Summary of Improve Machine Learning Carbon Footprint Using Nvidia Gpu and Mixed Precision Training For Classification Models — Part I, by Andrew Antonopoulos
Improve Machine Learning carbon footprint using Nvidia GPU and Mixed Precision training for classification models – Part I
by Andrew Antonopoulos
First submitted to arxiv on: 12 Sep 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The medium-difficulty summary is as follows: This paper compares the power consumption of training a classification machine learning model using default floating point (32-bit) versus Nvidia mixed precision (16-bit and 32-bit). A custom PC was built to perform experiments, varying ML hyperparameters such as batch size, neurons, and epochs. Power consumption data was collected from GPU, CPU, RAM, and manual wattmeter readings. Benchmarking tests with default hyperparameter values were used as a reference. The results showed a positive outcome when using mixed precision combined with specific hyperparameters, reducing power consumption by 7-11 Watts compared to the benchmarking test. Inferential statistics, including ANOVA and T-test, were used to compare means. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The low-difficulty summary is as follows: This paper compares two ways of training a machine learning model on a computer to see which one uses less power. They built a special PC and tried different settings for the model. They also measured how much power was being used by the computer’s parts, like the graphics card and processor. The results showed that using the mixed precision method with specific settings can reduce power consumption by 7-11 Watts compared to just using the default setting. This could help reduce carbon emissions from computers. |
Keywords
» Artificial intelligence » Classification » Hyperparameter » Machine learning » Precision
