I had a question about the GPU necessary to run NVIDIA’s Jarvis SDK. My laptop has an RTX Quadro 3000 with Max-Q design - this has 6 GB of GPU memory. The documentation for the Jarvis Support Matrix says that users need at least 16 GB of GPU memory to run all the models (ASR, NLP, and TTS).
I have gotten Jarvis to run on my computer by commenting out everything except the TTS model in the
config.sh file that comes with the Quick Start Guide (the TTS model only needs 3.5 GB of GPU memory). My question is - is the 16 GB strictly necessary? For example, will the Jarvis speech server simply not start if it doesn’t detect a powerful enough GPU? Or will preparing the models simply take a long time on my computer?
My apologies if this is a question with a known answer.