Perplexity bets on local hardware for sovereign, private AI workloads

Perplexity CEO Aravind Srinivas says users increasingly want AI agents to run on their own hardware to strengthen privacy, cybersecurity, intellectual-property protection, and digital sovereignty while controlling costs. In an interview with CNBC’s Kate Rooney, he described Perplexity’s fully local DGX Spark deployment and a hybrid model that keeps sensitive data on-premises while still accessing cloud-based frontier models.
Source

👉@sysadminoff

https://4sysops.com/archives/perplexity-bets-on-local-hardware-for-sovereign-private-ai-workloads/

📰 Canonical shuttering some of its legacy chat channels

The Ubuntu Pastebin went in June, IRC gets demoted next.

🔗 Source: https://www.theregister.com/os-platforms/2026/09/04/canonical-shuttering-some-of-its-legacy-chat-channels/5294524

#ubuntu

👉@sysadminoff

https://theregister.com/os-platforms/2026/09/04/canonical-shuttering-some-of-its-legacy-chat-channels/5294524

Ergo Framework 3.3

Для тех, кто еще не знаком с фреймворком: он реализует акторную модель — дизайн-паттерн из мира Erlang/Elixir. Это не виртуальная машина BEAM, а скорее аналог библиотеки OTP для построения распределенных систем.
Этот релиз получился очень большим, поэтому я остановлюсь на ключевых изменениях.
( читать дальше... )



 elixir, erlang, framework, go, от автора

👉@sysadminoff

https://www.linux.org.ru/news/opensource/18374006

Новые версии Wine 11.17, Wine-staging 11.17, DXVK 3.1, D7VK 2.1 и Vkd3d 2.1

Опубликован экспериментальный выпуск открытой реализации Win32 API - Wine 11.17. С момента выпуска 11.16 был закрыто 44 отчёта об ошибках и внесено 387 изменений.

👉@sysadminoff

https://www.opennet.ru/opennews/art.shtml?num=66212

Реформирование процесса утверждения значительных изменений в GNOME

Разработчики GNOME обсуждают внедрение нового процесса принятия решений, основанного на публикации предложений в форме RFC (Request for Comments) и их последующего совместного утверждения. Участникам GNOME Foundation предлагается подключиться к обсуждению и высказать свои замечания до 4 октября. Модель на основе RFC планируют применять только в особых случаях, для разрешения тупиковых ситуаций, утверждения глобальных изменений в проекте, затрагивающих методы работы участников или код разных приложений и подсистем, а также архитектурных изменений и крупных переработок интерфейса.

👉@sysadminoff

https://www.opennet.ru/opennews/art.shtml?num=66213

📰 I Tried This Rust Tool, and It Immediately Made Bash Modern

A Rust-powered Readline replacement that brings intelligent suggestions, fuzzy search, syntax highlighting and modern completion to Bash.

🔗 Source:

#rust

👉@sysadminoff

https://feed.itsfoss.com/link/24361/17440361/flyline

📰 Brave Says Its Browser Loads Pages 20% Faster Than Major Rivals

New Brave benchmarks put the privacy-focused browser ahead of Chrome, Edge, and Firefox across several resource-usage metrics.

🔗 Source: /

#firefox

👉@sysadminoff

https://linuxiac.com/brave-says-its-browser-loads-pages-20-faster-than-major-rivals

📰 GNOME Stakeholders Discussing RFC Process For Solving Major Technical Changes

GNOME has been working to establish a formal request for comments (RFC) process for solving major technical questions and changes moving forward. A drafted RFC is now under discussion by GNOME stakeholders for moving this process forward...

🔗 Source: https://www.phoronix.com/news/GNOME-RFC-Process-Discussion

#gnome

👉@sysadminoff

https://phoronix.com/news/GNOME-RFC-Process-Discussion

Cisco’s Unified Edge brings AI inference to distributed sites

Cisco is turning edge locations into full AI infrastructure with Unified Edge, a modular platform that combines compute, networking, and storage for real-time inference outside the data center. The system supports CPUs and GPUs, up to 120 TB of storage, redundant power and cooling, and 25-gigabit networking, while Cisco Intersight provides centralized management across distributed deployments.
Source

👉@sysadminoff

https://4sysops.com/archives/ciscos-unified-edge-brings-ai-inference-to-distributed-sites/

Microsoft confirms Registry workaround still keeps uBlock Origin working in Edge

Microsoft has confirmed that a Registry-based workaround can still keep uBlock Origin and other Manifest V2 extensions running in Edge, despite the browser’s planned phase-out. The change follows Microsoft’s announcement that Edge will eventually disable these extensions, giving organizations that still depend on MV2 extensions temporary breathing room.
Source

👉@sysadminoff

https://4sysops.com/archives/microsoft-confirms-registry-workaround-still-keeps-ublock-origin-working-in-edge/

📰 Linux Patches Posted For Initial Enablement Of The Apple A18 Pro + MacBook Neo

In addition to Linux developers focused on Apple Silicon being busy enabling functionality for the M3 and newer SoCs under the Linux kernel, there are also developers working on enabling the A18 Pro SoC used by the new, budget-friendly MacBook Neo laptop. Patches were posted this week for initial enablement of the A18 Pro and the MacBook Neo Device Tree... But even with these patches, the Linux support is still severely limited for now...

🔗 Source: https://www.phoronix.com/news/Apple-A18-Pro-MacBook-Neo-Patch

#kernel #linux

👉@sysadminoff

https://phoronix.com/news/Apple-A18-Pro-MacBook-Neo-Patch

Perplexity reveals internal serving architecture behind its search embeddings and ranking workloads

Perplexity has published the internal serving architecture behind its search embeddings and ranking workloads, exposing a three-layer design built to keep request processing, batch scheduling and GPU inference from blocking one another. The Ivy, Tulip and ROSE components support both low-latency searches and large indexing jobs, while also underpinning Perplexity’s broader AI search and API infrastructure.
Source

👉@sysadminoff

https://4sysops.com/archives/perplexity-reveals-internal-serving-architecture-behind-its-search-embeddings-and-ranking-workloads/

OpenAI’s GPT-6 Astra guide for devs

OpenAI is advising developers to prompt GPT-6 Astra to act on implied intent instead of repeatedly asking for clarification, while also limiting its tendency to produce canned AI language and excessive test runs. The guidance adds operational detail to GPT-6 Astra’s critical cybersecurity classification, making prompt and context management important parts of deploying the model safely.
Source

👉@sysadminoff

https://4sysops.com/archives/openais-gpt-6-astra-guide-for-devs/

OpenAI publishes several benchmarks for GPT-6 Astra

OpenAI has published several benchmarks for GPT-6 Astra, highlighting major gains in computer use, software engineering, scientific reasoning, and cybersecurity. The model achieved 98% on FrontierMath Tier 4, 99.9% on ARC-AGI-3, and 100% on ExploitBench, while Enterprise access remains disabled by default as Astra rolls out through ChatGPT, the OpenAI API, Microsoft Azure, and AWS Bedrock.
Source

👉@sysadminoff

https://4sysops.com/archives/openai-publishes-several-benchmarks-for-gpt-6-astra/

Доступно пользовательское окружение Noctalia 5.0

Опубликован выпуск проекта Noctalia 5.0, развивающего пользовательское окружение, способное функционировать поверх любых композитных менеджеров, поддерживающих Wayland-протокол layer-shell, например.

👉@sysadminoff

https://www.opennet.ru/opennews/art.shtml?num=66215

📰 Nouveau Lands Display Fixes For NVIDIA Blackwell GPUs As It Works Toward HDMI 2.1

This week's Direct Rendering Manager (DRM) fixes that were merged ahead of Sunday's Linux 7.3-rc2 release contain a number of fixes to the open-source Nouveau driver's display support for current-generation NVIDIA Blackwell GPUs...

🔗 Source: https://www.phoronix.com/news/Nouveau-Blackwell-Display-Fixes

#linux #opensource

👉@sysadminoff

https://phoronix.com/news/Nouveau-Blackwell-Display-Fixes