Thanks, I thought I was forgetting something. I had that interposer sitting in a drawer for the last year.
Ahh… I recall seeing something about that, but for some reason was thinking only the 1x8i mode was affected. I’ll roll back and give it another try.
I’m seeing some strange results with the new board,
Windows sees the GPU, but GPUZ shows some odd results, and I can’t get an external monitor to work with it. Anyone know whats going on?
Most likely nvidia error 43. Just requires running a fix script available on egpu forums.
Thanks, I’ll try that out when I get back on Friday
No way. This is the same issue as my pcie x16 board. What do you need to do to solve it?
Updating in this thread as well as the “update thread”.
I have sold and shipped the last adapter. I hate that the manufacturer did what they did, but we should be kind to our fellow Framework OCuLink users who do purchase from the manufacturer’s listing.
I’m sure the manufacturer has sold a few as well, so I am confident that my adapter is in (or has been in for those that have switched to Filip’s adapter) over 150 laptops.
2 Likes
Appreciate you Kyle, I constantly go between my oculink setup or the 5070 now.
Can’t wait for that oculink kit from FW at this point
1 Like
Rolled back to 3.07. No joy. I am reasonably confident the cable is the issue, but I cannot know for certain until the other cable arrives. Sadly, tracking says it’s still a week out. I guess it’s OK because I have a lot of other stuff going on right now that I really should be focusing on. 
Go into device manager and open the GPU. If it says something like error code 43, its definitely that and you just need to run a fix script available over at eGPU.io forums. You can easily find it by just googling.
Note: this is specific to Nvidia GPUs on Windows, if you were to run Linux it would just work without any kind of extra step.
Device manager says the device is working properly however i still am having the “ghosting” issue where GPU is visible under display adapters (in device manager) however it cannot output a display signal nor does it appear properly in GPU-Z
That seems like a different issue. Most likely something funky going on with the link, dunno what though.
1 Like
I’m encountering a similar issue to yours however its not the nvidia error 43. Please let me know if you are able to solve it
My 8i cable arrived much sooner than anticipated. Obviously, I couldn’t resist not doing the things I should be focusing on to take about 30 minutes to confirm things are working.
Flashed the board, flipped the necessary switches, and… well, it didn’t work at first. Then I cycled the power on the “eGPU” and timed the power-on just before powering on the FW16. I pretty much knew right away things would work because LEDs started flashing pretty much right away on the PCIe M.2 adapter board I test with. Sure enough, once logged in I was able to see the SSD installed in the eGPU (or eSSD? LOL).
For giggles I went back to the 2x4i setup and tried the same process. Still doesn’t work, so I’m pretty sure it’s the cable.
Sadly, I have to get back to other things now and probably won’t have much time to test over the coming days, but here is another sneak peek at the enclosure.
It’s hard to see, but there is a slot for the OCuLink cable in the orange section on the right side directly above the PSU. This allows for a setup where ALL cables are coming out the “back” of the enclosure. I will plan on having a similar slot on the “front”.
3 Likes
Looks great!
This is the reason why I want to switch the board to 2x 4i connectors in the future instead since the dev kit will be THE x8 setup most likely. Finding good quality Y-cables is a pain.
1 Like
what?
Damn. That’s rough, man. Though I dont think its entirely theft. But it definitely became part of the collection of the PCB manufacturer.
Perhaps, its also just mere coincidence. Someone probably saw it and was “inspired”. Its not that hard a design, minus the high-speed bit.
I will have to look at the trace layout, but I believe its indeed different.
.. yeah from the picture the trace look different. Yours seem to attempted to length match between the pairs, which is not necessary.
And yours have resistor. They dont. Maybe its one of the “failed” prototypes?
Hey there. I explained this in other posts. The pictures on the listing are identical (not similar; identical) to the original design the manufacturer produced for me, and not one of those samples worked. The updated design with the matched length pairs worked.
I ordered an adapter from the listing, and it is actually the updated design (again, identical and not just similar).
When I say “identical”, I mean right down to what’s printed on the PCB.
I agree that calling it “theft” isn’t completely accurate, but the measurements came from me, the money for producing the initial test pieces came from me, and the testing was all done by me (the manufacturer didn’t have a FW16). The manufacturer put in a small amount of effort and took on exactly zero risk but is reaping the rewards (hopefully not many now, but who knows).
5 Likes
Lamo they stole the broken design.
Really. Not working? I thought I had ran some stuff off of it.
Not gen 4, no, but.
How can I get my hand on one of your newer ones? Your site says they are out of stock