Giovanni's blog

As Phoronix and other outlets have reported, a third major vendor of X86 is utilizing a new 64-bit only chip architecture called  X86-S , a project Intel abandoned because it got some pushback from the legacy linux community and also didn't have a market apparently it wanted to invest in. While RosaicLabs has been given an X86 license, there is still some unconfirmed details about which company or companies are actually implementing the X86S and APX. https://www.phoronix.com/forums/forum/hardware/processors-memory/1656925-mysterious-x86-cpu-already-has-apx-x86s-where-intel-left-off-for-legacy-free-x86   However, regardless of who, what is known is that the interest in the Atom core is notable for several reasons. First, my long-requested chip design since 2011 to integrate a solar Pentium using near threshold voltage may have a new chip architect who may be receptive to my ideas (I am being extremely optimistic, of course). While I have focused on 32-bit code because of the ...

Building on recent X86 news, I ran this through a Gemini Prompt : https://github.com/hatonthecat/solarcircuits/commit/ebbc2dfca3f293927b0a75a191695b03db609227 https://www.imec-int.com/en/articles/disrupting-dram-roadmap-capacitor-less-igzo-dram-technology https://en.wikipedia.org/wiki/ Intel_5-level_paging https://www.phoronix.com/news/ x86-CPU-With-APX-x86S   https://hwbusters.com/news/ x86s-is-back-from-the-dead- and-the-vendor-reviving-it- still-isnt-intel-or-amd/ https://en.kocpc.com.tw/ archives/25278 " PM64-only means the core boots straight into 64-bit protected mode" https://www.techradar.com/pro/ why-is-intel-sharing-its-atom- cpu-technology-with-a-start- up-the-answer-could-be-in-low- power-x86-design https://www.intel.com/content/www/us/en/products/sku/207904/intel-atom-x6200fe-processor-1-5m-cache-1-00-ghz/specifications.html https://www.intel.com/content/ www/us/en/products/platforms/ details/elkhart-lake.html   https://en.wikipedia.org/wiki/ Tremont_(microarchi...

[I originally drafted this article two weeks ago (August 1st, but then had been delayed and other errands] Most of it has been left intact, with the exception of a brief conclusion added.] I remember YouTube before it was acquired by Google in 2006. Its interface has changed significantly over the years, for better and worse. Higher bittrate, 360 degree videos, and all sorts of analytics. But I also noticed it wastes bandwidth. I'm not referring to preloading a video that isn't watched- yes it used to play the entire video in the past, where now it only caches up to 90-120 seconds to save bandwidth in case the user doesn't watch the entire video. I'm also not referring to AutoPlay, which I also disable on all my devices that I'm logged into. I'm aware there are countless extensions that allow smartbuffering, some which are great on browsers that still support extensions (or aren't routinely filtered by Chrome, which is owned by the same company as Youtube). ...

Red Hat 7.2 (2001) with 8MB RAM Red Hat 7.2 System Information  Two nights ago (technically Saturday night) I spent some time using free tokens from Replit, Github Codespace, and Gemini to understand for myself what goes into kernel and system development. While I have read Linux From Scratch excerpts and countless resources, sometimes it is more interesting (or at a loss of ideas?) to see AI attempt to develop an operating system image. The first few attempts I got a non bootable image, around 5-6MB in size. The later ones determined what was missing and I ended up with a stalled bootup process (further than before) at around 20-29MB (the request was to make it less than 30MB, which it did). Those sizes were requested because I knew they were possible and bootable (such as Tiny Core Linux and SliTaz 's core ISO- their images are also bootable in under 30MB). However, without knowing enough steps, the code assistant can sometimes repeat the same mistake a few times, and it wastes...

PSI Raining on the CPU Scheduler   As I began familiarizing myself with various Yocto builds, I watched a few Yocto tutorials on optimizing compilation times, and came across a 2022 video  "making bitbake Psi-chic" and other tutorials to enable  PSI at GRUB.conf for Yocto builds (2022 Slides ).  from p.7&8 of the 2022 Slides & /layers/bitbake/lib/bb/runqueue.py (it appears the BB_PRESSURE_MAX_{CPU|IO|MEMORY} was not found in local.conf)   However, the feature was added only in kernel versions 4.2 through 5.15, where a proc/pressure folder exists. According to this presentation , a new process needs to be used to enable it for kernel versions after that (e.g. 5.20+). Some distros do not even compile it so you have to check which kernel and distro version you have. ( Edit: 7-12-26 : Some 2023 forums confirm it was removed after kernel 5.15, and it needs to be recompiled with a distro if someone wants to use it. Also, it confirms Arch and Zen kernel use ...

Last fall, I tested my first Yocto (5.0) linux builds on a couple machines from last decade: a 2011 Lenovo ThinkPad Edge 420 with an i3 2350M and 6GB of RAM, and a Ryzen 5 1600 with 16GB of RAM. The latter is quite capable , but I honestly found it more interesting to see the dual core i3 (w/ 4 threads) manage to compile a  core-image-minimal.  I even wrote an installation log about the process (I recorded video with the Ryzen using Ubuntu's built-in CPU-based GNOME Recorder, since it has 12 threads and the GPU (a Radeon 6450 or R5 240) I was using didn't have at least VCE 1.0 ), whereas I only took screenshots on the i3 2350m during load , to limit crashing) With the release of Yocto 6.0.1 LTS Wrynose , I was curious to test out BitBake, on a PC in between the performance of the i3 2350m and the Ryzen 5 1600. My 2015 FX8320E with 8 threads (it's called an 8 core but it's really a quad core with 2 threads each). I built this PC with parts from MicroCenter and Newegg, ...