Can an RTX 5080 Handle Heavy LLM Data Consolidation?
I’m trying to figure out whether running a local LLM on an RTX 5080 would be practical for a data-heavy project.
The goal would be to take a large amount of lab-related data and compile it into one clean reference file. This would include things like:
- Lab providers
- Lab test names
- Prices
- Descriptions
- Biomarkers included
- CPT/test codes
- Provider links
- Category/grouping logic
- Duplicate or equivalent test matching
It would not just be basic copy/paste cleanup. Some reasoning would be needed to correctly categorize tests, recognize similar panels across providers, clean inconsistent naming, and structure everything into a usable dataset.
Would a local model on a 5080 be capable of doing this well, assuming the data is chunked properly? Or would the context limits / accuracy issues make this a bad use case?
Also, what model would be the best fit for this kind of task? I’m more interested in accuracy, structured output, and data cleanup than creative writing.
I’m not trying to train a model from scratch. More like using an LLM as a data normalization / research assistant to help build a large reference file.
Specs: 9800X3D, 32gb DDR5, RTX 5080 (spare 3060 12gb I can sidekick if needed)