Hello everyone, I'm not sure how many of you are attempting to add an external GPU, but I have been trying for a while now to get the Bolt to detect the graphics card with no luck. Advice and tips from other online forums have not helped so I thought to ask the UDOO Bolt community what suggestions you all might have. I am not an expert with computers, but I like to try to learn in my spare time, this issue has me stumped. I am using Windows 10 64bit OS and I intend on using an ASUS Dual GTX 1070 8GB, and I have attached the following to the Bolt. https://www.amazon.com/gp/product/B074Z5YKXJ/ref=ppx_yo_dt_b_asin_title_o02_s00?ie=UTF8&psc=1 https://www.microsatacables.com/pci-express-4x-male-to-pcie-16x-female-riser-cable-pci-e16x-4x Any help resolving the issue would be greatly appreciated. Thank you.
You need to supply 12V power to the adapter (notice that 4-pin Molex connector???) You can find some pictures in this BOLT as well as UDOO x86 forum.
Thank you for your response. I apologize for not mentioning earlier that I do have a separate psu attached to the 4-pin molex connector and there is an indicator light that illuminates on the adapter when the Bolt is powered on.
First, sufficiently powering the GPU is a necessary condition... 1. If the GPU card has its own 4-pin (or 6-pin) power receptacle, 12V power definitely needs to be supplied there. 2. PSU must have enough 12V ? Amp power supplied to GPU, per GPU card's requirement 3. For troubleshooting purpose, do not use the second extension board as pictured in post #1. Just use the first adapter (key-M NGFF to PCIe x4) and plug GPU card directly into that adapter. Note: the second extension cable is not meant to carry high current at all, it's just for PCIex4 signaling only.
Hello @css_hello Which of the M.2 is the best for the external gpu setup, I ask because right now I'm using my 2280 Key M port with an NVME, should i use the 2260 Key B one? How the performance will be affected from one port to the other? Thanks
Main difference is BOLT's key-M has 4-lane PCIe while key-B only has 2 (latter see my UDOO x86 thread, need to cut a "B" slot to fit). The real world difference might be minimum, if it is not dataflow-bound (often times it is not.) Think: if one can live with reducing the max. of 16-lane PCIe (in a graphics card application) down to 4 (in the case of key-M), what hit will you feel the pain when down to 2?
Will try the 2 lanes way, lol with the approach of the 1650 GPU and the pico PSU, let's see what happens. Thanks
@css_hello, Thank you for identifying the issue with the x4 to x16 ribbon. I will replace that with a new riser this week.
