2.7 KiB
2.7 KiB
跨 region，region 扩缩容、kvs 要迁移、服务什么场景、怎么迁移、怎么定义 cost model
claudian claude code for obsidian
deepwiki
context as memory, LLM as CPU: https://rlancemartin.github.io/2025/06/23/context_engineering/
通过向某些特定网址发送 HTTP 请求（比如google.com/generate_204），根据它们的回复，判断当前是否在线。相比于 ping 8.8.8.8 使用 ICMP protocol 只检查了 DNS，能够检查 full stack 的 DNS, TCP, HTTP
Dijkstra’s 10 commandments for academic research https://www.cs.utexas.edu/~EWD/
- Raise your standards as high as you can live with, avoid wasting your time on routine problems, and always try to work as closely as possible at the boundary of your abilities. Do this because it is the only way of discovering how that boundary should be moved forward.
- We all like our work to be socially relevant and scientifically sound. If we can find a topic satisfying both desires, we are lucky; if the two targets are in conflict with each other, let the requirement of scientific soundness prevail.
- Never tackle a problem of which you can be pretty sure that (now or in the near future) it will be tackled by others who are, in relation to that problem, at least as competent and well-equipped as you are.
- Write as if your work is going to be studied by a thousand people.
- Don't get enamored with the complexities you have learned to live with (be they of your own making or imported). The lurking suspicion that something could be simplified is the world's richest source of rewarding challenges.
- Before embarking on an ambitious project, try to kill it.
- Remember that research with a big R is rarely mission-oriented and plan in terms of decades, not years. Resist all pressure —be it financial or cultural— to do work that is of ephemeral significance at best.
- Don't strive for recognition (in whatever form): recognition should not be your goal, but a symptom that your work has been worthwhile.
- Avoid involvement in projects so vague that their failure could remain invisible: such involvement tends to corrupt one's scientific integrity.
- Striving for perfection is ultimately the only justification for the academic enterprise; if you don't feel comfortable with this goal —e.g. because you think it too presumptuous—, stay out!
视频格式转换： handbrake: https://handbrake.fr/
The craft of research The design of everyday things
https://github.com/xlite-dev/CUDA-Learn-Notes
- transformer 各个模块的具体计算流程
- attention 的各类计算方式，计算、访存的复杂度分析，分别是如何优化的？如何做到硬件友好
- 各种并行策略是如何实现的？如何切分？如何综合？