A Call For Measurements: Isolating the Thunderbolt Effect.
Fry's sent me an update the Gigabyte GA-Alpine-Ridge is back-order and no ETA. I'm doing 1440 and 4K benchmarks in the mean time.
I cannot find the actual quote button. It disappears once you log in. Anyways, I have found with an Asrock Z270M Extreme4 motherboard, the Asrock AIC works, the Gigabyte works, and the MSI Thunderbolt M3 AIC is detected (USB-C ports fell of PCB before I could fully test). I remember somebody saying the Asus ThunderboltEX 3 is locked down or something, but never believe anything you read on the internet, right? It might work, but it is the one I do not own to test.
You probably have it sorted by now, but still, just worth mentioning that the AICs seem to work cross-brand, even if they will never admit to it or officially support it.
Pending: Add my system information and expected eGPU configuration to my signature to give context to my posts
These are the critical CUDA-Z host-to-device (H2D) values I've seen posted here and elsewhere that can be used to gauge relative bandwidth to each other:
x4 3.0 M.2 = 2940MiB/s = 24.66Gbps
TB3 = 2260MiB/s = 18.96Gbps
TB2 = 1250MiB/s = 10.49Gbps
TB1 = 790MiB/s = 6.63Gbps
EC2 = 380MiB/s = 3.19Gbps
I have measured the full 22Gbps = 2750MB/s on Thunderbolt 3 to my eGPU. I know M.2 is 3.94GB/s theoretically, and we have the Samsung 970 Pro advertising 3500MB/s reads. So these are a little off. Just curious though, what is EC2 aside from Elastic Compute Cloud (Amazon)? It does not seem to match the context of ports.
EC2 = ExpressCard 2.0 = PCIe 2.0 x1 = 5 GT/s = 4 Gbps = 500 MB/s = 476.8 MiB/s
Effective transfer speed is more like 400 MB/s = 381.4 MiB/s (PCIe protocol overhead, etc).