Hehehe, I meant to take only 2-3 pictures but got carried away and have about 20-30 of the whole thing instead.
I have the ASRock Fatal1ty Z270 Gaming-ITX/ac coming in for review so I'll just go ahead and do a review of the AKiTiO as well.
-
Save 15% on ALL SFF Network merch, until Dec 31st! Use code SFF2024 at checkout. Click here!
An eGPU dock that's actually SFF (name TBD)
- Thread starter jeshikat
- Start date
Hehehe, I meant to take only 2-3 pictures but got carried away and have about 20-30 of the whole thing instead.
I have the ASRock Fatal1ty Z270 Gaming-ITX/ac coming in for review so I'll just go ahead and do a review of the AKiTiO as well.
The reason I thanked you is because I love where this project is going and wouldn't be possible without someone like you "sponsoring" the EGPU box.
Sorry for the late notice, I wasn't sure of my schedule until now. But I'll be doing a short stream starting at 4:00pm Central (click here to see in your local time zone).
Ok, Episode 2 is now uploaded:
Episode 3 (the one I recorded yesterday) is exporting from Premiere now.
Episode 3 (the one I recorded yesterday) is exporting from Premiere now.
I'm happy to see that enough people are bothered by the size to start a project like this. I've only recently gotten into SFF and ended up wanting to go for the (controversial) eGPU + Skull Canyon NUC setup. The problem was that the eGPU docks out are about as big as an SFF case which is ridiculous. (Like the Razer Core vs DAN A4-SFX)
I'll be following this thread closely, in fact I signed up just to follow this thread.
I'll be following this thread closely, in fact I signed up just to follow this thread.
Which means that the dock plus and ITX APU build will be around the size of the A4. Sounds pretty reasonable to me.
Is there any chance of desoldering the PCI socket from the Aiki..whatever PCB to replace it with a smaller socket?
Or just straight up take out the dev board (cause apparently it's cheaper to buy the enclosures then it is to buy the dev board?) and do a custom PCB for the PCI slot? Then you could have a standard "vertical" PCI socket. How difficult is it to design for PCI? Given that risers are a thing I'm going to assume that timings aren't too important?
The AD-DC lines up with the GPU thickness so incredibly well that it'd be a shame to waste so much space just to re-use the (not-so-complicated-looking) secodary PCB
Or just straight up take out the dev board (cause apparently it's cheaper to buy the enclosures then it is to buy the dev board?) and do a custom PCB for the PCI slot? Then you could have a standard "vertical" PCI socket. How difficult is it to design for PCI? Given that risers are a thing I'm going to assume that timings aren't too important?
The AD-DC lines up with the GPU thickness so incredibly well that it'd be a shame to waste so much space just to re-use the (not-so-complicated-looking) secodary PCB
Timings are critical, but it's the relative timings between pairs that are important. Increasing total length while keeping all traces the same length makes you vulnerable to interference and eventually resistive losses (managed by using shielded risers and limiting total length and number of connectors in the chain), but if you fail to keep the path length the same for all lanes then the link won;t work at all. Custom PCIe risers can be designed, but it's not a simple as just linking the right pins together.
My warped mind sees this:
Maybe incorporate something rougly resembling optical illusions ? To show off the mindbending spatial magic going on.
(love perspective grids)
Maybe incorporate something rougly resembling optical illusions ? To show off the mindbending spatial magic going on.
(love perspective grids)
That was the idea, but it turns out trying to turn optical illusions into vents that are functional, good-looking, and manufacturable is difficult 
Similar threads
