Why is there no ryzen AI on the 16

Good day. I am new to the framework ecosystem. I am very interested in what you are doing but i have a question.

A 13 inch for me is way too small a plscreen for me to read/use. I am just curious as to why there isn’t a 16 that utilizes the new Ryzen AI cpu? If it is an option, I didnt see it available.

1 Like

So, I don’t have anything concrete since Framework doesn’t announce anything until they’re nearly ready to launch, but they did say they were still working on the FW16 (with little else mentioned). I have a feeling that they might eventually put the Ryzen AI on the FW16. I believe we don’t have it yet because of a few reasons: AMD wanted soldered RAM, I’ve heard there was some issues with low PCIe lane count on the Ryzen AI chips too which wouldn’t be good for the FW16, and heat dissipation is probably another big issue. Although I do think we’re more likely to get a different AMD chip (more PCIe lanes with different core configuration) or less likely an Intel one. I also suspect they’ll release a new GPU around the same time.

When this will be I have no idea. I’m sure we have a long wait.

4 Likes

Could be waiting for the compatibility of the newer LPCAMM2 RAM

2 Likes

If framework released a new mainboard like the one in the desktop for the FW16 with 128G or MORE - I would buy it in a heartbeat. And I don’t care if the RAM is soldered or not. I have a use case for local LLM work and this would make it possible.

2 Likes

My personal theory is in the design of the board itself.

Instead of the heatsink largely being contained within the form factor of the board, the Framework 16 boards have heat pipes that extend above the board to where the expansion bay is. This is significant as, around when the Framework 13 got its first replacement boards, an external enclosure was released for old boards to be reused.

The heat pipes make a similar design for the 16 boards a bit trickier, and its not like they can just recommend people remove the Liquid Metal for an external enclosure given the nature of LM. Part of the delay could simply be trying to figure out a way to design a case for these boards and potentially the expansion bay system as well.

1 Like

with the FW16 you could use the room for the mainboard + expansion module since the Ryzen AI has a iGPU, you would just have to be clear about this on your ordering site.

I don’t think that’s a valid statement, because they are doing it for the Framework 13 right now, using DDR5. Non-soldered RAM. To me it seems like they prefer the F13 over the F16 for upgrades/support, as their “main product”.

The main reason to go for the Ryzen AI is the ability to use iGPU for AI tasks. Our current Ryzen 7840HS/7940HS is super bad for that, AMD 780M doesn’t like to use GTT memory very well.

With the lower-tier AI processors, they are using socketed RAM. But those processors, such as the AI 9 HX 370, lack the amount of native PCIe lanes that the FW 16 utilizes.

I think the AI Max are the ones that essentially require soldered RAM at this point, in order to take advantage of some of the primary benefits of those processors.

I don’t think they are too many differences between 7840HS and HX 370.

https://www.cpu-monkey.com/en/compare_cpu-amd_ryzen_ai_9_hx_370-vs-amd_ryzen_7_7840hs

You will get extra battery life, small performance gains, and way better iGPU for computing purposes.

I think the main issue is the fewer PCIe lanes on the AI 7/9 processors. 16 wouldn’t be enough for all the ports, two NVME slots, Wi-Fi card, and the lanes dedicated to the expansion module (dGPU, additional NVME drives, etc.).

1 Like

Yup, I do understand that, but what if AMD doesn’t release any new processor with that enough PCIe lines for laptops? This is something they should have planned.

I seen few laptops using AI 300 with far more demanding GPUs like 5090 (latest Razer Blade 16 for example).

Now, if they have to release some motherboard with less features (soldered RAM or 1 NVME slot) but upgraded components, I think many people would still go for it, as long the results make it worth for upgrade.

The main purpose of Framework is to replace most components easily, and reuse as much as possible everything (screen, keyboard, chassis, USB slots, etc)

1 Like

I’m sure many would be okay with compromises, and Framework may even be working on ways around some of the limitations, or working on some other alternative. I’m also pretty sure that for every person excited to get an AI 9 chip in the FW 16, there would be another person shouting that Framework was betraying their customers and ruining the FW 16 and watering down the design, and destroying the modularity and universal compatibility, etc., etc., if Framework announced that the new AMD AI 300 mainboards would only have one NVME slot, or couldn’t use anything in the expansion bay, etc. Unfortunately, it’s impossible to please everyone, so they have to make tough decisions. Maybe they have a workaround and it is almost ready. Who knows.

Framework chooses not to release details of what is in the works until it is virtually ready, so at this point, all we can do is speculate. And since the lack of PCIe lanes seems to be a factor that would limit some of the core functionality they have built into the FW 16, I speculate that as one reason they may have chosen to hold off on making the Ryzen AI chips available in FW16 mainboards.

1 Like

The AMD Fire Range processor was announced … but the delivery was a bit later than the others AMD processors. I suspect this is what Framework will use for the FW16.
It has enough PCI lines for everything the FW16 has today… however it has a very limited iGPU and basically requires to have a dGPU.

It is the reason why, even if I was very reluctant to buy an outdated processor generation, I ended up purchasing the FW16 almost 2 months ago. I do need a usable iGPU and the Fire Range processor would not work for me. (Plus my current laptop was on his last leg).