News AMD silently bumped up memory specifications for Ryzen AI 300 CPUs — Strix Point now supports LPDDR5X-8000 as opposed to LPDDR5X-7500

[Admin]

A new ultraportable notebook from HP with the AMD Ryzen AI 300 "Strix Point" processor, set for release this December, features LPDDR5X-8000 memory which is a step up from the standard LPDDR5X-7500.

AMD silently bumped up memory specifications for Ryzen AI 300 CPUs — Strix Point now supports LPDDR5X-8000 as opposed to LPDDR5X-7500 : Read more

 

[JamesJones44]

They now support LPDDR5X-7500 memory instead of the previously stated LPDDR5X-7500

They now support LPDDR5X-8000 memory ...

 

[Roland Of Gilead]

They now support LPDDR5X-8000 memory ...

It's further down in the next paragraph too!

 

[NinoPino]

At author, please correct the typos :

"LPDDR5X-7500 memory instead of the previously stated LPDDR5X-7500"

"The jump from LPDDR5X-7500 to LPDDR5X-7500"

 

[DS426]

Now all AMD needs is ThreadRipper mobile.

 

[SonoraTechnical]

Meh,
I'd like to see a Ryzen 9 AI HX 395 16 ZEN5c cores for 32 threads, RDNA 3.5 with 24 CUs, XDNA 2 AI, LPDDR5X-8000 quad channel memory support, and support for dual PCIe5 NVMe for Socket AM5 with a slight bump in TDP for an increase in base and boost speeds. Would make for a very nice Mini-ITX box.

 

[DS426]

Meh,
I'd like to see a Ryzen 9 AI HX 395 16 ZEN5c cores for 32 threads, RDNA 3.5 with 24 CUs, XDNA 2 AI, LPDDR5X-8000 quad channel memory support, and support for dual PCIe5 NVMe for Socket AM5 with a slight bump in TDP for an increase in base and boost speeds. Would make for a very nice Mini-ITX box.

Yep, I'm definitely thinking along the lines of mostly or all Zen 5c cores. I'm not sure what Zen 5 Threadripper will look like but I'm one to assume that it will have some Z5c's.

 

[RUSerious]

Meh,
I'd like to see a Ryzen 9 AI HX 395 16 ZEN5c cores for 32 threads, RDNA 3.5 with 24 CUs, XDNA 2 AI, LPDDR5X-8000 quad channel memory support, and support for dual PCIe5 NVMe for Socket AM5 with a slight bump in TDP for an increase in base and boost speeds. Would make for a very nice Mini-ITX box.

Strix Halo will likely give you 16 cores. You can always dial in the wattage limits on Ryzen to get whatever TDP you are looking for. Why buy a Ford when the Ferrari only costs twice as much.

 

[abufrejoval]

I wish there was a link to the spec, because that's notoriously hard to find with AMD...

Because I'd like to know if the 32GB limit for anything LPDDR5 is for real, and if it's a limitation which AMD chose or common with all of LPDDR5 today: yes, I'd love to know where this limit--if real--comes from.

And since Chrstimas is drawing closer: would there be any chance you'd be able to run LPDDR5 say on the lower part of physical memory, where the iGPU is taking its chunk, while the upper part might still be implemented as SO-DIMM for that extra bit of expandability where bandwidth for CPU stuff isn't as crucial as paging would be even slower...

 

[Pemalite]

Tom Hardware needs to hire people specifically to check articles for accuracy before publication.

"They now support LPDDR5X-7500 memory instead of the previously stated LPDDR5X-7500." - IS the same thing, they made the error multiple times.

 

[Mama Changa]

Strix Halo will likely give you 16 cores. You can always dial in the wattage limits on Ryzen to get whatever TDP you are looking for. Why buy a Ford when the Ferrari only costs twice as much.

Halo is 8+8 design with support for LPDDR5X 8533 on a 256 bit bus.
Fire range will be a 16+0 design to replace Dragon range. There are no plans at all for an all 5c high core count model.

 

[The Historical Fidelity]

A new ultraportable notebook from HP with the AMD Ryzen AI 300 "Strix Point" processor, set for release this December, features LPDDR5X-8000 memory which is a step up from the standard LPDDR5X-7500.

AMD silently bumped up memory specifications for Ryzen AI 300 CPUs — Strix Point now supports LPDDR5X-8000 as opposed to LPDDR5X-7500 : Read more

“They now support LPDDR5X-7500 memory instead of the previously stated LPDDR5X-7500.”

Please correct this typo, very sloppy writing, was this your first draft?

 

[watzupken]

Feels like they are trying to pull ahead of Intel by bumping up the memory speed, which helps with iGPU performance.