After installing it from the app store, it tells me that my driver is not compatible with the control center. I wanted to Check the NVidia Control Center for CUDA support, but it wasn't installed. However, calling tf.test.is_gpu_available() from python CLI returns False. It installed tensorflow-gpu and necessary packages, so it seems they weren't installed. I manually activated the "faceswap" conda environment and installed tensorflow-gpu via pip. How can I make sure the GPU is used for training? I chose the "original" trainer and "Gpus" ist set to 1.Īfter starting, "Setting Faceswap backend to NVIDIA" is displayed, so I guess faceswap is configured correctly. However, when I start training, it is even slower than on my old machine (currently 3,2 EGs/sec, old machine 4 EGs/sec) and the GPU usage remains below 4%. I chose NVidia-Support, and HELP -> Output System Information shows a RTX 2080 Super as GPU_0. I was impatient and just installed faceswap via the installer on the pre-installed Windows 10. Before I played around with faceswap on my old Linux Machine (just a quad core i7) to get the hang of it. Server.io_loop.add_callback(lambda: send_data(table.I just got my new PC with a GeForce 2080 Super. Server = add_perspective_tables_to_panel_server(server, tables) # some app (a data source, a "per-user" part, and a "server part")į())] So I’m sure some stuff is wrong (not just the missing perspective theme/style) because the world seems to have changed a lot since I last looked # Note: I had never used python 3.9, panel 1.3.1, or perspective 1.9.4 (or tornado 6.3, or bokeh 3…) until making this post (I am stuck using many years older versions of everything).
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |