(no title)
philsquared_ | 1 year ago
I think as the technology adapts, the utility will be having chips on your existing hardware that can provide this functionality. Will LLMs always need super high end gpu's to process requests or will the algorithms improve enough to allow quick speeds with lesser hardware? I will tell you one thing for certain. NVDA's stock price is not based on any possibility of algorithm improvements and only based on the idea that more and more resources will be needed. It's based on the idea that there will always be the need for massive data centers to process these requests..
No comments yet.