CandleFocus

AI Training in GPUs: The Way Forward for Miners

Author: Angelica Morris
Cryptocurrency mining was a lucrative endeavor that offered miners staggering profits until Ethereum switched to the proof-of-stake protocol. However, with the soaring popularity of AI, miners are presented with a new opportunity to make use of their GPUs. But when it comes to using these haspower-loving cards for AI, it is less than ideal. This is because AI training requires GPUs with an abundant vRAM (Video Random Access Memory). Unlike cryptocurrency mining that focuses on the number of computations your GPU can crunch per second, AI training is more vRAM oriented as it necessitates GPUs that are capable of handling and storing vast amounts of data simultaneously.

As expected, the level of vRAM needed for mining crypto and AI differ greatly. For instance, a competent GPU for mining crypto may only have 4GB of vRAM, whereas OpenAI requires Nvidia’s A100 and V100 models with 32 and 80GB of vRAM respectively. Though some miners with lesser vRAM GPUs can still be employed for smaller AI models or tasks such as Stable Diffusion, but utilizing them for training larger AI models will only be futile.

However, some crypto mining companies have already made the change to AI operations. Omega AI, Hive Blockchain and Hut8 Mining are all trying out AI operations. This solution, while not as profitable as your good old altcoin mining, can still serve as a plan C for miners who are unsuccessful in their ASIC operations.

Related News