Setup:
MBP 16" 2019 Base, i7, 5300m
Lg Ultrafine 5k
Zotac RTX 3060Ti Driver: 460.89 (latest)
Hello all, I'm having trouble getting acceptable performance from my new zotac rtx 3060ti. Previously I was using a rtx 2060 and got expected performance using loopback mode, within 10% of desktop numbers. My setup looks like: Lg Ultrafine 5k -> MBP in loopback mode. Egpu -> MBP.
With my new rtx 3060ti, I'm only getting +10% compared to the 2060 which translates to negligible gains in games. Performance is almost identical whether at 1080p or 1400p or 4k. I'm monitoring my gpu and cpu usage and GPU usage is always at 90%+ and CPU usage usually hovers around 50% in benchmarks so I shouldn't be CPU limited.
RTX2060 heaven benchmark: 2100 -- 1080p extreme 8AA,
RTX3060Ti heaven benchmark: 2500 -- 1080p extreme 8AA
Things I have tried:
* drivers that came with the rtx 3060 with no update
* uninstall bootcamp and clean install, but yielded the exact same numbers.
* use different thunderbolt cables
* using just the laptop monitor, no ultrafine, exact same numbers
* Limiting CPU power and unlimiting it. Exact same performance whether it's running @ 2.5ghz or 3.9ghz
Starting to think this Zotac is a POS, sadly there's not really any stock readily available for me to try a different GPU. Also don't have my desktop anymore so I can't plug in the GPU to a desktop.
Any suggestions or anyone else with a GPU similar to RTX 3060Ti can vouch for loopback performance numbers?
To do: Create my signature with system and expected eGPU configuration information to give context to my posts. I have no builds.
.
@theobserver, Loopback performance is not great and can differ between cards. For example, Radeon Polaris (RX 580) performs great but Navi (RX 5700) loses a lot more performance in loopback mode. This could be a similar situation.
LinkedIn | Twitter | Facebook | Youtube | Instagram
• external graphics card builds
• best laptops for external GPU
• eGPU enclosure buyer's guide
2021 14" Microsoft Surface Laptop Studio [11th,4C,H] + RTX 2080 Ti @ 32Gbps-TB4 (WD_Black D50) + Win11 [build link]
@itsage, thanks for the insight! I saw several threads where loopback performance was only within 2%-10% of non-loopback mode in higher resolutions, still can't reach those numbers on my setup though. When I get a non-loopback external monitor/ better RTX30x0 card I'll report back on this thread with some results
To do: Create my signature with system and expected eGPU configuration information to give context to my posts. I have no builds.
.
looking forward how exernal monitor is improving the output 
i am running 5600xt in loop back mode on my laptop
Definitely something more than just loopback performance is going on here.
Here's my setup:
iMac 27" 2019 MRQY2LL/A upgraded with M2 NVMe SSD Barracuda 1Tb and 40GB RAM
iGPU - Intel Graphics UHD 630 - at present without drivers (Microsoft Basic Display Adapter)
dGPU Radeon 570X 4GB VRAM - at present disabled on the level of Intel(R) Xeon(R) E3 - 1200/1500 v5/6th Gen Intel(R) Core(TM) PCIe Controller (x16) - 1901
eGPU - nVidia RTX 3060 12GB miniATX in Aorus Gaming Box (upgraded from GTX 1070) with external monitor 17" LG running miserly 1280*1024
loaded with help of automate-eGPU EFI.
Will report more after I run heaven benchmark and try some more tricks
Performance feels even worse than 1070 was giving me. I suspect that has something to do with iGPU drivers not being installed and thus some burden on video switching internally on OS level instead of Optimus, although should not really be an issue with external monitor.
GPU appears to be 90+% loaded and CPU hovering around 60% same as @theobserver described.
MSFS 2020 barely pumps out 30 FPS and I'd rather stay silent about (~10 fps) VR that definitely requires loopbacking video over Thunderbolt.