16gb on the latest pro phones, but the system doesn’t let any app use more than 5gb. haven’t tested this myself yet though. maybe I’ll do that tonight
COMPUTING
-
Wall Street’s Best Ideas: Nvidia Gains While Google, Astera Face Pressure
By
–
Wall Street’s Best Ideas: Astera, Google out, Kopin, Nvidia, in The new year starts with a raft of new recommendations, including dumping pricey Astera Labs, avoiding Google and Uber, snapping up Kopin, and hanging onto Nvidia. https://
thetechnologyletter.com/the-posts/wall
-streets-best-ideas-masayoshi-sons-vanity-strikes-again-mth2x
… $NVDA $GOOGL $UBER $KOPN -
iOS Unified Memory Pools Benefits for AI
By
–
i’m sure there are some good ones, but iOS benefits from unified memory pools
-
Battery consumption during AI inference remains high but manageable
By
–
it’s still high, but this is only at the time of inference, so if you have a question or two you won’t really see your battery drop in realtime or something like that
-
Essential GPU Glossary for AI and Machine Learning
By
–
GPU Glossary https://
bit.ly/4gheq9X
#AI #MachineLearning #DeepLearning #LLMs #DataScience -
SambaNova Cloud Speed Comparison with AWS on Llama 3.1 405B
By
–
Legend has it AWS is still going 🫢
— SambaNova (@SambaNovaAI) 2 janvier 2025
See how SambaNova Cloud's speed compares with your own eyes 👀
Experience faster than fast on Llama 3.1 405B. #AI #GenAI #FastAI
Try it yourself ⤵️Legend has it AWS is still going See how SambaNova Cloud's speed compares with your own eyes Experience faster than fast on Llama 3.1 405B. #AI #GenAI #FastAI Try it yourself
-
Data Infrastructure Redefined: Sovereign Clouds and PaaS Evolution
By
–
Why #2025 will redefine #data infrastructure: 11 expert insights on sovereign #clouds, exploding data, PaaS and more https://
venturebeat.com/data-infrastru
cture/why-2025-will-redefine-data-infrastructure-11-expert-insights-on-sovereign-clouds-exploding-data-paas-and-more/
… via @VentureBeat -
WordPress categories for AI topics
By
–
The numbers are from ARC Prize, in the quoted post (two quotes down)
-
Inference Performance Not Bad For Extended Usage
By
–
it’s not bad, you won’t be constantly inferencing for hours