For OpenAI, I’d assume that a GPU is dedicated to your task from the point you press enter to the point it finishes writing. I would think most of the 700 million barely use ChatGPT and a small proportion use it a lot and likely would need to pay due to the limits. Most of the time you have the website/app open I’d think you are either reading what it has written, writing something or it’s just open in the background, so ChatGPT isn’t doing anything in that time. If we assume 20 queries a week taking 25 seconds each. That’s 8.33 minutes a week. That would mean a single GPU could serve up to 1209 users, meaning for 700 million users you’d need at least 578,703 GPUs. Sam Altman has said OpenAI is due to have over a million GPUs by the end of year.

I’ve found that the inference speed on newer GPUs is barely faster than older ones (perhaps it’s memory speed limited?). They could be using older clusters of V100, A100 or even H100 GPUs for inference if they can get the model to fit or multiple GPUs if it doesn’t fit. A100s were available in 40GB and 80GB versions.

I would think they use a queuing system to allocate your message to a GPU. Slurm is widely used in HPC compute clusters, so might use that, though likely they have rolled their own system for inference.

[1] https://history-computer.com/palm-pilot-guide/ [2] https://www.zdnet.com/article/pocket-pc-sales-1-million-and-...

The latest Nvidia driver no longer supports the K40, so you’ll have to use version 470 (or lower, officially Nvidia says 460, but 470 seems to work). That supports CUDA 11.4 natively. Newer versions of CUDA 11.x are supported: https://docs.nvidia.com/deploy/cuda-compatibility/index.html though CUDA 12 is not.

In my testing, a system with a single RTX3060 was faster in tensorflow than with 3 K40s and probably close to the performance of 4 k40s.

If you are considering other GPUs, there are some good benchmarks here (The RTX3060 is not there, though the GTX1080Ti was almost the same performance in the tensorflow test they run): https://lambdalabs.com/gpu-benchmarks

As others have said Google CoLab is free option you can use.

Interchangeable lens cameras now all have video features and increasingly most of the improvements are in that area. When SLRs are used for video the mirror needs to be flipped up and auto-focus system that is used for photos can’t be used, so the camera need another one on the sensor. In this case the mirror is redundant and the viewfinder can’t be used.

Tracking of fast moving subject is difficult with a SLR, the SLR cannot see the image in viewfinder mode only a focus module can, which likely only has a few hundred focus points (or less) and those points often don’t reach the edge of the frame. Additionally mirrorless cameras are able track a subject eye using AI and keep that in focus. A SLR cannot do this in the viewfinder mode as the focus sensor does not have nearly enough resolution to recognise small item like an eye or to know that it is an eye.

Burst shooting is also difficult on a SLR, for each shot the mirror needs to flip up and down and the focus module use a brief period to change focus. Canon is/was the leader in sports photography cameras. The highest end Canon SLR camera can do 16fps with autofocus, but 20fps with the mirror up. The Sony a1 (mirrorless) can do 30fps. These fast shooting rates are only possible with mirrorless cameras.

Phone apps seem to be modelled on webapps, typically the server is run by the app maker, installs are easy and updates are automatic. Additionally, phone apps automatically go on the home screen and have notifications, which means you’ll open them more often.

You can also unlock using a tang server: https://www.networkshinobi.com/clevis-and-tang-network-bound...