jdecourval

5 ms·

1.
by jdecourval 2y ago
Location: QC, Canada Remote: Remote or Hybrid Willing to relocate: Anywhere in QC, Canada Technologies: C++, C, Python, Kotlin, PostgreSQL, SQLite, CMake, Linux, STM32, Docker, SIP, high-availability distributed systems. Résumé
2.
by jdecourval 2y ago
RSS should be available already from my website: https://jdecourval.com/index.xml Not sure how often I will be posting, though!
3.
by jdecourval 2y ago
With the computer idle, just a terminal open, screen on, but minimum brightness, I measure 4W power usage, so about 18h of autonomy. Real-world autonomy is probably closer to 8-12h which I think is really good.
4.
by jdecourval 2y ago
Not perfect, but `systemd-cat` is indeed a very good idea. Thanks.
5.
by jdecourval 2y ago
Fair point! Anything to suggest? I only confirmed that the GPU is indeed clocking correctly, and using the appropriate power profiles. The games I tried all ran flawlessly, but I had less success with ROCm or Vulkan acceleration in Lllama.c
6.
by jdecourval 2y ago
The truth is, I realized this too late, and now I would also need to, maybe not reinstall, but at least reformat. So I have no idea what sort of impact this has, but clearly, this is not huge. What supposedly really kills performance is if
7.
by jdecourval 2y ago
Exactly. This guide makes the computer my own. I'm always astonished when I see ArchLinux users openly praising the distro. It's great... but for people like me. I especially wouldn't recommend using it on a laptop, since ski
8.
by jdecourval 2y ago
Yes, something similar with snapshots happened to me at a previous job. That led to the filesystem getting 100% full, and with btrfs, you can not even delete files anymore if you reach that point. This specific bug has been fixed since then
9.
by jdecourval 2y ago
Thank you! :)
10.
Optimizing Linux for an AMD Xiaomi/Framework Laptop(jdecourval.com)
74 points by jdecourval 2y ago|34 comments
11.
E2E LLM finetuning on a single 24GB GPU(github.com)
1 points by jdecourval 2y ago|1 comments
12.
by jdecourval 2y ago
I have experience working with Large Language Models (LLMs) during my tenure at my previous company. However, one aspect I didn't get the opportunity to explore there was fine-tuning my own LLM, so I decided to experiment. While there