Day: 4 November 2023

MITRE ATT&CK

From another security maillist, Mitre is mentioned a lot but till this week I didnt really dig a bit about. So copy/paste:

MITRE ATT&CK® is a globally-accessible knowledge base of adversary tactics and techniques based on real-world observations. The ATT&CK knowledge base is used as a foundation for the development of specific threat models and methodologies in the private sector, in government, and in the cybersecurity product and service community.

In similar subject, at some point, I would like to see how vulnerable my VPS is. Still not sure if would be usable or how to use Mittre to do that. At least to get some audit/basics done and improve my “security” knowledge a bit. As usual… time.

Google Spanner

From an email list, I read something about Gmail migration to Spanner. I was a bit surprised because I use gmail and didnt know anything about it. That email sent me to this page. That migration had to be a monster one! More details here. From the first page, I had a bit more info about Falcon. In summary, that is part of a bigger picture about building the “AI-driven” future infrastructure.

FP8-LM

From the AlphaSignal email list, that most of the times go over my lame knowledge, I found this piece of info, quite interesting:

FP8-LM: Training FP8 Large Language Models

Goal: Optimize LLM training with FP8 low-bit data formats.
Issue: High cost of LLM computational resources.
Solution: FP8 automatic mixed-precision framework for LLMs.
Results: Reduced memory by 42%, increased speed by 64%.
Insight: FP8 maintains accuracy, optimizes training efficiency.

Repo. Paper

This is something I want to really understand at one point. FP (Floating-Point) instructions can be from several sizes (8, 16, 32, 64). So the bigger, the better precision. I guess for some scientific tasks that is important. But looks like for AI, with FP8 could be good enough.

Limits Computer Performance

Reading across this blog, I came to this statement:

What limits computer performance today is predictability, and the two big ones are instruction/branch predictability, and data locality.

That is from this interview. I dont kown Jim Keller but it is a long and interesting conversation. I liked it when he says he was the laziest person at Tesla!

And actually I found a tab from his company