Hackernews posts about M2 Ultra

M2 Ultra is a powerful system-on-chip (SoC) designed by Apple that enables impressive performance and efficiency for artificial intelligence (AI) workloads, such as large language models and high-speed video processing.

1. M2 Ultra can run 128 streams of Llama 2 7B in parallel (github.com)
   268 points by behnamoh over 1 year ago | 173 comments
2. Running a 180B parameter LLM on a single Apple M2 Ultra (twitter.com)
   255 points by tbruckner almost 2 years ago | 141 comments
3. DeepSeek R1 671B running on 2 M2 Ultras faster than reading speed (twitter.com)
   96 points by thyrox 5 months ago | 29 comments
4. M3 Max Chip Around as Fast as M2 Ultra in Early Benchmark Results (www.macrumors.com)
   89 points by jitl over 1 year ago | 100 comments
5. Apple to Power AI Features with M2 Ultra Servers (www.macrumors.com)
   38 points by mikhael about 1 year ago | 60 comments
6. Apple claims M2 Ultra "can train ML workloads, like LLMs" (old.reddit.com)
   33 points by behnamoh almost 2 years ago | 7 comments
7. Full F16 precision 34B Code Llama at >20T/s on M2 Ultra (twitter.com)
   17 points by Jimmc414 almost 2 years ago | 2 comments
8. Apple to Power AI Features with M2 Ultra Servers (www.macrumors.com)
   16 points by maattp about 1 year ago | 3 comments
9. Why Intel and AMD don't make chips like the M2 Max and M2 Ultra (www.xda-developers.com)
   10 points by Tuldok almost 2 years ago | 4 comments
10. gg: "M2 Ultra is the absolute best personal LLM inference node you can buy." (github.com)
    8 points by behnamoh over 1 year ago | discuss
11. Apple Reportedly Building M2 Ultra and M4-Powered AI Servers (www.macrumors.com)
    5 points by larrysalibra about 1 year ago | 3 comments
12. Apple to Power AI Features with M2 Ultra Servers (www.macrumors.com)
    5 points by robbomacrae about 1 year ago | discuss
13. M4 Max Chip Up to 25% Faster Than M2 Ultra in First Benchmark Results (www.macrumors.com)
    4 points by retskrad 8 months ago | discuss
14. M3 Ultra Is Slower Than M2 Ultra for Language Models (old.reddit.com)
    3 points by behnamoh 4 months ago | 4 comments
15. M2 Ultra runs codellama 34B F16 at 150 token per second (twitter.com)
    3 points by behnamoh over 1 year ago | 3 comments
16. 100 tokens/s, 4-bit Mistral 7B in MLX on M2 Ultra (faster than llama.cpp) (twitter.com)
    3 points by tosh over 1 year ago | discuss
17. MacBook Pro with M3 Max Chip on Par with M2 Ultra in Performance (9to5mac.com)
    3 points by retskrad over 1 year ago | discuss
18. Apple predicted the rise of local LLMs, hence the M2 Ultra
    2 points by behnamoh over 1 year ago | 3 comments
19. Ask HN: Running LLMs locally–what hardware should I get–M2 Ultra, PC, cloud GPU?
    2 points by behnamoh over 1 year ago | 2 comments
20. Ask HN: Case Study on Training LLMs on Apple M2 Ultra Neural Engine?
    1 points by pritambarhate about 1 year ago | 1 comments
21. Apple developed a chip for the Apple car that is similar to four M2 Ultra (www.bloomberg.com)
    1 points by xrayarx over 1 year ago | 1 comments
22. Llama on Mac M2 Ultra (Literally) (github.com)
    1 points by behnamoh over 1 year ago | discuss
23. Maxed Out Mac Studio's Trade-In Value ≃ Basic MacBook Pro with M1 Pro
    6 points by behnamoh 2 months ago | 2 comments
24. Ask HN: Do you use a second local build server?
    4 points by valty almost 2 years ago | 1 comments
25. Ask HN: Best Setup for LLM training and inference with $30k?
    1 points by behnamoh 4 months ago | 1 comments
26. Show HN: Horizon Chart for System Monitoring in terminal (github.com)
    1 points by zqkba over 1 year ago | discuss
27. The Mac Pro and Studio won't get the M4 nod until mid-2025 (www.theverge.com)
    2 points by doener about 1 year ago | 1 comments
28. M2 Macs cannot power more than 3 USB/Thunderbolt devices (old.reddit.com)
    2 points by behnamoh over 1 year ago | discuss
29. Self Driving Cars: Apple Silicon Equivalent to Four M3 Ultra SoCs (wccftech.com)
    1 points by doener over 1 year ago | discuss
30. Qcom elite uses per-cluster DVFS, with cores in a cluster share the same freq
    2 points by firstlovelife 8 months ago | discuss