Skip to content

Uptime Lab

Hackathon every day
  • Home
  • Instagram
  • Contacts
  • English
    • English
    • Русский

About

uptime.lab
MaCluster IV M5 Max cluster 72 CPU and 128 GPU co MaCluster IV M5 Max cluster 72 CPU and 128 GPU cores, 512GB unified Ram in total Each MacBook is connected to all the others with Thunderbolt 5 (up to 120Gbit/s). But I’ll have to use Wi-Fi to connect to the cluster Wonder how it will perform in local AI inference?
Just for the sake of experiment. This system runs Just for the sake of experiment. This system runs the Nemotron 3 Super 120B without any issues. It operates at speeds of up to 77 tokens per second. Which is actually very fast. As CPU Intel N150 and "just" 16GB Ram ATX power supply is required for the GPU
Unusual stateful AI cluster 512 GB of unified RAM Unusual stateful AI cluster 512 GB of unified RAM, 80 ARM cores and 4 PFLOPs FP4 and a NVMe NAS for data storage and as a switch
Better than one local AI agent might be only four Better than one local AI agent might be only four local AI agents. Any ideas how to use them effectively to safe cloud tokens?
A long time ago, I ordered a soldering station fro A long time ago, I ordered a soldering station from iFixit and filmed what was in the box besides the station itself. I think it's really cool This is a screwdriver (without bits) And a iFixit spudger
​Will this build ever pay off compared to cloud st ​Will this build ever pay off compared to cloud storage? Not today ​Will it be easier to set up? No, definitely not ​Will you get complete control over your data? The ability to run various apps and virtual machines? And most importantly — absolute cloud independence? OH YES. 🔥 ​Behind the scenes of this pure tech ASMR build: ​🚀 A CPU that outperforms the latest Synology NAS 🪙 ​16GB of LPDDR5 (6400) RAM ​⚡️ Typical power consumption is only around 30W ​🌐 2.5GbE network ports for ultra-fast transfers ​🤫 And it runs virtually silent ​Which side are you on: the convenience of the cloud or total control? Let me know in the comments 👇
ZimaBoard 2 Essentially a handheld server. A devic ZimaBoard 2 Essentially a handheld server. A device with a unique form factor. I once thought that it was a competitor to my Compute Blade. But in reality, it's a very different device. Very different indeed. I plan to try it out as an x86 server, which I've been missing in my home lab. It's just that I mainly have ARM. I plan to install Docker and possibly run tests for future projects. However, I'm also curious to see how it will perform as a NAS. And if you connect a video card, you can run AI. Is anyone already using this board at home? The specifications look very tempting: Intel® N150 quad-core CPU, 3.6 GHz Up to 16GB LPDDR5x Up to 64GB eMMC 2 × 2.5GbE LAN ports 2 × SATA 3.0 2 × USB 3.1 1 × PCIe 3.0 1 × Mini DisplayPort 1.4, 4K@60Hz I added links to Link Tree in my profile if you are interested in where to buy it and how much it costs. The documentation is quite impressive as well, by the way.
Bambu lab Vortek in work I really enjoy watching Bambu lab Vortek in work I really enjoy watching this printer work. Despite the comments and doubts, the 3D-printer has already clocked up more than 1,000 hours without any maintenance whatsoever. It's about time. (And the H2D has clocked up more than 2,500 hours)
But can it run Crysis? But can it run Crysis?
Open in Instagram

Categories

  • Cool stuff
  • Finished
  • In Progress
  • Main
  • Raspberry Pi

Other projects

  • UniEDC: Painless direct connection to any Linux edge device
  • Raid on Raspberry Pi with Compute Blade
  • Compute Blade Auto-Reboot in case of problems
  • Compute Blade PCIe 3.0
  • Firmware update Raspberry Pi CM5 (Compute Module 5)
Uptime Lab | Copyright © 2021-2026 Uptime Lab, Ivan Kuleshov.

All the content and materials appearing on uplab.pro are the property of their respective owner. Unauthorized use and/or duplication of this material without express and written permission from this site’s author and/or owner is strictly prohibited.