#bare-metal — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #bare-metal, aggregated by home.social.
-
Are you looking to deploy robust IT infrastructure in South Asia? 🛑 Stop routing traffic through distant international hubs!
Explore iDatam's Bare-Metal Dedicated Servers in India! Located in prime hubs like Mumbai, Bangalore, Chennai, and New Delhi, you get unshared Intel/AMD compute power, unmetered bandwidth, and sub-millisecond local latency.
Check out our available hardware here: https://www.idatam.com/dedicated-servers-india/
#WebHosting #SysAdmin #DataCenter #Linux #India #BareMetal #TechNews
-
Looking for the ideal dedicated server specs for Docker and Kubernetes workloads? 🐳⚙️
Our new engineering guide is live! We break down how to optimize:
🔹 AMD EPYC & Intel Xeon high core-count CPUs
🔹 DDR5 ECC memory ratios to prevent OOMKilled errors
🔹 PCIe Gen 5 NVMe storage for etcd stability
🔹 100Gbps networking & DPU offloadingRead the full tutorial here:
https://www.fitservers.com/blogs/dedicated-server-specs-docker-kubernetes/#Kubernetes #Docker #DevOps #BareMetal #SysAdmin #CloudNative
-
Are you looking to deploy robust IT infrastructure in Europe? 🛑 Stop dealing with slow cross-Atlantic routing!
Explore iDatam's Bare-Metal Dedicated Servers in the Netherlands! With locations in Amsterdam, you get unshared Intel/AMD processing power, unmetered bandwidth, and ultra-low latency across the EU via the AMS-IX.
Check out our available hardware here: https://www.idatam.com/dedicated-servers-netherlands/
#WebHosting #SysAdmin #DataCenter #Linux #Amsterdam #BareMetal #TechNews
-
Are you looking to deploy robust IT infrastructure in Europe? 🛑 Stop dealing with slow cross-Atlantic routing.
Explore iDatam's Bare-Metal Dedicated Servers in Germany! With locations in Frankfurt, Munich, Berlin, and Nuremberg, you get unshared Intel/AMD processing power, unmetered bandwidth, and ultra-low latency across the EU.
Check out our available hardware here: https://www.idatam.com/dedicated-servers-germany/
#WebHosting #SysAdmin #DataCenter #Linux #Germany #BareMetal #TechNews
-
The control plane tax is gone. Next: your hypervisor tax.
Introducing kMetal 1.0 by CLASTIX!
Real #Kubernetes tenant isolation on shared #BareMetal: a separate kernel + network per tenant. No dedicated hardware, no separate #hypervisor layer to license or run.
-
⌨️ BMC64 5.0.0 released.
“BMC64 is a #C64 emulator for the #RaspberryPi. Recent updates: Improvements for Wi-Fi and Ethernet, BBS access, C64 OS networking and USB. And support for #8BitDo Retro devices. #BMC64 is tested on Raspberry Pi Zero W, Zero 2 W, 2B, 3A+ and 3B+.”
-
“The #Amiga -derived #AROS Research #OperatingSystem has just been ported to boot #BareMetal on the #RaspberryPi, in both 32-bit and 64-bit versions.”
https://hackaday.com/2026/08/23/amiga-inspired-aros-goes-bare-metal-on-raspberry-pi/
-
Shared server misconfigurations can cost US businesses millions in CCPA & GDPR penalties.
Multi-tenant cloud & shared hosting expose you to:
❌ Cross-tenant leaks ("Noisy Neighbors")
❌ Shared IP blacklisting
❌ Lack of root-level security controlBare metal dedicated servers provide 100% physical data isolation, strict datacenter control, and audit-ready logging to keep your infrastructure compliant.
https://www.servers99.com/blog/ccpa-gdpr-compliance-dedicated-servers/
#DataPrivacy #GDPR #CCPA #CyberSecurity #BareMetal #SelfHosting
-
Мой код терял бы 230 байт из 231. Чтобы это увидеть, пришлось патчить QEMU
Через четыре часа после первой статьи пришёл комментарий: в моём коде не хватает трёх инструкций, и на настоящей плате он рассыплется. Читатель arteast был прав. Сел воспроизводить поломку. Собрал два варианта, свой и правильный, прогнал на строке в 231 байт. Вывод побайтово одинаковый. Замедлил линию до предела. Ничего не изменилось. Написал программу, которая смотрит, бывает ли передатчик занят хоть когда-нибудь. При выводе в консоль не бывает. Моя ошибка внутри эмулятора не проявляется вообще. Проверить себя было нечем. Пришлось чинить эмулятор. Ну, чини!
-
Оценка скорости реакции RTOS на события
продолжение темы RTOS, начало см. https://habr.com/ru/posts/1060466/ https://habr.com/p/1062346/ Получив результат, а именно, запуск SMP версии RTOS, задался вопросом, а какая же плата в виде ухудшения временнЫх характеристик взимается за межъядерное взаимодействие сервисов RTOS? Достал из пыльного чулана давно написанный тест производительнсти сервисов RTOS и причесал с целью использования в одно- и много-ядерном режиме.
-
El sesgo tecnológico actual es absurdo: muchas organizaciones prefieren el gasto recurrente y la dependencia de plataformas de terceros antes que gestionar su propia infraestructura.
Le tienen más miedo a asumir la responsabilidad técnica que al desperdicio financiero. La soberanía digital no solo ahorra recursos, expone la inercia del sector. ⚙️
#SelfHosted #DeClouding #SysAdmin #Linux #OpenSource #BareMetal #SoberaniaDigital #DevOps
-
La «repatriación de la nube» ya es una realidad. Pagar fortunas por servicios en AWS o Azure sin entender el motor debajo ya no es sostenible.
Cada vez más empresas están volviendo al hardware propio y a la administración Bare Metal: menor costo, cero dependencia y soberanía real sobre los datos.
El control volvió a la infraestructura propia. ⚙️
#DeClouding #BareMetal #Linux #SelfHosted #DevOps #SysAdmin #OpenSource #Privacidad
-
¿Docker o Bare Metal para un servidor de Minecraft? 📦 vs 🖥️
Docker (usando imágenes como itzg/minecraft-server) es ideal para aislamiento y despliegue rápido. Sin embargo, en servidores con cientos de plugins o modpacks pesados, el acceso directo en Bare Metal exprime cada ciclo de CPU al reducir el overhead de E/S.
La arquitectura manda. ⚙️
#SelfHosted #Minecraft #Docker #BareMetal #Linux #SysAdmin #DevOps
-
RTT без JLINK
“Обнаруженный в кустах рояль” под названием IAR i-Jet TRACE позволяет полноценную отладку SMP (Symmetric Multiprocessing) приложения, запущенного на нескольких ядрах многоядерного процессора, в данном случае это Allwinner T113. Полноценная это значит: возможность старт-стоп отдельно каждого ядра, установка breakpoints для каждого ядра, просмотр регистров каждого ядра
-
Stop burning money on raw TFLOPS—Generative AI inference is memory-bound!
While NVIDIA H100 and H200 share identical compute, memory bandwidth changes everything:
• H200 (141GB HBM3e @ 4.8 TB/s) gives 1.9x faster LLM inference
• 1.1 TB node VRAM fits 70B parameter models without sharding
• Zero 84°C thermal throttling using high-density bare metal cooling
• No 10–15% cloud hypervisor tax for maximum token output speeds -
🤖💻 Frame: Un nou server grafic X11 pentru Linux scris în întregime în Assembly
Într-o epocă în care industria Linux migrează masiv către Wayland și dezvoltă software complex în limbaje moderne, un proiect incredibil vine să zguduie status quo-ul prin tehnica pură: Frame. Acesta este un server grafic compatibil X11, scris de la zero în întregime în Assembly (ASM).
Proiectul reprezintă un tur de forță ingineresc, demonstrând ce se poate obține atunci când elimini complet straturile intermediare de cod, compilatoarele greoaie și dependințele masive, vorbind direct pe limba procesorului.
Iată caracteristicile uluitoare ale acestui proiect minimalist:
🔹 Performanță brută și latență zero
Deoarece codul este scris direct în Assembly, Frame ocolește toate optimizările (sau limitările) pe care un compilator de C sau C++ le-ar impune.Latență minimă: Timpul de răspuns între input-ul utilizatorului (mouse/tastatură) și randarea pe ecran este redus la minimul absolut permis de hardware.
Fără supraîncărcare (Zero Overhead): Serverul nu încarcă biblioteci externe gigantice, scăzând drastic numărul de instrucțiuni executate de procesor pentru fiecare cadru (frame) randat.
🔹 Dimensiune microscopică a binarului
Spre deosebire de serverul clasic Xorg, care are nevoie de zeci sau sute de megabiți de biblioteci și dependențe pentru a funcționa, binarul executabil al Frame ocupă doar câțiva kilobiți. Întregul server grafic poate fi stocat pe o veche dischetă, fiind probabil cel mai mic server de afișare funcțional creat vreodată pentru Linux.🔹 Compatibilitate cu protocolul X11
În ciuda naturii sale experimentale și a limbajului de nivel jos în care este scris, Frame implementează elementele de bază ale protocolului X11. Acest lucru înseamnă că poate, teoretic, să comunice cu clienți X standard, permițând rularea aplicațiilor grafice de bază din ecosistemul Linux direct pe această fundație ultra-ușoară.🔹 Eficiență extremă a resurselor
Consumul de memorie RAM al serverului se măsoară în câțiva megabiți (sau chiar mai puțin în modul idle), făcându-l o opțiune fascinantă pentru:Sisteme integrate (Embedded): Dispozitive cu resurse hardware extrem de limitate, unde fiecare octet de RAM contează.
Sisteme de restaurare și recuperare: Medii de boot minimale care au nevoie de o interfață grafică rapidă, fără a încărca RAM-disk-ul.
Frame este un proiect manifest dedicat programării retro-futuriste, o dovadă vie că optimizarea extremă nu a murit și că bătrânul protocol X11 poate fi reinventat în cel mai pur mod cu putință.
#OpenSource #FrameServer #Assembly #X11 #LinuxGraphics #BareMetal #LowLevel #Minimalism #TechNews #Linuxiac
-
At some point, every application became a landlord—and the rent included my privacy, data and increasingly creative interpretations of consent.
So I rented two bare-metal servers, deployed far too many databases, and built one very expensive way to say: my life is not an untapped dataset.
https://augustini.wtf/blog/why-i-built-a-bare-metal-cluster/
#SelfHosting #Privacy #DigitalSovereignty #BareMetal #OpenSource
-
О Raspberry Pi Pico, RP2040 и Baremetal
В этой статье пойдет речь о первом поколении микроконтроллеров от Raspberry Pi Foundation — RP2040 на основе двух ядер ARM Cortex‑M0+, и о плате, которая базируется на этом чипе — Raspberry Pi Pico. Статья призвана облегчить понимание некоторых принципов работы RP2040 и микроконтроллеров в целом.
-
Stop relying on blind scripts to install Pterodactyl Panel. When they fail, you get a 500 error with no context. 🎮
Our SRE guide walks you through a clean manual deployment on Bare Metal:
• Secure WebSocket console routing (port 8080)
• Fix root file ownership permission traps
• Restore flatlining CPU/RAM telemetry graphsReclaim absolute control over your game hosting performance.
Read the full guide on our site:
https://www.irexta.com/tutorials/install-pterodactyl-panel-bare-metal/ -
Ich habe seit ca. 5 Jahren Nextcloud stabil laufen. Allerdings nur mit den Standartsapps, also keine externen APPs oder Plugins.
Das schwierigste ist wenn NC, eine PHP-Version braucht die das OS nicht ausliefert, oder nicht die passende DatenbankVersion für die neue NC-Version verfügbar. Das ist doof.
Ehrlich ohne CLI-Kenntnisse lieber Fingerweg vom Selberhosten.
-
BareMetal RAM Dumper – Bare-metal x86 tool for Cold Boot Attack experiments
https://github.com/pIat0n/BareMetal-RAM-Dumper
#HackerNews #BareMetal #RAM #Dumper #Cold #Boot #Attack #x86 #Security #Research #Tool
-
#Crowdsec has been re-enabled on #Midgaard - the #Kubernetes cluster running #MSTDNDK. We're doing this without CloudFlare and virtualization - it's #baremetal and homegrown #firewall all the way.
Should you experience connectivity issues, either as a user or a federated instance, please reach out to us.
-
🛠️ alexellis/awesome-baremetal
Bare-metal is awesome. Let's share our favourite tools.
Shares a curated list of favorite bare-metal tools for developers, system administrators, and hardware enthusiasts
⭐ Stars: 1910
📅 Last Update: Jun 10, 2026https://github.com/alexellis/awesome-baremetal
#selfhosted #homelab #selfhost #selfhosting #opensource #baremetal #tools
-
Agentic AI destroys old hardware assumptions.
Generative AI demanded massive GPUs. Agentic AI demands extreme CPU core density. Stop overprovisioning expensive accelerators! Learn why:
The 1:8 CPU-to-GPU ratio is dead (1:2 or 1:1 is the future)
GPUs sit idle waiting for CPU orchestration
Autonomous agents require MicroVM sandboxing for security
Public cloud egress fees will bleed your budget
Read the full guide by @irexta:
🔗 https://www.irexta.com/blogs/agentic-ai-cpu-gpu-hardware-requirements/ -
Optimize AI Cluster Networks with Multi-Rail RoCEv2
Standard Ethernet stalls GPU training with packet drops and ECMP hash collisions. Master the SRE fabric playbook: Bypass the OS kernel with RDMA, enforce lossless PFC (use watchdogs to prevent deadlocks!), and use Multi-Rail PCIe affinity to dedicate physical NICs directly to GPUs.
Read the bare-metal architecture guide by @ServerMO:
🔗 https://www.servermo.com/blogs/multi-rail-rocev2-ai-cluster/#SRE #DevOps #AI #Networking #BareMetal #RoCEv2 #MachineLearning