For Stable Diffusion 3, we measured the training throughput for the 2B Multimodal Diffusion Transformer (MMDiT) architecture model. Gaudi 2 trained images 1.5x faster than the H100-80GB, and 3x faster than A100-80GB GPU’s when scaled up to 32 nodes. (2/3)
AI HARDWARE
-
Intel Gaudi 2 vs Nvidia A100 H100 Training Speed Comparison
By
–
In this installment of "Behind the Compute", a series dedicated to offering insights for others to harness the power of generative AI, we compared the training speed of @Intel Gaudi 2 accelerators versus @Nvidia
's A100 and H100 for two of our models. (1/3) -
Accelerate Enterprise AI Initiatives with Optimized IT Infrastructure
By
–
Get Ahead with Your Enterprise #AI initiatives! Watch this @PureStorage TechTalk to learn how you can accelerate your AI priorities with top tips for IT infrastructure optimized for AI: http://
purefla.sh/3wwG7JM -
Pure Storage at NVIDIA GTC24: GenAI RAG Demos and Expert Meetings
By
–
Don't miss @PureStorage at #NVIDIA #GTC24. Visit the Pure Booth (#1529). Talk with their experts. See the #GenAI RAG demos. Attend their theater session. Register for 20% OFF registration at http://
purefla.sh/3OYYnSn where you can also book a live meeting with their experts, or if -
HONOR’s Remarkable AI Device Ecosystem Showcased at MWC2024
By
–
I am amazed % by all of this… @Honorglobal Illuminates the Future of #AI in Smart Devices at #MWC24 See highlights from #MWC2024 of HONOR’s remarkable device ecosystem at: https://
bit.ly/3vRpvw7 And read more about each of the HONOR Magic products in the news here: -
Assessing mass production capabilities for Figure’s AI robots
By
–
+ let’s see if Figure can accelerate But unclear if they will have any mass production lines available any time soon
-
Jensen Huang continues minting money in AI hardware boom
By
–
yes huge profits there; should have clarified that (was focused on the GenAI companies). Jensen will continue to mint money.
-
GPU Infrastructure Evolution: From Personal Desktops to University Investment
By
–
Even I as a @UCBerkeley grad student had 8 gpus in my (non uni) desktop in 2012 (electricity bills weren't cheap…). Time to invest @Stanford
-
AI Now Institute Report on Compute and AI Policy
By
–
And this stellar report from @AINowInstitute — @sarahbmyers and Jai Vipra 9/ https://
ainowinstitute.org/publication/po
licy/compute-and-ai
… -
Gaming Future Predictions: Next Decade Technological Evolution
By
–
Can't wait for gaming in 5-10 years