Trackpoint and MXM.
Trackpoint is an accessibility thing. I’m thankfully young and my arms work fine, but for people with limited mobility, Trackpoint is very helpful. It’s not under patent or anything, it was introduced over 20 years ago and there are other companies making keyboards with it.
MXM is not wanted for upgradeability, I’d be fine if the Framework shipped with an MXM Rage 128 Pro. Right now, I can vouch for KDE Plasma 5 running buttery smooth on a GMA 950 with 64MB VRAM. I understand the adage of “you’ll want to upgrade the CPU at the same time anyway so why bother”. It’s wanted for extra repairability. A faulty GPU could pretty much end a laptop right then and there. It’s bad enough if the laptop is a few years old, way worse if it came from the factory with a dead iGPU. The MXM scene is a little dire right now for sure, but I feel like this is something that should be looked into because it’s one less thing that can render your whole mainboard e-waste, and it might help revitalize MXM at least a little as well. That’s especially important for me, who likes to get a decade of use out of my hardware.
Chip failure is very low as far as failure rate is concerned and if the iGPU were dead from the factory…then it would obviously be covered under warranty and of no real concern to you, especially since QA would prevent such a laptop from even being produced, Intel won’t sell a defective chip like that. If the iGPU is defective, they bin it as an F-SKU and market it as such. MXM seems on it’s way out and Framework doesn’t have the market share to in any way halt or arrest its demise. Besides, MXM would actually create another point of failure and hurt repairability as the likelihood of finding replacement parts will steadily shrink as MXM vanishes into obscurity.
That still generates a fair bit of waste since you have to burn the oil to get it back to Framework instead of just ordering a new card.
They might not, but that doesn’t mean it can’t get damaged afterward.
Isn’t DDR4L SODIMM also vanishingly rare, at least on laptops of this sort? Sure, it’s more common than MXM, but both are usually most found on gaming laptops. The most recent MXM release I can see on TPU is the Quadro RTX 5000, but I also follow the Powerboard Tyche project, and their supplier is not only still making MXM cards, but they’ve just recently updated to GDDR6.
I fail to see how going from “literally unrepairable” to “theoretically repairable if it’s still being made” is a downgrade. At worst, there’s no diffrence, but I can hop on eBay right now and get myself a Voodoo3 3000 PCI if I want one. You would have had a point about the socket corroding or breaking off, but even then, sockets on IBM 5150s are just fine, as long as they use good quality parts it shouldn’t be an issue.
Yeah but that’s all hypothetical, Intel checks the dies before shipping them to any manufacturer for integration, it wouldn’t even leave the fab like that.
Again hypothetical and that also applies to any MXM GPU as well being damaged and if the CPU die gets damaged and the GPU die doesn’t…it’s still dead Jim.
I don’t know of any SODIMMS that are DDR4L, I’d actually be interested in seeing that.
Your link says otherwise. They had to switch suppliers and AMD is no longer producing MXM cards. That’s a direct quote.
True, but that’s like saying that socketed RAM is pointless for repairability because if the CPU gets damaged the whole system is dead. I’d love for socketed CPUs to be back too, but MXM is at least a start and easier to get ahold of for low power parts.
From my understanding (and a quick sanity check), it seems like most DDR4 SODIMMs are DDR4L? It wasn’t in the DDR4 spec as of 2019 but 1.05v was the agreed upon unofficial standard. But then other sources say it’s all regular DDR4… I assume some weirdness happened, the same weirdness that caused 1GB PC133 SODIMMs to exist, but that it wasn’t official. Whether DDR4 or DDR4L, either one is rarer and rarer as time goes on.
Fair enough, that’s what I get for going off memory, but the fact there are GDDR6 MXM cards (which they brought up in this post, not that one) at all is a glimmer of hope. Dell is still selling the Alienware Area 51m that has both MXM and a socketed 2070 Super, so that’s promising.
ehhh, to an extent that is true. I’d argue that SO-DIMMS are much easier to get as a consumer vs MXM modules but you are right that if the GPU dies on an MXM modules it would be easier to replace. That still doesn’t negate the fact that having an iGPU doesn’t adversely impact repairability. The iGPU is built into the CPU package so if anything is going to damage the iGPU, it’ll damage the CPU too and that would end the board regardless of the presence of socketed RAM/GPU.
Now I realize why DDR4L sounded weird to me, I’m used to hearing it referred to as LP(low-power)DDR4 or LPDDR4x. Both of which are soldered modules and do not come in socketed form AFAIK.
I would expect a larger battery like 65-70W. The current 55Wh is simply too small now given the low battery life.
Does battery capacity not scale with size? I’m reasonably there are some large trade-offs for more energy dense batteries as well (such as shorter endurance).
That being said a large battery would of course be great!
No, LPDDR4 is a different thing. LPDDR3 coexisted with DDR3L (and apparently LPDDR1 existed and was used in the iPhone 3GS), where the latter was just DDR3 that was undervolted a little. DDR4L is technically unofficial but theoretically DDR4 at 1.05v, but otherwise the same chips as DDR4 in the same modules.
I found a Super User post where someone says they bought DDR4L 2133, so clearly it exists in some capacity. The two four answers on this Quora post also seem to agree DDR4L exists, though one is just talking about SODIMM and not specifically DDR4L.
It kinda scales with size. And I take this as a possible blocker on FW’s upgrading path as a larger-sized battery may not fit in the current model and may need a redesign on the chassis/boards. But thats the best way I can think of to deal with the current high battery-drain issue.
imo best way to deal with it is software! Designing a whole new chassis and mainboard would be counter-goals I think!
Would be great if we can configure the fan curve profiles in the BIOS too.
Barring that, I know of this great little program called fancontrol that I found on the LTT forums. Give it a shot, it can even be configured to start on boot or multiple profiles I think.
In Linux, thanks to ectool, you can actually do better and control the fan curves without going into the BIOS:
There’s also a neat little daemon which lets you do more advanced fan control strategies:
Yea, I just was thinking of a hardware level one so it doesn’t matter what OS do I switch to.
But you are right, being able to control at this level would be similar to Mac OSX where they are even able to achieve Power States as well as even downclocking even below Intel’s spec sheet. Maybe I will play around with it tonight.
I’ve seen SIM card Expansion Cards requested multiple times.
However, how about an integrated SIM card on the mainboard?
This way, an expansion slot is saved for something else. Also, if the SIM expansion cards are eventually released, it would be a good way to enable dual-SIM configurations.
It’s not just a SIM card it’s all the network (frequency) transmitters and receivers and antenna that have to be worked into the main board
It would be best to save the functionality as an expansion card for multiple reasons. One is that each region of the world uses different frequencies and tuning antennae for multiple regions would add complexity. More basically, there will many users that do not want additional capabilities attached to the machine that could be used for tracking (myself included).
This makes me feel like even an expansion card would be unlikely.
I am aware of the different world frequency standards, but more so for signal creating devices (WiFi APs).
From my experience, phones seem to work without problem in multiple countries. So, do devices that use WiFi.
What would be the differences in this case between a laptop using a SIM card versus a laptop using a WiFi card?
That is a good point about privacy concerns.
because they have modems that supports many different bands/frequencies for each standard, so its likely that at least one of them is in use in the country you’re in. It’s not always optimal or sometimes there isnt one, but it generally works out
EDIT: It just occurred to me that I should probably add that this situation was NOT always the case. Years ago you needed specific phones to travel to specific countries. This is what ‘dual/triple/quad band’ was about. The more bands it had the more likely it was to work in different countries. It’s rarely an issue with phones now, but modem modules can still be specific to certain regions.
Wifi is generally much the same wherever you are, the different local regs only allow or disallow certain channels or restrict transmit power.