We handle this.
We handle this.
I discovered what appears to be the original slide for reviewing the footnote. The note on POR suggests that just because something is listed on the slide doesn't guarantee it will become a product in that form. There could be variations or no changes at all. Looking into CAMM shows it was agreed upon by the industry via JEDEC. Two key points stand out: we might use replaceable LPDDR chips that were usually soldered, allowing future capacity upgrades without immediate decisions. Also, historically LPDDR often offered greater speed compared to the standard DDR of its era.
I see Camm likely offering higher speed than DDR at the same cost with improved motherboard and better timing traces. It seems like LPDDR Camm2 isn't the standard Dell version—it's a newer iteration that returns to single-channel per stick design, making it simpler but less flexible. With Camm, I wonder if it might reduce the appeal of soldering, since you still benefit from memory soldering without being locked into it.
The modular design offers manufacturers significant advantages. Most SODIMM parts are DDR-only; LPDDR isn’t modular at all, though power efficiency and performance gains justify the compromises for mobile applications. Similar to how LPDDR5 reaches 6400MT/s at 1.05V, you should look for a DDR5 module that matches that speed. LPDDR5x also delivers 8533MT/s at the same voltage, or Samsung claims comparable speed with 20% lower power consumption. CAMM supports modularity without sacrificing performance or efficiency. High-performance DDR5 modules require at least 1.1V; remember that power consumption depends on voltage squared. The top binned DDR5 options cost much more than standard LPDDR5 chips and use 10% more power, assuming resistance remains constant. Modularity benefits not only customers but also reduces SKU inventory for manufacturers and improves repairability for service teams. Designing memory routing on boards is one of the most intricate and time-intensive aspects of motherboard engineering—typically around two months per SKU. With faster memory speeds, the complexity has increased, though better computer tools are helping. I don’t believe modularity will disappear entirely on smartphones, nor will Apple likely make it mandatory for their devices. Companies such as HP, ASUS, and Dell, which manage hundreds of SKUs each generation, will be particularly interested in these advancements.
My prediction: CAMM might substitute SODIMMs in their existing applications. Tighter designs will still use soldered connections.