, this smells like the dell-only gpu format that was labeled as an "open" standard which promptly, no one adopted. They killed off mxm to make a few more bucks. HP still uses mxm in their highest end zbook.
As an owner of many precision laptops, my current one will be my last.
Click to expand...
Do you recall what that "dell-only gpu" standard was? I'm not familiar....
 
Where exactly is the sodimm speed limit coming from, and why can't it be addressed the same way that the dimm connection on desktop systems is upgraded with each new generation of DDR?

I'm trying to understand if this is addressing a real problem, or is just a "problem" being exploited with the intent of achieving punitively expensive/difficult aftermarket upgrades so that Dell'll be able to gorge itself of overpriced ram upgrades at time of purchase without the extra expense of having to maintain multiple PCBs with different soldiered ram configurations.
 
DanNeely said:
Where exactly is the sodimm speed limit coming from, and why can't it be addressed the same way that the dimm connection on desktop systems is upgraded with each new generation of DDR?

I'm trying to understand if this is addressing a real problem, or is just a "problem" being exploited with the intent of achieving punitively expensive/difficult aftermarket upgrades so that Dell'll be able to gorge itself of overpriced ram upgrades at time of purchase without the extra expense of having to maintain multiple PCBs with different soldiered ram configurations.
Click to expand...
IIRC it's size of the module itself - limited number of traces or the like, but I only saw it in passing while digging for something else on JDEC ram details.

In theory this is an open standard - anyone can make a CAMM module, much like anyone can make an M.2 drive now - which, ok - sure - that's not bad. Given that we've always had standards that change over time (SIMM -> DIMM -> different types of DIMM and so on), that's ok - if it's an actual open standard.
 
DanNeely said:
Where exactly is the sodimm speed limit coming from, and why can't it be addressed the same way that the dimm connection on desktop systems is upgraded with each new generation of DDR?

I'm trying to understand if this is addressing a real problem, or is just a "problem" being exploited with the intent of achieving punitively expensive/difficult aftermarket upgrades so that Dell'll be able to gorge itself of overpriced ram upgrades at time of purchase without the extra expense of having to maintain multiple PCBs with different soldiered ram configurations.
Click to expand...
The problem lies with the socket design of the SODIMM modules themselves, the Small Outline Dual Inline Memory Module has a pretty weird tracing requirement which elongates the traces adding lots of latency. In theory, there isn't a limitation on the specification itself but the tracing plus the placement needed to make it actually fit in a laptop adds unacceptable amounts of latency for the DDR5 spec which limits its practical application to DRR5 4800. Timings and speeds get even worse as you add more modules so if you build a laptop with 4 memory modules using SODIMM chips you are practically limited to DDR5 4000 speeds due to the tracing length between the CPU and the last module.
It should be noted that the CAMM design is a variation on the JEDEC standards and doesn't do anything weird, it is just a new connector. This allows you to use an Interposer if you are so inclined to use traditional modules.
1665168617101.png
 
Last edited:
Lakados said:
The problem lies with the socket design of the SODIMM modules themselves, the Small Outline Dual Inline Memory Module has a pretty weird tracing requirement which elongates the traces adding lots of latency. In theory, there isn't a limitation on the specification itself but the tracing plus the placement needed to make it actually fit in a laptop adds unacceptable amounts of latency for the DDR5 spec which limits its practical application to DRR5 4800. Timings and speeds get even worse as you add more modules so if you build a laptop with 4 memory modules using SODIMM chips you are practically limited to DDR5 4000 speeds due to the tracing length between the CPU and the last module.
It should be noted that the CAMM design is a variation on the JEDEC standards and doesn't do anything weird, it is just a new connector. This allows you to use an Interposer if you are so inclined to use traditional modules.
Click to expand...

That still seems like it should be something for JEDEC (or whoever owns the SODIMM) standard to be addressing.
 
DanNeely said:
FFS!
Click to expand...
I know right, you just have to laugh at the absurdity of it all.
But I have to correct myself HP owns the patent for DIMM, Intel owns the patent for SO-DIMM, so that's my bad.
Intel very well could correct it at a later point, I would trust that to be an issue on their radar as they do a lot of mobile devices.
 
scottypippin said:
Do you recall what that "dell-only gpu" standard was? I'm not familiar....
Click to expand...
It's called dell DGFF. It was the dell proprietary MXM replacement that no one ended up using except dell. Thus, owners of extremely expensive workstation laptops were forced into only using dell only gpus. Before you could use any mxm-b or mxm-a module within reason (ones that supported LVDS, ones that supported edp) and if you flashed a dell bios.
They took a very cheap and easy to implement MXM format and added unnecessary complexity and cost to pretend to shave off a fraction of a mm in z-height from laptop thickness.
It's a pile of nonsense and a pure business play. Dell can go screw off.
 

Attachments

  • s-l1600.jpg
    s-l1600.jpg
    427.9 KB · Views: 0
  • s-l1600-1.jpg
    s-l1600-1.jpg
    59 KB · Views: 0
commissioneranthony said:
It's called dell DGFF. It was the dell proprietary MXM replacement that no one ended up using except dell. Thus, owners of extremely expensive workstation laptops were forced into only using dell only gpus. Before you could use any mxm-b or mxm-a module within reason (ones that supported LVDS, ones that supported edp) and if you flashed a dell bios.
They took a very cheap and easy to implement MXM format and added unnecessary complexity and cost to pretend to shave off a fraction of a mm in z-height from laptop thickness.
It's a pile of nonsense and a pure business play. Dell can go screw off.
Click to expand...
*facepalm*
Yeah, I'll pass on that. Thanks for digging that info up!
 
Lakados said:
Well the high end Lenovo and HP moble workstations are moving over to soldering the ram down, so I would argue the Dell solution is still cheaper and more practical.
Click to expand...
Yeah, if the RAM is at least upgradeable then it is a good thing.

The lower-end Dell laptops already have some models that have the RAM soldered on. RAM goes bad and the whole motherboard has to be replaced.

User needs more RAM... Then they have to get a whole new laptop.

The Precision Workstation models with DDR4 currently have 4 SO-DIMM slots and can take up to 128GB RAM.

The new ones, the 7670 and 7770, are already using the CAMM modules. There are only 2 slots but can take 64GB modules and run at DDR5-4800.

I haven't had a chance to play with these yet but the performance has got to be great.
 
cyclone3d said:
Yeah, if the RAM is at least upgradeable then it is a good thing.

The lower-end Dell laptops already have some models that have the RAM soldered on. RAM goes bad and the whole motherboard has to be replaced.

User needs more RAM... Then they have to get a whole new laptop.

The Precision Workstation models with DDR4 currently have 4 SO-DIMM slots and can take up to 128GB RAM.

The new ones, the 7670 and 7770, are already using the CAMM modules. There are only 2 slots but can take 64GB modules and run at DDR5-4800.

I haven't had a chance to play with these yet but the performance has got to be great.
Click to expand...
This is going to be a bigger fight than not in the coming years, if the rumor mill is somewhat correct then both AMD and Intel are going to be selling more SoC solutions for their next-generation mobile platforms and far fewer socket-based ones. So both are looking at having the CPUs and Ram soldered down and going far more apple like in their mobile strategies.
 
Lakados said:
This is going to be a bigger fight than not in the coming years, if the rumor mill is somewhat correct then both AMD and Intel are going to be selling more SoC solutions for their next-generation mobile platforms and far fewer socket-based ones. So both are looking at having the CPUs and Ram soldered down and going far more apple like in their mobile strategies.
Click to expand...
The CPUs for the most part are already all soldered on, even on the Precision models.
 
cyclone3d said:
The CPUs for the most part are already all soldered on, even on the Precision models.
Click to expand...
Swapping out a laptop CPU has never been a common thing to do even among power users. That's very different from ram, where getting the smallest ram capacity and doing an after market upgrade has often been a way to save money at purchase time
 
DanNeely said:
Swapping out a laptop CPU has never been a common thing to do even among power users. That's very different from ram, where getting the smallest ram capacity and doing an after market upgrade has often been a way to save money at purchase time
Click to expand...
Storage too, I'm sure none of you would be surprised by OEMs using class 30 or below for most solid state offerings, then charging you much much more for a class 40+ upgrade.
Best usually to just order their cheapest config in a class 20 or what ever, replace it with aftermarket class 40's of sufficient size and just clone your images to those.
USB base NVME docks make short work of that process.
 
DanNeely said:
Swapping out a laptop CPU has never been a common thing to do even among power users. That's very different from ram, where getting the smallest ram capacity and doing an after market upgrade has often been a way to save money at purchase time
Click to expand...
I've done laptop CPU upgrades quite a few times.

Now with newer laptops you have to buy a board with everything soldered on instead of just being able to upgrade what you want to.

All in the name of making devices thinner.
 
cyclone3d said:
I've done laptop CPU upgrades quite a few times.

Now with newer laptops you have to buy a board with everything soldered on instead of just being able to upgrade what you want to.

All in the name of making devices thinner.
Click to expand...
You’re literally the only person I know that’s done a laptop CPU upgrade then - that’s exceedingly rare.
 
lopoetve said:
You’re literally the only person I know that’s done a laptop CPU upgrade then - that’s exceedingly rare.
Click to expand...

I did one in the mid-2000s, but back in the day, they still used sockets.
 
lopoetve said:
You’re literally the only person I know that’s done a laptop CPU upgrade then - that’s exceedingly rare.
Click to expand...
There are plenty of forums posts (not necessarily here) asking about CPU upgrades and GPU upgrades for laptops.

Not so much anymore, but the Dell Precision 4600 - 4800 / 6600 - 6800 were upgradeable.. had the option of either Nvidia or AMD.

The 4600 is the newest one I did an upgrade on.

These were 2nd through 4th gen Core I series.
 
cyclone3d said:
There are plenty of forums posts (not necessarily here) asking about CPU upgrades and GPU upgrades for laptops.

Not so much anymore, but the Dell Precision 4600 - 4800 / 6600 - 6800 were upgradeable.. had the option of either Nvidia or AMD.

The 4600 is the newest one I did an upgrade on.

These were 2nd through 4th gen Core I series.
Click to expand...
Huh. Fascinating. I was mostly apple during that period, but I also tend to buy a laptop for a designed life of 2 years max. By then it’s time to upgrade all of it anyway- and tends to be rather beat up. And I bought what I needed to start with the exception of ram. I did do ram upgrades over the years.
 
Hopefully JEDEC standardizes the module form factors, just an agreed upon interconnect is useless if every laptop maker has its own set of sizes.
 
DanNeely said:
Hopefully JEDEC standardizes the module form factors, just an agreed upon interconnect is useless if every laptop maker has its own set of sizes.
Click to expand...
Looks like that is the part they are working on now.
https://www.pcworld.com/article/1473126/camm-the-future-of-laptop-memory-has-arrived.html
“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.

Potentially sucks for people who have workstations using the 0.5 specification which is what the existing Dell laptops are using, it sounds like the size and hole placement may be different for 1.0

"Schnell actually created the original CAMM—or Compression Attached Memory Module—design for Dell last year. JEDEC’s CAMM standard will be based on that CAMM design but is likely to be somewhat different as companies hammer it out."
It also seems they are not doing away with SO-DIMM, just using CAMM for speeds exceeding DDR5-6400

"Some of the motivation for expediency likely comes from the fast-approaching “brick wall” facing laptops when SO-DIMMs hit at DDR5/6400.

Schnell said the CAMM spec is far from finalized, but the first JEDEC CAMM modules should take over right where SO-DIMM ends at 6400."
 
Does anyone have a link to the paper or working group study showing the downfalls of sodimm vs camm? I can imagine that reducing latency is important, but again, why not just update the sodimm spec? Why not just require tighter PCB trace routing and length matching requirements to reduce ringing or require additional clock re-timers (which are used with other high bandwidth communications) if the sodimm modules are placed too far outside of a designated trace length from the main CPU?
This just seems like a pure money play to me without real hard information to back it up.
It's just too easy for large manufacturers to use their business might to change an established standard for reasons none other than new revenue streams.
 
commissioneranthony said:
It's called dell DGFF. It was the dell proprietary MXM replacement that no one ended up using except dell. Thus, owners of extremely expensive workstation laptops were forced into only using dell only gpus. Before you could use any mxm-b or mxm-a module within reason (ones that supported LVDS, ones that supported edp) and if you flashed a dell bios.
They took a very cheap and easy to implement MXM format and added unnecessary complexity and cost to pretend to shave off a fraction of a mm in z-height from laptop thickness.
It's a pile of nonsense and a pure business play. Dell can go screw off.
Click to expand...
I don’t buy Dell due to their history of proprietary shit. This adds nothing to change my view of them.
 
commissioneranthony said:
Does anyone have a link to the paper or working group study showing the downfalls of sodimm vs camm? I can imagine that reducing latency is important, but again, why not just update the sodimm spec? Why not just require tighter PCB trace routing and length matching requirements to reduce ringing or require additional clock re-timers (which are used with other high bandwidth communications) if the sodimm modules are placed too far outside of a designated trace length from the main CPU?
This just seems like a pure money play to me without real hard information to back it up.
It's just too easy for large manufacturers to use their business might to change an established standard for reasons none other than new revenue streams.
Click to expand...
Pretty soon they’ll sell ram size as a service
 
You must log in or register to reply here.
Back
Top