I have built my first system around 5 years ago, the main purpose is to do data crunching/machine learning/deep learning related tasks. Most of my hardware was decent at the time (given my budget ~$3K).
- CPU: i7-6700K
- GPU: 1080
- Memory: 32gb
- Disk: 512gb ssd
I have been finding with this current setup of mine, a lot of data crunching job is becoming more and more difficult especially for deep learning where GPU is heavily used. What I wish to have improvements on:
- More CPU cores (faster parallel execution for example, and hopefully decent per core speed as well)
- 1080 is pretty old now given all the new gtx 30x series e.g. 3090
- newer architecture than enables faster calculation (where 1080 does not have)
- more gpu memory to hold more data (my current card has 8gb which often was not enough)
- 32gbs is often not enough as well, i would imagine 64gbs would make things easier (of course 128gb is even better)
- its also proven 512gb to be not enough, 1tb would make some of my past tasks easier.
Given this current situation, what would you recommend? Switching out each parts with the latest one? if so, then i guess it comes to the matter of budget, but then i also wonder what do people normally do with their old/current system then?
it feels like a dilemma to me, because the old one is not too incapable that cannot do anything and to be thrown away, but its still not strong enough to do current tasks well, how do you decide? Getting a new one? (wallet takes hit and also need to think of a way to use old system) or replace parts? (probably a lot work and the latest parts most likely wont play well with my old system e.g. old motherboard)