it will be. but I think releasing a not working PCB design is not a very responsible action for the open source community. it is link dumping trash on the internet. I will release it once it worked. and let other’s improve it.
3 Likes
We can review the design in closed group, I can volunteer for that. We could for example have a discord for it.
2 Likes
I would think at this point Framework would notice the length of this tread and understand how much the community wants this solution and the work we are putting in that they would step in and release an official oculink 8i board to give us some goodwill for not having any updates for us.
6 Likes
Sorry no time, need to sell out and reek in that sweet sweet AI money
but still the board is flawed because the fan is not working tho… Working on revisions now.
7 Likes
Fantastic, really looking forward to it.
1 Like
Like many others here, I have been in this from the very beginning and could not agree more. It is truly mind boggling how long this has been going on. I for one have, after many attempts, different EGPUs partly from China, over several years slowly managed to implement a very good OCuLink 4i solution that looks very clean. And it was worth the effort. But it showed me how difficult “forcing” technology to work is, and how unexpectedly lengthy.
Framework is my favourite company in tech, but I keep realising that this is more and more because of the community. Which is a real shame. Because the degree to which Framework as a company is unresponsive to its Framework 16 laptop community is getting very surprising.
And I have to say it. If we come to the year 2026 without a significant update for this laptop, then I would consider it a failure.
1 Like
My goodness, that’s so amazing! I cannot overstate how much “I want one” once we get there ;).
1 Like
100% - If we don’t get an RX 9070m or similar by 1h 2026, I don’t think we ever will. And the whole idea of modular graphics was kind of pointless.
I’ve been lurking this thread for a long time and considering turning off the notifications because it’s depressing.
Yeah, but the community is making progress. After all, it has led to the only Western notebook that can have a clean OCuLink 4i connection without damaging your notebook. And it will lead to the only notebook in the world with an OCuLink 8i connection. Think about thunderbolt 5, people have talked about it for a decade, and where are we? I think a Razer Blade 18 and an ASUS EGPU.. lol.
What really surprises me is that Framework as a company doesn’t care about such a low-hanging fruit. I mean, as a marketing staff at Framework I would seriously bite into my table. The Framework 16 could be the ONLY notebook in the world with OCuLink 8i, FASTER than Thunderbolt 5 which the entire industry has been chasing for a decade - and they don’t do it… Fastest notebook connection in the world - desktop speed in notebook format. And in my opinion they should have built an 8i EGPU. The company would have an unique product.
1 Like
Well, eGPU is a pretty niche thing to begin with. OcuLink is probably less popular than Thunderbolt due to the lack of hotplug and signal integrity not being a guaranteed thing. And OcuLink 8i is another small niche within a niche (OcuLink) within a niche (eGPU). And if you count Framework 16 as another niche, you could see how this achievement might not matter much, even if shouted to the world via an LTT video. Would it drive sales in a significant way? How many people would actually get an 8i setup? There aren’t many 8i eGPUs available, so many people would graduate from complaining about the lack of OcuLink expansion bay board to the lack of the eGPU adapter which is tested and warrantied by Framework. And suddenly Framework has even more engineering burden without, I’d argue, much of a payoff…
3 Likes
I know, I know.. ;). But I just wanted to also ask for an 8i board to come with the dual m.2 board they made - it can’t be so difficult to build it… ;).
2 Likes
Agree, would be so cool to have a first party solution. Robust, reliable. If ever that happens, I’ll get it on the day of release for sure.
1 Like
And I am more serious about an EGPU in general. It would simply fit the sustainability idea too. To truly replace a desktop. Could also be thunderbolt 5 later on, or OCuLink, whatever. Yes, I will refrain from entering the discussion about the Framework desktop (not sure what the staff over at Overtek is thinking about the achievement of building a 4.5l desktop that CANNOT fit a GPU). Just hope they also consider EGPUs in the future.
1 Like
Agree, an endgame for me would be a laptop with Thunderbolt 5 to use an eGPU as a single cable docking station for desktop use and to enable desktop level modularity. While still having dGPU for gaming on the go. (But for this we have to wait on AMD to deliver a chip with TB5)
3 Likes
Fully agree too. And I get it, it’s a private consumer preference. But still, I personally have the feeling that this would be really good for sustainability too.
Agree, PCIe is the standard, so supporting a PCIe GPU is the most sustainable thing.
But as I’m saying those words, unless you’re running a 5090 with some cherry-picked games, you should be fine with a 4-lane OcuLink, which is achievable already.
And by the time Nvidia comes up with a new 1200W GPU, we’ll probably have TB5 everywhere.
1 Like
the difference is. Pcie 4.0x8 is better than 4.0x4.
even 3.0x8 is going to be better than 4.0x4
even if the 3.0x8 have the same bandwidth, it is not going to be the same as 4.0x4. Think it as a 8 lane road vs a 4 lane highway. 8 lane is going to be less stressed than 4 lane. Even if the speed of the 4 lane is faster.
5 Likes
Yes, I looked it up and I think with my RTX 3090 I am merely looking at a 10% performance increase compared to OCuLink 4i, but any frame counts ;). And to be honest it is also for future proofing, I just want the fastest possible connection if, at any point, it makes sense to upgrade for me too. As you and others have mentioned, a 5090 for example can certainly profit from 8i to reduce bottleneck.