@ Elliot, I assume by ‘bay’ you mean how I am connecting the rx7600 pciex16 to my computer. The path is rx7600 GPU onto a adt-link k43sg which outputs to a nvme card ( I am having to daisy chain this to a usb4 rated nvme enclosure as the framework mianboard only has 1 nvme slot and it is occupied by the ssd with my OS on it). In hindsight I should have bought the ut3g. The bay has the option for lower powered GPUs to be powered by the bay but given the power requirements of a rx7600 I am powering the RX7600 directly from the 550w atx psu (using an 8 pin connector). the k43sg itself gets power from a second power line of the same atx 550w psu. I could double check but I think when I used one of the power meters on the connect between the mainboard and the external GPU (specifically it was the connection to the nvme enclosure) it measured about 3w (I don’t think my power meter can sense direction so I am guessing that was pulling power from the mainboard).
Update: as I had some time to do some more testing. By going into device manager an disabling the integrated GPU (AMD 760m) so that only my eGPU AMD rx7600 is enabled I am able to get slightly better performance but a crash still occurs. Note that the way I am testing is with a game called “Dakar desert rally” as it seem to to be fairly constant in it’s crashing of my system ( it always happens in the fist tutorial stage and I can easily tell when performance is a bit better because I get further on the course). additionally if i drop the resolution to 18 from 1440 the games worked significantly longer but did eventually crash
I would still keep the iGPU enabled just in case you need to power it on without the dGPU because there wouldn’t be a GPU for Windows to use if you were running it.
Can you try to check Event Viewer in windows to see if it caught any errors?
I’m suspecting that the NVMe to USB 4 enclosure you have might be the issue. Do you have any details on the model number of the enclosure or what chip it is using?
below is the enclosure I am using
https://www.amazon.com/gp/product/B0CR3PY8YL/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&psc=1
I can force a few more crashes and recheck for events but have not been seeing any logs in the past ( usually I will just see the errors as it start back up lik 'system restarted from unexpected shutdown")
Does the enclosure get very hot when running? After doing some research it seems like a pretty good quality enclosure except for the fact that it gets very hot.
(source:https://www.reddit.com/r/UsbCHardware/comments/1ea60n9/nvme_or_not_to_nvme_is_the_question_enclosures/)
Also the controller used in the enclosure seems to be rated mainly for NVMe SSDs although I’m not sure what effect that would have. A reddit post says that the controller used in the enclosure is the ASM2464. Although doing some research shows that there are two revisions of the controller. One seems to be designed mainly for storage devices (ASM2464PD) while the other one is designed for both storage devices and other PCIe components (ASM2464PDX). The PDX model allows for an external PD controller although not sure what it effect it could have.
@ Elliot, thank you for the research into this…a heat issue at the nvme enclosure seems very plausible.I managed to force a shutdown and immediately checked the temp with my heat gun (showed 33 to 37C). I am unsure if this is hot enough to be considered an issue. I also looked at the event log and saw “critical: The system has rebooted without cleanly shutting down first. This error could be caused if the system stopped responding, crashed, or lost power unexpectedly”. if possible I would like to confirm that the enclosure heat is the issue before dishing out $110 for a ut3g that would eliminate the enclosure in the system altogether.
I can’t confirm this but 37C seems alright to me. If it has a crash log, then install WinDbg from the Microsoft store and try to get it to read the crash log.
I had some critical errors before on another device and using WinDbg helped me trace it down to a memory issue and it no longer occurred after swapping out my memory sticks. If you want, you can either post the output from WinDbg here on the forum, or do some research online yourself.
unfortunately there are no real logs beyond the one above but that is clearly created after the system has already restarted, not when it crashes. as I wrote this my system crashesdin the manor it does when gaming. this is the first time it has ever done this without a game running. nvme enclosure as the connection was about 40c.
Do you have any kind of heatsink that you could put on the enclosure to cool it down? Try keeping it cool and seeing if the crash behavior still occurs.
I dont really have any kind of hea sink that would work for this because of the way the connect is made at the eng of the k43Asg (see pic).
As much as I wanted to wait, I did pull the trigger on getting the ut3g. interestingly the aliexpress seller had two versions, 1 with ASM2464PD and 1 with ASM2464PDX. I ordered the latter and is should be here between 10/7 and 10/12. Once I get that in I can do some testing to see if the issue is fixed. What I am now expecting is the issue is that the k43sg is designed to deliver 64Gbps whereas the enclosure was only designed for 40Gbps and is heating up then crashing to prevent damage.
Know anyone interested in a k43sg for sale??
Hopefully this resolves your issue. If any further issues pop up, feel free to respond to this post or make a new one.