M1 Max
A high-performance Apple-designed chip used in the Mac Studio and MacBook Pro. It features a unified memory architecture and supports local LLM inference.
Timeline (last 90 days)
The replay test was conducted on an M1 Max, with the 27B using the production build prism-b10709 and the 9B requiring prism-b10777; the 27B reproduced 273 of 275 production decisions, and the two edge differences may come from Metal/CUDA numerical differences.
In actual benchmark threads for llama.cpp, the 2021 M1 Max achieved a decode write speed of 61 tokens/s, higher than the 2024 M4 Pro's 51 tokens/s.
Running a 27B model on M1 Max with a 32,000-token prompt requires waiting more than 500 seconds before the first character is output, whereas the M5 Max takes less than one minute (approximately 40 seconds with optimized software).
The notch design on MacBook Pro began with the M1 Pro/M1 Max models released in October 2021.
在 M1 Max 上以 10 个工作线程运行时，reinfors 引擎可达到 8,000+ 步/秒，但该数字并非与 Gymnasium 的 AsyncVectorEnv 进行同等条件对比
All Facts (7)
M1 Pro/Max配备了两个独立的Media Engine，可同时处理多路视频流
85%Unverified苹果从2021年推出搭载M1 Pro/M1 Max芯片的MacBook Pro开始引入刘海屏设计
60%UnverifiedThe replay test was conducted on an M1 Max, with the 27B using the production build prism-b10709 and the 9B requiring prism-b10777; the 27B reproduced 273 of 275 production decisions, and the two edge differences may come from Metal/CUDA numerical differences.
50%UnverifiedRunning a 27B model on M1 Max with a 32,000-token prompt requires waiting more than 500 seconds before the first character is output, whereas the M5 Max takes less than one minute (approximately 40 seconds with optimized software).
50%UnverifiedIn actual benchmark threads for llama.cpp, the 2021 M1 Max achieved a decode write speed of 61 tokens/s, higher than the 2024 M4 Pro's 51 tokens/s.
50%UnverifiedThe notch design on MacBook Pro began with the M1 Pro/M1 Max models released in October 2021.
50%Unverified在 M1 Max 上以 10 个工作线程运行时，reinfors 引擎可达到 8,000+ 步/秒，但该数字并非与 Gymnasium 的 AsyncVectorEnv 进行同等条件对比
50%