I still think you’re missing the point. Okay, I’ll concede that some applications might benefit from a closer-to-square aspect ratio. The point I was making, however, is that aspect ratio is of limited relevance. For a given pixel width (and here I’m talking about “virtual pixels”, after accounting for scaling), a 3:2 display is strictly larger than a 16:9 display.
A 1920×1280 display is strictly larger than a 1920×1080 display. You can fit exactly the same stuff horizontally, but you have more vertical space. If an application manages that space that poorly… maybe don’t make it as tall as the entire screen (more room for something else!).
The problem with the current Framework display is that it is effectively 1500 “virtual pixels” wide, which is narrower than the 1920 we’ve become used to, and even that requires fractional scaling which has its own problems. A 3000×2000 display would be strictly better; still only 1500 “virtual pixels” but at least with integer scaling, though realistically, 1500 or even 1600 is not wide enough to have anything side-by-side (I speak from experience). A 15" 3840×2560 screen would probably be the best case, and I would certainly prefer that over a 3840×2160 screen.
If you’re arguing that a 1920×1080 screen would be better… well, sure. But it’s not because of the aspect ratio, it’s better because it has more horizontal “real estate”. But a 1920×1280 screen would be strictly better than 1920×1080.
My argument over the last essay is that for a given amount of pixels, a “wider” screen rather than a “tall” screen is more effective because it is suitable for more workloads.
Yes! Thank you. Now please, if you want to argue, just message me and not clutter up this thread any longer.
Here is my “design” for a 15"6 inch PC. Make it 16:9 like said by most people here so you may play videos at native scale + place two windows side by side.
I would keep it all the same for the internals, except one of the ports would be much bigger to accept a new kind of EGPU card on USB C.
The size of the slot is the same as CPU+FAN+1 mem; it will have to extract 30-40W of heat.
It’s not just the bezel though. Presumably you want your screen and chassis to be the same size, so this necessitates introducing a new top cover and chassis in addition to the bezel (and sourcing the screen, of course).
Right, sorry. You want the chassis, bezel (outer dimensions), and input cover to all be the same size. Obviously the screen will be smaller (and will match the inner bezel dimensions).
It depend on how much engineering framework would want to put into the late model.
In theory not much can be done if they want to support the current gen motherboard. However if they employ creative techniques they can have the PCIe lanes extended to an empty area and kit it out with MXM graphics.
The problem is, however, that because Framework had gone with the “built-in dongles” in their current, rather consumer-friendly way (rather than some pin headers that have reduced footprint), the motherboard is very small and cramming additional power delivery (among other required circuitry for anything else) is a challenging task. Even with 4-layer or even 6-layer boards because there is just no physical room for you to mount them.
I feel like using an eGPU enclosure is good enough
Doable, but overtly complicated. As I mentioned, Framework had it designed so there isn’t much opportunity for you to have the … even the PCIe lanes for the GPU. Remember the chipset have 20 or 24 lanes, 16 of which is taken up by the four Thunderbolts and 4 additional ones by storage (and more by WiFi/BT if applicable) and thus even if you mamaged to shove in a dGPU there’d be other things.
To be fair. I don’t think all four need to be thunderbolt. The others can be USB3.2+DPalt+PD ports or simply USB ports altogether. This will allow the allocation of PCIe to a discrete GPU, but a redesigned mainboard is still very highly a necessity, although not necessarily a bad thing consider the limitations the current mainboard have.
Yes, framework have made the most repairable laptop ever (basically that), but the laptop isn’t necessarily performant or competitive to other slightly less repairable designs.
Again, throwing out the Latitude5520/Precision3560 example.
That machine (link) have two Thunderbolt, 2 USB, a MicroSD, Ethernet, and HDMI (and obviously headphone jack), for a price similar to Frameworks. You get touchscreen, two NVMe slot, WLAN and WWAN slots, and the option to carry dGPU, huge battery and smartcard readers.
If framework is to release a 15 inch, they would be designing a big board with all of that (or at least, most of them) in mind, and just put their “built-in adapters” on the side. Not too difficult. But if they say “we are never going to release a 15 inch so to waste money and time to compete with others”, fair argument.
Yeah but since the speed doubles every gen, thats 32 lanes of 4.0, so plenty of lanes available to give 4 to storage and say 8 to a dGPU and 4 lanes for each expansion for a total of 28 lanes of 4.0, leaving 4 lanes for WiFi or whatever else
16 lanes of 4.0 would be less than useful if a dGPU is wanted
dGPU mostly all utilize a x8 interface, so that’s that.
4 for each expansion: as I mentioned. Could use less if absolute necessary.
4 lanes for WiFi: Afaik Wi-Fi cards (even the high-end killers) aren’t demanding and a x1 is sufficient.
SSD: As a person that think any NVMe trying to sell to you with “high performance” is selling snake oil, I think a x2 is as good as a x4.
The product brief mention having 8x gen 4, beside the 16x gen 5, which is very intriguing.
I don’t know how you would split the gen 5 lanes to gen 4 without a mux/chipset of some sort. This take up room and board complexity.
Very interesting mixed bag of beams. Integrated USB controller, too.
They don’t have to make the two size compatiable, but that would be a “nice to have”. Especially if people continue to expect the 15 inch to have beefy performance. Which they should, there is no point in making a glorified 15 inch Chromebook that runs windows
I would imagine having the small PCB offset to one side and have some “USB-C extender” on the other side. Which is another problem by itself as more complexity (on signal traces) = dirty signal and Thunderbolt dont like dirty signals.
But, but. As much as we are here and cry out “framework make a beefy 15 inch model with better everything”, we need to remember that Framework is a company that dont own the tools to make the laptops – this current one is more of a contract between framework team and a OEM, since equipment for making parts (for the laptop) is a tremendous cost paid upfront.
PCB fab, too. Maybe if they have made enough profit + reinvest + more attention from investors they will be able to make more cool and super cool stuff for more people.
Consider that, the current 14 inch suddenly looks quite good, actually (which it is). Since other “OEM” laptops are very trash
Concerning the talk about GPUs in a 15" laptop.
One thing to keep in mind is of course the end user. I assume that most people her can change their stuf in their PC.
Also for the other 99% users they will not WANT to do it theirselfs. And if some want to do it the dongle thingin seems a very good inovation. USB C is a proven consumer grade technology.
I have seen MXM boards in previous laptops, they would be the main cause of the failure of my laptop system (connector beeing loos)! Making things reparable can make you slide on less robust hardware (think fairephone 2). It was broken in 12 months although I never moved it
Now imagine an average enduser playing with MXM boards will sign the end for framework IMHO as support will explode. Moreover I have never seen a commercialy available MXM board… their must be reasons why MSI HP …never did.
That isn’t terribly constructive, if you want to get your point across, elaborate on it. You haven’t even stated what aspect ratio you would prefer and why.
Dell Precision.
The problem is that most connectors (include RAM connectors) are not that secure and sometimes vibration or shock can dislodge a otherwise good connection. In that case simply reseat the module will help.
Or, like Dell, come with a innovative mxm connector in their Precision 7730.
They have little “ribbon cable bridge” that screws onto both the motherboard and the card. One bridge per “connector”. Not all cards use all 4 connectors (I think it have to do with external video back-feed over thunderbolt)
There’s nothing preventing you from engineering things to make them modular and replaceable. It’s just whether you want to or not.
Why will the support explode if the user is given materials to educate themselves to troubleshoot and repair their own devices?!
yes, but it also make it cheaper, at least in that case. Plus, I don’t think modularity is a excuse of using cheap parts. It’s a excuse to be able to support/maintain the same machine for 5, 10, or 25 years. Depend on application/industry of use.
It’s back to the same “why repair” story. Because things will break. And when it’s broken you can at least, replace the broken part with a good part. Just like how a broken piston ring/bent push rod shouldn’t end a car.
Our school still rocks Pentium M Thinkpad T43p? And here is the maintainance manual, freely available. Archive org.
16:9 – 2560 * 1440
16:10 – 2560 * 1600
(screaming on top of their lungs complaining about the lack of 160 vertical pixels)
¯\_(ツ)_/¯
Because its only a difference of 160 pixels, I can’t care more than 0%. However, 3:2 at 2560*1706 is just a waste of pixels. And I will protest about it if I have difficulty seeing things at native resolution with no scaling.
