2019 15" MacBook Pro (Vega20) [9th,8C,H] + RTX 2080 Ti @ 32Gbps-TB3 (Mantiz Venus) + Win10 [csn666]
- 2019 15'' Macbook Pro
- dGPU: Radeon Pro Vega 20
- CPU: Intel i9-9880H
- Win10: Installed with bootcamp
- Mac OS Mojave 10.14 (eGPU only available on Win10 because I'm using NVIDIA card)
- no external monitor, internal display only.
Mantiz Venus + GALAX HOF 2080 Ti PLUS + Mantiz TB3 cable
1. Assemble eGpu.
2. Installing win10 with Boot camp assistant.
3. Restart with win10
4. Uninstall AMD drive in win10 with DDU and reboot
5. Download the AMD drive from the official website, install and then reboot
6. Plug-in the eGPU, download the official NVIDIA drive, shut down the laptop
7. Un-plug the eGPU, boot the laptop with win10 and log in.
8. plug-in the eGPU, wait several seconds until the OS recognizes it
9A. To use the eGPU for gaming, search Display in start menu, find Advanced Graphic Settings, and browse your game executable file(*.exe). Set 'high performance' for it, which means the eGPU will be used.
9B. To use the eGPU for deep learning, you should download and install cuda & cudnn.
Host Pinned to Device: 2140.24 MiB/s
Host Pageable to Device: 1769 MiB/s
Device to Host Pinned: 2704.39 MiB/s
Device to Host Pageable: 2118.88 MiB/s
Device to Device: 209.855 GiB/s
Comments (eg: how has the eGPU improved your workflow or gaming)
The performance loss due to TB3 is obvious, but my laptop is now like a gaming PC anyway. I can now train DNNs(deep neural networks) much faster.
Pending: Add my system information and expected eGPU configuration to my signature to give context to my posts