Not saying Framework should get these. They are an SoC which likely means soldered ram and storage. Who even knows about the Linux support. Still, would be interesting to see an upgrade friendly version of this in the future. The video goes over some tasks that take really good advantage of the NPU.
The Snapdragon has 16 pcie 4.0 lanes in this mobile variant. 8 of which are not actually being used for anything; according to Linus. I wonder if this would make it easier for Framework to port itās I/O to the USB C expansion ports scheme.
Also, that battery life. As a 12th gen Intel Framework owner; the battery life is terrible.
Also, if Framework could offer a thinkpad or thinkpad-like keyboard option that would be great too.
They are an SoC which likely means soldered ram and storage.
Storage can be replaceable. The new Yoga Slim 7x with the X Elite has an M.2 2242 SSD.
Seems like the ram of other laptops coming with the X Elite out are soldered to the mother board - so I donāt see why it couldnāt be replaceable in theory
Lpddr5x works woth the snapdragon processors, so in theory lpcamm2 should allow for upgradable memory on snapdragon-based mainboards.
Theres even a lenovo device out already with lpcamm2, so its not too far ouside the realm of posibility for framework to come out with an lpcamm2 + snapdragon system. Itd be very in-line with their mission.
Canāt do lpcamm with lpddr4, was my first thought too though.
On the other hand the qualcom stuff, however much they massaged their numbers, is going to be a lot more competitive with the intel and amd platforms than any risc-v platform in the next couple years, especially this one.
Forgive me for being the dumb guy in the room, but I also have been really hoping for the snapdragon X SOC to come to the framework. I currently run a MacBook almost solely for the reason of being able to just close the lid and throw it in my backpack as well as I never have to turn it off before putting it down and itās always ready to go the second I open it.
I really donāt want to buy another MacBook but the constant ready and battery life seems no windows system can touch (I recently test drive a XPS 14 with a 155H) it was the closet Iāve seen but still wasnāt able to match the āconstantly readyā state of my MacBook.
I feel like even having the SOC/RAM be soldered to the board is okay for framework sense when you want an upgrade could replace the mobo.
So is there a reason or way that the RISC-V chip is better than the snapdragon X?
That is a very slippery slope, especially since replacing the mainboard to get more ram is already pretty much the status quo in laptops which is exactly the issue.
In this case since the soldered memory is the absolute maximum the platform can handle it is kind of ok.
The one coming to framework? Performance wise not even close, hell itāll struggle to keep up with a raspberry pi. This isnāt really much of a general consumer product but a platform for testing an development.
I mean, as a 13th gen user, Iāll have to buy new RAM anyway when I do any upgrade, since nothing new will support the DDR4 that I have.
What Iād really like is something like 16/32GB soldered quad-channel + 32/48/64GB LPCAMM2 module. I really want to see what framework does with the new AMD chips (and Iād especially like a Strix Halo FW13 mobo).
As one other has said; this is barely more powerful than a Raspberry Pi. I found this out once I began reading more comments a few hours after I posted. To clarify my earlier statement; I meant I would prefer to buy an IPC competitive, mostly ready for daily laptop use, RISC-V chip.
I do not. Keeping all the separate possible variant skus of ram soldiered and ssd boards would drastically drive up Frameworkās logistic costs. SODIMM modules/slots are very inefficient. Once LPCAMM2 boards are out the difference will narrow between Framework and soldiered vendors. I am a software developer-myself and my peers have already standardized on 32 gigs of RAM and are slowly moving to 64 gigs. Most laptop users do not need that. I also like to overprovision for storage-especially for m.2 ssds; you have far fewer connections for that compared to sata ssds.
I have that right now with my Framework Intel 12th gen with Fedora Linux. LinusTechTips also referenced a past video he did where only ASUS and Framework properly implemented power states for x86 Windows laptops.
The battery life is absolutely terrible (less than my 10th gen work issued thinkpad), but I also havenāt gotten around to updating the bios. Bios updating is broken on Linux and I have to use uefi method which may wipe my uefi memory. One of my tech friends recently bought the newest AMD version of the framework. He says the battery life is comparable to his wifeās M3 macbook with Fedora Linux. He did not use Windows as we do not use it in our line of software development work.
Others should feel free to correct me but the M3 has worse battery life than the M1? @Jameson did not say they had a M1, but I am assuming.
thats really sad. i really want a good daily driver alternative but really dont want to spend 4k on another mbp.
I do not. Keeping all the separate possible variant skus of ram soldiered and ssd boards would drastically drive up Frameworkās logistic costs.
i do get that. how ever even if they charged me 1k for a new snapdragon board and more ram (i understand those are tied together being an SOC) i would be okay with it compared to the alternative (apple)
It is for developers who want to use a riscv platform. Not for daily use or for regular users. Please keep that in mind. That is why I clarified my statement.
Even performance aside, that thing definitely isnāt daily driver grade even for people developing for it.
The snapdragon should be able to use lpcamm as it uses lpddr5. SOC means stuff like the memory controller and chipset and some of the io is integrated, putting the memory on there too is a few steps further.
First got my 32GB RAM laptop back in 2010 (W510)ā¦went to 64GB around 2016 IIRC (P50)ā¦ Slowly moving to needing 128GB+ (work related use cases).
That got me thinking, from a purely out-of-landfill / environmental cost perspective (financial cost aside), shouldnāt manufacturers just stick with max spec memory modules, to maximize the physical itemās lifetime? That way, regardless of where the memory module ends up, itās always āusefulā to that generation of hardware as oppose to having this āsizeā / use case restriction.
What do you use it for? I do kernel development and that means VMs. Edit: my friend does Kubernetes work so minikube.
They do not care about that. They care about upcharging you and charging you again once you realize you do not have enough memory. Apple calls 8 gigs enough for laptop users for a reason.
Nice. Off topic: been thinking of a putting together an Epyc workstation system so I can put in 256 gigs of ram for an OpenShift environment or as a kernel build farm. 2nd gen Epyc cpus were very cheap a few months ago before a Youtuber talked about them-now I struggle to find them on ebay.
