After playing around with Ollama platform, the API does not use the GPU or at least performance is not listed in Task Manager in Windows 11. After researching and shopping for new cards, the NVIDIA RTX 4090 is the best choice for me. The RTX Workstation cards offer performance that is half the 4090 but with the same amount of VRAM for around the same price. The clear solution for AI is a GeForce card… There are faster RTX Workstation cards, but are higher than my budget of $3,770. If my budget was $5,000 greater, I would include a few workstation graphics cards to increase the amount of fast GPU memory in a server implementation. Keeping with the original budget, the plan is to invest in a 256GB RAM upgrade incase offloading layers from VRAM is not much of a performance hit.

Regardless, the RTX 4090 is still a GPU just made for real-time renders and less for training AI models. And I have a number of processes on my list:

• Gather News articles
• RSS Feeds
• Import Wikipedia articles
• Import Other Wikia articles
• Visual Reasoning Folding
• Image2Text
• Text2Image
• LLava and Wikipedia Image Parsing
• Books Operations
• Books Concept Recognition
• Books Context Recognition
• Books What to Learn
• Books Summary
• Dictionary Reasoning
• Bible Reasoning
• Import Stackechange articles
• News Article summary folding
• News Article fallacy checking
• News Article deductive folding
• News Article inductive folding
• Reddit deductive reasoning
• Reddit fallacy checking
• Stackexchange deductive sylogisms
• Stackexchange inductive sylogisms
• Stackexchange fallacy checking
• Create a reality definition with the sylogisms
• Comedy Video and Audio Folding

---

Filed under: Uncategorized - @ June 9, 2024 4:04 pm