Hey everyone,
I have been working on a serviceable, mid-range laptop (Dell Inspiron 15 3520 series) and chugging along with AddaxAI processing, but as my projects at work have grown, processing time has become a bottleneck and it has become clear that I need a higher performance workstation to speed things up. I also conduct bioacoustics work and run BirdNET to process the audio.
My budget is $3-5k and I am strongly considering a NZXT machine. Either the Player Two Prime or the Player Three. My understanding is that AddaxAI v7 supports GPU usage, and the upcoming BirdNET v3 should too I believe.
Does anyone have workstations that they bought for wildlife AI work and can share insights? It seems that gaming PCs are a good sector to look at. I would like to get some input from other folks in ecology before making a big purchase rather than just relying on gamer reviews.
Thanks!
Hi @elliotjohnston,
Yes, your understanding is right. AddaxAI v7 uses an NVIDIA GPU automatically, and both machines you linked will run it at full speed.
The GPU is what decides the processing speed. For real numbers per card, have a look at the MegaDetector benchmark timings: MegaDetector/megadetector.md at main · agentmorris/MegaDetector · GitHub. Both cards sit at the fast end of that table, so the difference with your laptop will be huge.
Between the two I would take the Player Two Prime. The GPUs are close to each other, maybe is the 5080 some 20% faster. But the cheaper machine has 12 CPU cores instead of 8, and that helps with video processing, which runs partly on the CPU. The extra cache on the 9800X3D is a gaming feature, for this work it does nothing.
About BirdNET I can’t say it for sure, that is not my product. But it runs on the same NVIDIA hardware, so I expect it to work fine. No guarantees though.
Cheers,
Peter
Cheers thanks for your thoughts on these machines Peter. I went with the Player Three given the faster GPU, even if modestly so. My camera trapping work is all photos at this point, but good to know about the CPU usage for videos if we capture those in the future.
1 Like