📰 Reworked System Call Entry Handling Slated For Linux 7.3
Stemming from looking at a proposed Linux kernel patch to alter the Linux kernel's system call number handling, veteran Linux kernel developer Thomas Gleixner went down a rabbit hole of the kernel's system call entry handling to make a number of clean-ups and improvements to the code. That rework to the system call entry handling is now expected to land for the Linux 7.3 kernel cycle...
🔗 Source:
#kernel #linux
👉@sysadminoff
https://www.phoronix.com/news/Rework-Sysclal-Entry-Linux-7.3
Stemming from looking at a proposed Linux kernel patch to alter the Linux kernel's system call number handling, veteran Linux kernel developer Thomas Gleixner went down a rabbit hole of the kernel's system call entry handling to make a number of clean-ups and improvements to the code. That rework to the system call entry handling is now expected to land for the Linux 7.3 kernel cycle...
🔗 Source:
#kernel #linux
👉@sysadminoff
https://www.phoronix.com/news/Rework-Sysclal-Entry-Linux-7.3
Phoronix
Reworked System Call Entry Handling Slated For Linux 7.3
Stemming from looking at a proposed Linux kernel patch to alter the Linux kernel's system call number handling, veteran Linux kernel developer Thomas Gleixner went down a rabbit hole of the kernel's system call entry handling to make a number of clean-ups…
ESP32 Keeps Tabs On Your Local Airspace | Hackaday
👉@sysadminoff
https://hackaday.com/2026/07/08/esp32-keeps-tabs-on-your-local-airspace/
👉@sysadminoff
https://hackaday.com/2026/07/08/esp32-keeps-tabs-on-your-local-airspace/
Hackaday
ESP32 Keeps Tabs On Your Local Airspace
We know, we know. Despite being called ESP32-Plane-Radar, this project from [Mateusz Juszczyk] isn’t actually using radar. But thanks to the round LCD this desktop gadget does a fantastic job…
Госдума России приняла закон «О поддержке развития технологий искусственного интеллекта в Российской Федерации»
Сразу во 2-м и 3-м чтении 8.07.2026 был принят очень важный для целой отрасли закон. Закон носит рамочный характер и предполагает, что правительство выпустит массу подзаконных актов, регулирующих отдельные технические и правовые вопросы. Закон, в случае одобрения Советом Федерации и Президентом РФ, вступит в действие в основном уже с 1 сентября 2026 г., хотя полностью все положения заработают с 1 марта 2027 г. В ситуации уже состоявшегося внедрения моделей, там где потребуется применение только национальных или суверенных моделей, их можно будет продолжать использовать до 1 сентября 2032 г.
( читать дальше... )
llm, госрегулирование, искусственный интеллект, россия
👉@sysadminoff
https://www.linux.org.ru/news/ai/18337915
Сразу во 2-м и 3-м чтении 8.07.2026 был принят очень важный для целой отрасли закон. Закон носит рамочный характер и предполагает, что правительство выпустит массу подзаконных актов, регулирующих отдельные технические и правовые вопросы. Закон, в случае одобрения Советом Федерации и Президентом РФ, вступит в действие в основном уже с 1 сентября 2026 г., хотя полностью все положения заработают с 1 марта 2027 г. В ситуации уже состоявшегося внедрения моделей, там где потребуется применение только национальных или суверенных моделей, их можно будет продолжать использовать до 1 сентября 2032 г.
( читать дальше... )
llm, госрегулирование, искусственный интеллект, россия
👉@sysadminoff
https://www.linux.org.ru/news/ai/18337915
😁1
📰 Heavy Metal Death Can is a retro survival horror set on a Cold War–era submarine - now on Linux
Heavy Metal Death Can released back in May and just recently the developer updated the game to include Native Linux and macOS support.Read the full article on GamingOnLinux.
🔗 Source:
#linux
👉@sysadminoff
https://www.gamingonlinux.com/2026/07/heavy-metal-death-can-is-a-retro-survival-horror-set-on-a-cold-war-era-submarine-now-on-linux/
Heavy Metal Death Can released back in May and just recently the developer updated the game to include Native Linux and macOS support.Read the full article on GamingOnLinux.
🔗 Source:
#linux
👉@sysadminoff
https://www.gamingonlinux.com/2026/07/heavy-metal-death-can-is-a-retro-survival-horror-set-on-a-cold-war-era-submarine-now-on-linux/
GamingOnLinux
Heavy Metal Death Can is a retro survival horror set on a Cold War–era submarine - now on Linux
Heavy Metal Death Can released back in May and just recently the developer updated the game to include Native Linux and macOS support.
China accelerates AI deployment as Western skepticism slows adoption
China has rapidly integrated artificial intelligence across its economy, reaching a penetration rate of nearly 43% in daily life and industrial sectors. The nation now holds approximately 60% of the world's AI patents, supported by a core industry valuation exceeding $174 billion. This aggressive rollout is driven by a strategic commitment to digital transformation during the current five-year economic plan.
Source
👉@sysadminoff
https://4sysops.com/archives/china-accelerates-ai-deployment-as-western-skepticism-slows-adoption/
China has rapidly integrated artificial intelligence across its economy, reaching a penetration rate of nearly 43% in daily life and industrial sectors. The nation now holds approximately 60% of the world's AI patents, supported by a core industry valuation exceeding $174 billion. This aggressive rollout is driven by a strategic commitment to digital transformation during the current five-year economic plan.
Source
👉@sysadminoff
https://4sysops.com/archives/china-accelerates-ai-deployment-as-western-skepticism-slows-adoption/
Orange Pi 6 SBC delivers 45 TOPS AI with dual 2.5GbE
👉@sysadminoff
https://www.electronics-lab.com/orange-pi-6-sbc-packs-cix-p1-soc-45-tops-ai-dual-2-5gbe-and-lpddr5/
👉@sysadminoff
https://www.electronics-lab.com/orange-pi-6-sbc-packs-cix-p1-soc-45-tops-ai-dual-2-5gbe-and-lpddr5/
Electronics-Lab
Orange Pi 6 SBC delivers 45 TOPS AI with dual 2.5GbE
Orange Pi 6 SBC features a CIX P1 12-core SoC, 45 TOPS AI, dual 2.5GbE, LPDDR5 memory, and dual PCIe 4.0 NVMe slots.
German AI consortium launches Soofi S 30B-A3B sovereign open-source model
A German research consortium has released Soofi S 30B-A3B, an open-source large language model designed for high performance in both German and English. The model utilizes a Mixture-of-Experts (MoE) architecture, which contains 31.6 billion total parameters but only activates 3.2 billion per token to reduce computational costs. It was trained on 27 trillion tokens, with the German language share intentionally increased to 15.3% to outperform existing European sovereign baselines.
Source
👉@sysadminoff
https://4sysops.com/archives/german-ai-consortium-launches-soofi-s-30b-a3b-sovereign-open-source-model/
A German research consortium has released Soofi S 30B-A3B, an open-source large language model designed for high performance in both German and English. The model utilizes a Mixture-of-Experts (MoE) architecture, which contains 31.6 billion total parameters but only activates 3.2 billion per token to reduce computational costs. It was trained on 27 trillion tokens, with the German language share intentionally increased to 15.3% to outperform existing European sovereign baselines.
Source
👉@sysadminoff
https://4sysops.com/archives/german-ai-consortium-launches-soofi-s-30b-a3b-sovereign-open-source-model/
AI orchestration systems prioritize cost and efficiency over model size
The artificial intelligence industry is transitioning from a race for larger models toward the development of integrated orchestration systems. These platforms function as a harness, automatically routing tasks to the most appropriate model based on complexity, cost, and required data access. This shift allows organizations to utilize inexpensive open-weight models for routine operations while reserving premium proprietary models for high-stakes reasoning or coding tasks.
Source
👉@sysadminoff
https://4sysops.com/archives/ai-orchestration-systems-prioritize-cost-and-efficiency-over-model-size/
The artificial intelligence industry is transitioning from a race for larger models toward the development of integrated orchestration systems. These platforms function as a harness, automatically routing tasks to the most appropriate model based on complexity, cost, and required data access. This shift allows organizations to utilize inexpensive open-weight models for routine operations while reserving premium proprietary models for high-stakes reasoning or coding tasks.
Source
👉@sysadminoff
https://4sysops.com/archives/ai-orchestration-systems-prioritize-cost-and-efficiency-over-model-size/
📰 Slack Who? This Self-Hosted Chat App Just Went Open Source
Built by one developer over the past year, Chatto is a team chat solution that can be self-hosted.
🔗 Source:
#opensource
👉@sysadminoff
https://feed.itsfoss.com/link/24361/17379567/chatto-goes-open-source
Built by one developer over the past year, Chatto is a team chat solution that can be self-hosted.
🔗 Source:
#opensource
👉@sysadminoff
https://feed.itsfoss.com/link/24361/17379567/chatto-goes-open-source
It's FOSS
Slack Who? This Self-Hosted Chat App Just Went Open Source
Built by one developer over the past year, Chatto is a team chat solution that can be self-hosted.
Ubuntu 26.04 fixes Papers bug that sent PDF links to wrong page
Ubuntu 26.04 LTS is prepping a bug fix update to Papers, the document viewer that replaced Evince in 25.04, resolving several annoyances – including internal PDF links that jumped to the wrong page. The internal link snafu only occurred in some PDFs, not all, and typically took you to a page one off the actual target. Annoying. The fix stops two separate bits of code that both tried to set the zoom level when links jumped to fit-width or fit-page view. A knock-on quirk also saw the alt + p shortcut (which will jump back to the previous page) wouldn’t […]
You're reading Ubuntu 26.04 fixes Papers bug that sent PDF links to wrong page, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.
👉@sysadminoff
https://www.omgubuntu.co.uk/2026/07/ubuntu-26-04-papers-pdf-viewer-bug-fixes
Ubuntu 26.04 LTS is prepping a bug fix update to Papers, the document viewer that replaced Evince in 25.04, resolving several annoyances – including internal PDF links that jumped to the wrong page. The internal link snafu only occurred in some PDFs, not all, and typically took you to a page one off the actual target. Annoying. The fix stops two separate bits of code that both tried to set the zoom level when links jumped to fit-width or fit-page view. A knock-on quirk also saw the alt + p shortcut (which will jump back to the previous page) wouldn’t […]
You're reading Ubuntu 26.04 fixes Papers bug that sent PDF links to wrong page, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.
👉@sysadminoff
https://www.omgubuntu.co.uk/2026/07/ubuntu-26-04-papers-pdf-viewer-bug-fixes
📰 My Linux laptop runs better now that I treat it like a server
I'm my own sysadmin and I only have myself to blame.
🔗 Source:
#linux
👉@sysadminoff
https://www.xda-developers.com/my-linux-pc-runs-better-now-that-i-treat-it-like-a-server/
I'm my own sysadmin and I only have myself to blame.
🔗 Source:
#linux
👉@sysadminoff
https://www.xda-developers.com/my-linux-pc-runs-better-now-that-i-treat-it-like-a-server/
XDA
My Linux laptop runs better now that I treat it like a server
I'm my own sysadmin and I only have myself to blame
Инициатива по упрощению тестирования экспериментальных версий программ в GNOME OS
Разработчики GNOME OS, дистрибутива для разработчиков и тестировщиков GNOME, представили инициативу по созданию инструментария для упрощения тестирования экспериментальных версий программ в дистрибутивах, распространяемых в форме атомарно обновляемых монолитных системных образов. Развиваемый инструментарий GNOME OS Developer Tool Suite позволит собирать, распространять и тестировать расширения к дистрибутивам, поставляемым в форме системных образов. Инструментарий даст возможность разработчикам и тестировщикам изменять части системы и тестировать вносимые изменения, не нарушая при этом монолитность системного образа, не используя изолированные от системы контейнеры и не отключая механизмы обеспечения безопасности.
👉@sysadminoff
https://www.opennet.ru/opennews/art.shtml?num=65897
Разработчики GNOME OS, дистрибутива для разработчиков и тестировщиков GNOME, представили инициативу по созданию инструментария для упрощения тестирования экспериментальных версий программ в дистрибутивах, распространяемых в форме атомарно обновляемых монолитных системных образов. Развиваемый инструментарий GNOME OS Developer Tool Suite позволит собирать, распространять и тестировать расширения к дистрибутивам, поставляемым в форме системных образов. Инструментарий даст возможность разработчикам и тестировщикам изменять части системы и тестировать вносимые изменения, не нарушая при этом монолитность системного образа, не используя изолированные от системы контейнеры и не отключая механизмы обеспечения безопасности.
👉@sysadminoff
https://www.opennet.ru/opennews/art.shtml?num=65897
📰 Ubuntu 26.04 fixes Papers bug that sent PDF links to wrong page
Ubuntu 26.04 LTS is prepping a bug fix update to Papers, the document viewer that replaced Evince in 25.04, resolving several annoyances – including internal PDF links that jumped to the wrong page. The internal link snafu only occurred in some PDFs, not all, and typically took you to a page one off the actual target. Annoying.
🔗 Source:
#ubuntu
👉@sysadminoff
https://www.omgubuntu.co.uk/2026/07/ubuntu-26-04-papers-pdf-viewer-bug-fixes
Ubuntu 26.04 LTS is prepping a bug fix update to Papers, the document viewer that replaced Evince in 25.04, resolving several annoyances – including internal PDF links that jumped to the wrong page. The internal link snafu only occurred in some PDFs, not all, and typically took you to a page one off the actual target. Annoying.
🔗 Source:
#ubuntu
👉@sysadminoff
https://www.omgubuntu.co.uk/2026/07/ubuntu-26-04-papers-pdf-viewer-bug-fixes
OMG! Ubuntu
Ubuntu 26.04 fixes Papers bug that sent PDF links to wrong page
Ubuntu 26.04 LTS is prepping a bug fix update to Papers, the document viewer that replaced Evince in 25.04, resolving several annoyances – including
RDP Multipath with redundant TCP for Azure Virtual Desktop and Windows 365
RDP Multipath is a feature for Azure Virtual Desktop and Windows 365 Cloud PCs that keeps multiple network transport paths active simultaneously and automatically switches them when one degrades. Microsoft has made RDP Multipath with redundant TCP support generally available for both services. The feature activates automatically when the prerequisites are met, requiring no server-side configuration. This article explains how the feature works, what to prepare, and what to verify.
Source
👉@sysadminoff
https://4sysops.com/archives/rdp-multipath-with-redundant-tcp-for-azure-virtual-desktop-and-windows-365/
RDP Multipath is a feature for Azure Virtual Desktop and Windows 365 Cloud PCs that keeps multiple network transport paths active simultaneously and automatically switches them when one degrades. Microsoft has made RDP Multipath with redundant TCP support generally available for both services. The feature activates automatically when the prerequisites are met, requiring no server-side configuration. This article explains how the feature works, what to prepare, and what to verify.
Source
👉@sysadminoff
https://4sysops.com/archives/rdp-multipath-with-redundant-tcp-for-azure-virtual-desktop-and-windows-365/
OpenAI recommends outcome-first prompting for GPT-5.6 and Codex
OpenAI has released updated prompting guidance that prioritizes desired outcomes over rigid step-by-step instructions. This approach allows reasoning models like GPT-5.6 to independently determine the most efficient search, tool, or reasoning path to reach a goal. The framework is built around four optional components: goal, context, output format, and boundaries.
Source
👉@sysadminoff
https://4sysops.com/archives/openai-recommends-outcome-first-prompting-for-gpt-5-6-and-codex/
OpenAI has released updated prompting guidance that prioritizes desired outcomes over rigid step-by-step instructions. This approach allows reasoning models like GPT-5.6 to independently determine the most efficient search, tool, or reasoning path to reach a goal. The framework is built around four optional components: goal, context, output format, and boundaries.
Source
👉@sysadminoff
https://4sysops.com/archives/openai-recommends-outcome-first-prompting-for-gpt-5-6-and-codex/
Xi Jinping to headline 2026 World AI Conference in Shanghai
The 2026 World AI Conference and High-Level Meeting on Global AI Governance is scheduled to take place in Shanghai from July 17 to July 20. Chinese President Xi Jinping will attend the opening ceremony to deliver a keynote speech regarding the future of artificial intelligence. This appearance marks the first time the Chinese leader will attend the annual flagship event in person since its inception in 2018.
Source
👉@sysadminoff
https://4sysops.com/archives/xi-jinping-to-headline-2026-world-ai-conference-in-shanghai/
The 2026 World AI Conference and High-Level Meeting on Global AI Governance is scheduled to take place in Shanghai from July 17 to July 20. Chinese President Xi Jinping will attend the opening ceremony to deliver a keynote speech regarding the future of artificial intelligence. This appearance marks the first time the Chinese leader will attend the annual flagship event in person since its inception in 2018.
Source
👉@sysadminoff
https://4sysops.com/archives/xi-jinping-to-headline-2026-world-ai-conference-in-shanghai/
Chinese researchers achieve 100-fold AI speed boost using optical interconnects
Researchers at Peking University have developed an all-optical interconnect system that links standard electronic chips to significantly accelerate artificial intelligence processing. This breakthrough utilizes specialized hardware to replace traditional electrical connections between chips with high-speed optical "joints." By integrating these optical links with custom algorithms, the system achieves AI inference speeds up to 100 times faster than conventional methods.
Source
👉@sysadminoff
https://4sysops.com/archives/chinese-researchers-achieve-100-fold-ai-speed-boost-using-optical-interconnects/
Researchers at Peking University have developed an all-optical interconnect system that links standard electronic chips to significantly accelerate artificial intelligence processing. This breakthrough utilizes specialized hardware to replace traditional electrical connections between chips with high-speed optical "joints." By integrating these optical links with custom algorithms, the system achieves AI inference speeds up to 100 times faster than conventional methods.
Source
👉@sysadminoff
https://4sysops.com/archives/chinese-researchers-achieve-100-fold-ai-speed-boost-using-optical-interconnects/
obsidian-skills: Teach Claude Code How to Work with Obsidian
The post obsidian-skills: Teach Claude Code How to Work with Obsidian first appeared on Tecmint: Linux Howtos, Tutorials & Guides .If you’ve ever watched an AI coding agent mess up your Obsidian vault, this can help. I keep my second
The post obsidian-skills: Teach Claude Code How to Work with Obsidian first appeared on Tecmint: Linux Howtos, Tutorials & Guides.
👉@sysadminoff
https://www.tecmint.com/obsidian-skills-claude-code-linux/
The post obsidian-skills: Teach Claude Code How to Work with Obsidian first appeared on Tecmint: Linux Howtos, Tutorials & Guides .If you’ve ever watched an AI coding agent mess up your Obsidian vault, this can help. I keep my second
The post obsidian-skills: Teach Claude Code How to Work with Obsidian first appeared on Tecmint: Linux Howtos, Tutorials & Guides.
👉@sysadminoff
https://www.tecmint.com/obsidian-skills-claude-code-linux/
obsidian-skills: Teach Claude Code How to Work with Obsidian
obsidian-skills: Help Claude Code Read and Write Obsidian Notes
Learn how to install obsidian-skills on Linux and make Claude Code understand Obsidian's wikilinks, Bases, Canvas files, and other vault-specific features.
Анализ безопасности 281 VPN-приложения для Android
Исследователи из Мичиганского университета разработали инструментарий MVPNalyzer для выявления утечек информации и анализа работы VPN, и проверили с его помощью 281 VPN-приложение для платформы Android. Для изучения были отобраны наиболее популярные VPN-приложений, распространяемые через каталог Google Play. VPN-приложения, в которых были выявлены те или иные проблемы с безопасностью и приватностью, в сумме насчитывают 2.4 миллиарда установок. Основные выводы.
👉@sysadminoff
https://www.opennet.ru/opennews/art.shtml?num=65892
Исследователи из Мичиганского университета разработали инструментарий MVPNalyzer для выявления утечек информации и анализа работы VPN, и проверили с его помощью 281 VPN-приложение для платформы Android. Для изучения были отобраны наиболее популярные VPN-приложений, распространяемые через каталог Google Play. VPN-приложения, в которых были выявлены те или иные проблемы с безопасностью и приватностью, в сумме насчитывают 2.4 миллиарда установок. Основные выводы.
👉@sysadminoff
https://www.opennet.ru/opennews/art.shtml?num=65892
📰 A Raspberry Pi 5 Handheld That Replaces Plastic with Brass
🔗 Source:
👉@sysadminoff
https://www.hackster.io/news/a-raspberry-pi-5-handheld-that-replaces-plastic-with-brass-eb5e29d5b751
🔗 Source:
👉@sysadminoff
https://www.hackster.io/news/a-raspberry-pi-5-handheld-that-replaces-plastic-with-brass-eb5e29d5b751
Hackster.io
A Raspberry Pi 5 Handheld That Replaces Plastic with Brass
Why settle for a boring black rectangle when you can build this gorgeous custom Raspberry Pi 5 handheld?
OpenAI GPT-5.6 family arrives on Amazon Bedrock with tiered reasoning
Amazon Bedrock has added general availability for the OpenAI GPT-5.6 model family, consisting of the Sol, Terra, and Luna tiers. Sol serves as the flagship model for complex reasoning and autonomous coding, while Terra offers balanced performance for everyday production tasks. Luna provides a high-speed, low-cost option optimized for low-latency applications like real-time classification and summarization.
Source
👉@sysadminoff
https://4sysops.com/archives/openai-gpt-5-6-family-arrives-on-amazon-bedrock-with-tiered-reasoning/
Amazon Bedrock has added general availability for the OpenAI GPT-5.6 model family, consisting of the Sol, Terra, and Luna tiers. Sol serves as the flagship model for complex reasoning and autonomous coding, while Terra offers balanced performance for everyday production tasks. Luna provides a high-speed, low-cost option optimized for low-latency applications like real-time classification and summarization.
Source
👉@sysadminoff
https://4sysops.com/archives/openai-gpt-5-6-family-arrives-on-amazon-bedrock-with-tiered-reasoning/