News Intel's Lunar Lake MX mobile processor pictured, with LPDDR5X memory on board

I am guessing it uses 2x 64Gb (8GB) or 128Gb (16GB) LPDDR5x densities?

I wonder why they don't have an in-between size.
The 96Gb (12GB) density exists on catalogue for both Samsung and Micron, last I checked.
 
  • Like
Reactions: thestryker
I am guessing it uses 2x 64Gb (8GB) or 128Gb (16GB) LPDDR5x densities?
That's what every leak has indicated so far.
I wonder why they don't have an in-between size.
The 96Gb (12GB) density exists on catalogue for both Samsung and Micron, last I checked.
Nobody has been using them which really confuses me. I believe they've been listed as mass production for a while so the availability should be there. I'd primarily been watching the handheld space, because 24GB total capacity is perfect and shouldn't cost significantly more than a 16GB implementation.
 
This approach should use HBM technology and not DDR technology
Why should they do that? HBM is 2-3x more expensive, will the greater bandwidth provided be utilized to offset the added cost?

As on package DDR is already much faster than off chip DDR, there is less of a performance advantage for HBM.

Apple which has been doing on package memory since the M1 continues to utilize DDR standard (they started with LPDDR4X and quickly moved to LPDDR5 which they are still using on the M3). The bandwidth/latency advantage is tunable with on package DDR (High speed and large bit bus width). Commodity parts sure makes more sense to me here. HBM only is worth the cost if the application/use warrants it (H100, A100, B100/B200,etc...).
 
Last edited:
If they don't put 8 gb on the chip.and says that is enough like apple does it's good for me. Hbm it's just a dream... ddr5 just side by side with the cpu it's too good for u Poor consumers
 
Why should they do that? HBM is 2-3x more expensive, will the greater bandwidth provided be utilized to offset the added cost?

As on package DDR is already much faster than off chip DDR, there is less of a performance advantage for HBM.

Apple which has been doing on package memory since the M1 continues to utilize DDR standard (they started with LPDDR4X and quickly moved to LPDDR5 which they are still using on the M3). The bandwidth/latency advantage is tunable with on package DDR (High speed and large bit bus width). Commodity parts sure makes more sense to me here. HBM only is worth the cost if the application/use warrants it (H100, A100, B100/B200,etc...).
it is worth it to use HBM in laptops for onboard GPU ... the ARC GPU will run faster.