惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

N
Netflix TechBlog - Medium
G
Google Developers Blog
H
Hackread – Cybersecurity News, Data Breaches, AI and More
T
The Blog of Author Tim Ferriss
Microsoft Azure Blog
Microsoft Azure Blog
GbyAI
GbyAI
L
LangChain Blog
云风的 BLOG
云风的 BLOG
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
aimingoo的专栏
aimingoo的专栏
P
Proofpoint News Feed
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC
小众软件
小众软件
WordPress大学
WordPress大学
A
About on SuperTechFans
大猫的无限游戏
大猫的无限游戏
C
Check Point Blog
月光博客
月光博客
Stack Overflow Blog
Stack Overflow Blog
美团技术团队
Jina AI
Jina AI
T
Tailwind CSS Blog
Google DeepMind News
Google DeepMind News
D
Docker

Hacker News - Newest: "AI"

AI can't read an investor deck AI as an attorney? Student uses ChatGPT, Gemini to sue UW over alleged racial discrimination Hacking MCP Servers in AI Systems – The Rug Pull: Tool Changes After Approval GitHub - MeepCastana/KubeezCut: Free Web based video editor Can AI judge journalism? A Thiel-backed startup says yes, even if it risks chilling whistleblowers Coming soon: 10 Things That Matter in AI Right Now DARPA built an AI to fact-check enemy weapons claims What explains heterogeneity in AI adoption? When AI Meets Muscle: Context-Aware Electrical Stimulation Promises a New Way to Guide Human Movements - Department of Computer Science AI Changed How We Build. It Did Not Change What Matters. Linux rules on using AI-generated code - Copilot is OK, but humans must take 'full responsibility for the… Meta spins up AI version of Mark Zuckerberg to engage with employees Code Mode: Let Your AI Write Programs, Not Just Call Tools | TanStack Blog GitHub - Delavalom/graft: Go framework for building AI agents. Type-safe tools, multi-provider (OpenAI, Anthropic, Gemini, Bedrock), zero vendor SDKs. India's TCS tops estimates, says new AI models did not dent services demand Gen Z's fading AI hype Strong feeling: we are in a folded AI reality GitHub - machinarii/total-recall-catalog: A reference catalog of latest knowledge retrieval, memory & RAG systems GitHub - mensfeld/code-on-incus: Give each AI agent its own isolated machine with root, Docker, and systemd. Active defense detects and stops threats automatically.. Quantization, LoRA, and the 8% Problem: Benchmarking Local LLMs for Production AI Iran war: We spoke to the man making Lego-style AI videos that experts say are powerful propaganda Powell, Bessent discussed Anthropic's Mythos AI cyber threat with major U.S. banks GitHub - immartian/bellamem: Persistent belief-graph memory for AI agents. Retrieves decisive context by importance — not recency, not RAG, not /compact. recursive-mode: The Repo-Native Operating System for AI Engineering After the attack on Sam Altman's home, will AI CEO's go on the offensive? The biggest advance in AI since the LLM Opus 4.6 vs GPT 5.4 One Prompt Unity World Generation Test “AI polls” are fake polls Client Challenge Can AI be a 'child of God'? Inside Anthropic's meeting with Christian leaders
GitHub - brucehoult/k3_ai: Utility to start a program on ...
hasheddan · 2026-05-21 · via Hacker News - Newest: "AI"

Run command on K3's AI cores

Utility to run a command with arguments on the A100 "AI" cores on SpacemiT K3.

Quick start

$ git clone https://github.com/brucehoult/k3_ai
$ cd k3_ai
$ make
$ ai bash

Usage

ai cmd arg1 arg2 ...

aix path arg1 arg2 ...

Normally prefer to use ai. This is a shell script that converts a bare program name to the absolute path of the executable and then chains to aix. It's about 1ms slower than calling aix directly.

aix is a tiny assembly language program that needs the absolute or relative path of the program to run. It is written to be pure asm using only syscalls, no libraries, no dynamic linker, so as to be really sure that nothing has used any RVV instructions in this process, and therefore no decisions have been made based on the X100 core's shorter VLEN.

Examples

# just run a single program on the A100 cores
ai as hello.s -o hello.o

# same thing but maybe 1ms faster
aix /usr/bin/as hello.s -o hello.o

# run a whole build. All processes started by `make` will run on the A100 cores.
ai make -j8 test

# start a shell on the A100 cores. All programs run from it will be run only on the A100 cores
ai bash

Why you need it

The K3 has eight high performance Out-of-Order X100 cores with RISC-V vector length of 256 bits. These cores are what all normal Linux programs run on.

There are also eight lower performance in-order A100 "AI" cores similar to the X60 cores in the SpacemiT K1/M1 chips, but with long 1024 bit vector registers and ALU.

The "AI" cores can actually run all normal Linux programs. On standard scalar code they have approximately 40% the performance of the main X100 cores, but on certain kinds of vectorised code (for example Llama) they have considerably higher performance.

Even ignoring vectorised code, the A100 cores by themselves have more computing power than any previous RISC-V computer under $1000, including those using the Allwinner D1, StarFive JH7110, THead TH1520, SpacemiT K1/M1, or even Eswin EIC7700x.

As such, it is well worth using the A100 cores to run normal programs and adding 40% to the overall power of a K3 computer.

Because of the different vector length, it is dangerous to move a program between X100 and A100 cores -- in either direction -- once it has been started.

SpacemiT provides a mechanism to move an existing running program from the X100 cores to the A100 cores by writing the PID of the process to /proc/set_ai_thread but that is dangerous depending on what that process has already done.

This utility creates a new process and then moves it to the A100 cores before it has done anything dangerous, and then does EXEC() on the program you really wanted to run.