Dominion Energy, which supplies power to N Virginia (highest concentration of datacenters in US), is now saying no new datacenters b/c the grid cannot support their power needs. IMO it's irresponsible to have these hyperscaler aspirant convos w/o frontloading climate impact.
COMPUTING
-
GPU utilization troubleshooting: why only 30% despite full memory
By
–
yet here's proof that my GPU utilization is still trash. if forward or backward is ~always happening according to the profiler, and memory is decently full now (at least over 50%)… …then how is my GPU being used only on average 30% of the time? what am i missing here?! help
-
PyTorch GPU Utilization Issues: Profiling Forward Backward Pass
By
–
alright y'all. still having GPU utilization issues and against my better judgement i broke out the PyTorch profiler. posting a few screenshots. doesn't seem like I/O is an issue (no gaps) how is GPU util so bad (~30%) if either forward or backward is running almost always?
-
NAIRR Pilot Opens Computing Resources for Safe AI Research
By
–
Are you working to advance safe, secure and trustworthy AI? Apply now to get early access to advanced computing resources as part of the National Artificial Intelligence Research Resource (NAIRR) Pilot: https://
bit.ly/42sZgbD -
GPU Batch Size Memory Utilization Performance Analysis
By
–
actually the batch size is relatively low. memory about half full. do you think that would explain ~30% utilization?
-
Optimizing Transformer Training on Single GPU with HuggingFace
By
–
what are some easy ways for a normal person like me to speed up training a transformer on a single GPU? i'm using huggingface T5 implementation + defaults. i know about FlashAttention and BetterTransformer and torch compile? will these all work together? is there anything else?
-
AI and Generative AI Insights at GTC24 Conference
By
–
Get the insights to prime your organization for emerging #AI, #generativeAI, industrial digitalization, and accelerated computing opportunities. Learn more at our featured sessions at #GTC24. Register now > https://
nvda.ws/42uITet -
Modin: Accelerating Pandas on Large Datasets
By
–
Modin works well on larger datasets, where pandas becomes slow or runs out of memory Check the below example, Pandas (left) and Modin (right) performing the same pandas operations on a 2GB dataset. The only difference between the two examples is the import statement: