For featherweight or lightweight + VGA fans…
Do these things really have to be so expensive?
If a feather-light adapter is going to cost that much, I’d rather use an VGA dongle (for 1/10th of the price) for travel. Or a VGA+HDMI, which should cover most “dealing with a potentially outdated projector/screen” scenarios.
And then, a medium-super for stationary use.
Thinking about a middle-super option now. Could there be a better combination than just creating a middle dock and a separated eGPU?
If I got it correctly from the previous debates, there are two things that would make a middle sized dock or an eGPU expensive:
- (a) The TB3 controller
- (b) The case
If we create a middle dock and then an eGPU as modules or separate devices, we would need two TB3 controllers, making the combination of both devices expensive. So, could we design a middle dock that has the ability to host an entirely optional eGPU without improving the base cost for non-ePGU users?
Maybe, if we overcome (a) and (b).
- (a) By using a single TB3 controller for both: Creating one single device that works as a TB3 dock, but also supports an eGPU connected to the same controller via PCIe. There wouldn’t be any performance drop, no second TB3 controller, and no important added cost to the base dock.
- (b) By literally thinking “outside the box” and using no case (or just an optional case) for the ePGU: We could design the device so that the casing could be added by eGPU owners… or not. This would make the eGPU expansion much more affordable
and open to a growing DIY community. Or you could just buy the extra case and enjoy a polished solution.
Some other people here have already suggested this. Things would look a bit like this:
(1) A base middle-sized dock, with an (unobtrusive) included PCIe connector. As most of the price would go to the casing and the TB3 controller, having a PCIe connector wouldn’t make this much more expensive to non-eGPU users. There would be no integrated PSU, just pass-through USB PD or a basic PSU to power the V and some peripherals.
(2) A GPU + external PSU. If the dock included an easy way to attach it to a VESA mount and hide it away behind a monitor or under a table, I would go for this option, and I think a lot of DIYers would too. For some, an exposed GPU may even look cool:
(3) An optional case for the eGPU with an integrated PSU, for those wanting to have everything in a more protected package… at a significant added cost ($$$). This case would leave all the ports of the base dock exposed, so you would have a great all-in-one solution.
Some problems/challenges for this approach:
- Finding the right external PSU for (2).
- Achieving a good design that looks clean even if “half encased” (2), and solving a way to firmly attach the GPU to the dock.
- Making the dock’s video ports redundant when having a GPU plugged in. But as having them wouldn’t increase the prices significantly, I’d cope with that. And they wouldn’t be totally useless, as they could still be used to connect a monitor to the Intel iGPU.
- Really keeping the price of the base dock as close as possible to an equivalent non-GPU-enabled dock.
- … Any others you can think of?
Some of them are relatively easy or not deal breakers to me. What do you think? Is this too crazy?
- Innovative and very unique solution, very Eve-like.
- Cost-effective, having not only a modular design, but also a truly modular pricing, where you would not be paying for duplicated stuff (except some video ports).
- No extra performance drop for the GPU.
- Attractive to a very wide audience (middle-weight dock users + DIY eGPU users + polished eGPU users).
- … Any others you can think of?
Good idea, but I am worrying that having too many separate modules will cause additional latency…
That is really very expensive for only Single Display Support…
I think it should be possible to build this much cheaper…
Here’s a couple discussions we’ve already had about modularity concepts already. I think Xinjie may have already linked one of them earlier but it seemed like an appropriate time to loop them back in.
Thanks for linking back to those threads, there’s really been a lot of discussion on modularity and eGPUs
You could also link this thread in your post, to have everything together:
In the first one you link @Attiq arrived to the same conclusions: a modular approach for eGPU integration could be feasible if just one TB3 controller was used for the dock and the GPU. I would still add the “case-less” possibility, but there is really a lot to consider from @Attiq’s concept. It’s very forward-looking as it would allow future modules.
I think this modular approach should be really considered for inclusion as an option in the poll… if @Xinjie and the rest of the team see enough feasibility for giving it as an option.
This sounds like a great idea.
Just thinking out loud, if Donald ends up not being the super weight option or having the option to add an eGPU on in a modular way, than it should probably plug in via USB C and not TB3. This way users who wanted an eGPU could still get one and use Donald (if they wanted). Although I have heard some say that an eGPU has enough ports of it’s own that something like Donald wouldn’t really be needed at the same time, only separately if it was a travel friendly adapter.
I think There can be maybe 2 dock types, I will buy one maybe heavier, with more ports for a station at home and maybe have a smaller dock at work or maybe at the living room, or maybe even carry with me to use it the V on trips…kinda like the dock microsoft offered with their last windows phone
The Superweight (or size) option mainly relies on whether we would be to choose supplying a PSU with the system or not, that’s where the main bulk comes from. It also depends on whether people demand to have enclosure around the card, and what kind of enclosure is wanted. For an enclosure with PSU such as the HP Omen Accelerator, the costs add up.
Without PSU (BYOPSU or, if possible, we could think about a sold-separately “PSU brick”), the eGPU compartment will remain small in size and it would not necessarily amp up the cost all too much.
But this kind of lightweight eGPU option would then mean that either people will find a way to connect their own PSU to supply the GPU power, or we will have to sell a separate PSU for the system.
Could you please shortly explain “PSU”? What exactly makes it big? Are the two devices on the photos making the same???
IMHO, Donald can either do very many things really poorly or do a half dozen things very well.
Should Donald be fairly stationary like a traditional laptop dock, or more flexible/portable like a port replicator? Perhaps Donald should bridge a tablet to be a laptop like the Surface Book, or make the V capable of being a great entertainment center like a Bose iPod dock but for your V?
I think a hybrid of 2 of these ideas is aggressive, but compelling. The hybridization is what will draw many people in just like Surface Pro/Eve V did.
I like the idea of melding the fun of a Beats audio pill experience with the utility of a device charging thunderbolt port replicator. All the tech is stable, compatible for integration and can remain under a $250 price point whilst still maintaining a decent level of portability. I think Eve has a real game changer if they can hit a $199 price point. Probably possible with some creative engineering on the power electronics for charging and audio amplification.
Also, I’m a user who needs an eGPU. In no way do I think Donald should be have an eGPU capability.
I have a half dozen options and growing to get an eGPU. Let’s just make sure Donald doesn’t impede on an eGPU connected via TB3.
Power Supply (?) Unit (where you plug in power cord for your desktop, e.g.). Something that is enough for majority of the GPU cards ends up bulking up the size comparing to anything middle size/weight.
IMHO middleweight option is already about the size of a PSU, meaning that anything with PSU would already double the size and weight, putting it to heavyweight category?
Additionally if the card needs to be housed inside a case, the volume of the housing is dictated by rather the biggest GPUs out there than the smallest (You’re surrounding the space needed by the card by a case).
Easily ending up with a bulky box that doesnt resonate to people who just wanted a dock and prohibitively big to most who want it to be mobile.
Take both PSU and the enclosure out, though, and the dock and still support a GPU (it just needs PSU, and the card will be visible)
Does a GPU card always need a fan, even when its “free in the air”???
yes, but the retail GPUs already have fans in them. Other fans are only to remove the hot air from the closed system, and replace it with room temp. If the card is bare, the air is more or less always room temp
We need a nice “propeller”!!!
It needs at least a cooler as thermal output of modern chips is far too high to be cooled passively - even passive coolers for your PC case usually recommend one case fan for a slight airflow.
But this is going a tad offtopic - for further elaboration please consider making a thread
I personally would never pay this much for a dock, especially right after a major purchase like the V. If Donald has an eGPU then I understand the price being this high or higher, but a nice middleweight port extender should not cost anywhere close to this.
Unless Donald has a bunch of other features (other than ports and power) that actually raises it’s value, I can’t see myself paying more than $100. If I’m being unrealistic, then I apologize and I guess I’ll just live with the V’s great port selection.