An Open Source GPU Benchmarking Project: BenchDaddi

In this video, I'm excited to introduce BenchDaddi, an innovative open-source GPU benchmarking suite that I've developed. BenchDaddi empowers users to analyze GPU performance across various AI tasks with precision and ease. Throughout this tutorial, I'll walk you through the suite's repository structure, setup, and installation process, showcasing how to leverage its benchmarking scripts tailored for Transformers (BERT), RNNs (LSTM), and CNNs (ResNet50).
Throughout the tutorial, you'll learn how to assess your GPU's throughput, execution time, data transfer time, and memory usage for both training and inference. Whether you're interested in testing thermals, evaluating server noise, or simply curious about your GPU's capabilities, this video provides comprehensive coverage.
Moreover, this project represents the start of an open-source initiative aimed at establishing a reliable method for evaluating the performance of different GPUs. Collaboration is encouraged, so feel free to join us as we work towards refining GPU benchmarking practices together.
Github Repo: github.com/thedatadaddi/Bench...
HOW TO GET IN CONTACT WITH ME
🐦 X (Formerly Twitter): @TheDataDaddi
📧 Email: [e-mail]
💬 Discord: / discord
Feel free to connect with me on X (Formerly Twitter) or shoot me an email for any inquiries, questions, collaborations, or just to say hello! 👋
HOW TO SUPPORT MY CHANNEL
If you found this content useful, please consider buying me a coffee at the link below. This goes a long way in helping me through grad school and allows me to continue making the best content possible.
Buy Me a Coffee
www.buymeacoffee.com/TheDataD...
As a cryptocurrency enthusiast, I warmly welcome donations in crypto. If you're inclined to support my work this way, please feel free to use the following addresses:
Bitcoin (BTC) Address: bc1q3hh904l4uttmge6p58kjhrw4v9clnc6ec0jns7
Ethereum (ETH) Address: 0x733471ED0A46a317A10bf5ea71b399151A4bd6BE
Should you prefer to donate in a cryptocurrency other than Bitcoin or Ethereum, please don't hesitate to reach out, and I'll provide you with the appropriate wallet address.
Thanks for your support!

[@rohithkyla7595]

keen about the comparison video!

[@TheDataDaddi]

This will be coming very soon! Stay tuned.

[@agriculture7188]

do you have a recommendation for what gpu(s) I should purchase for my R720? I mostly plan on running LLMs, Stable Diffusion, and image detection models. I don’t have a super high budget and was considering a dual P100 setup but wanted the opinion of someone a little more educated in the ML field.

[@TheDataDaddi]

Hi there! Thanks so much for the question. So for all of those things on a low budget. I would probably recommend the p100. You could also go with the p40 for more VRAM. The p100 will handle quantization more efficiently and have high significantly higher throughput which will be important when working with LLMs. The p40 has higher VRAM to start with so you can load larger models, but the fp16 performance is really bad so its through put will be a lot worse (theoretically). As budget options though, I think these could still put some of the smaller open source LLMs within reach for you to start experimenting with. Hope this helps!

[@Meoraclee]

Um Hi daddy, Im having a trouble with building a pc to train ai (play game sometimes). With 2500$ budget should I aim for 2x3090 to run 48 GB VRAM or 2x4060ti 24GB VRAM ? Is there any better option in my case ?

[@TheDataDaddi]

Hi there. Thanks so much for the question! I think it depends a lot on your use case, but I would say if you plan on using it primarily for AI workloads the 3090s would be a better choice because of the higher VRAM and ability to support NVLINK. However, if you want to focus more on gaming and some AI work loads I would choose the 4060s and put more money towards other components like a better CPU. You could also just go with a single RTX 4090. This would give you great performance for AI workloads and gaming with budget enough for other high quality components.