❌

Normal view

There are new articles available, click to refresh the page.
Today — 23 July 2026Main stream

Industrial GPU Adapted for the Desktop

✇Hackaday
By: Bryan Cockfield
23 July 2026 at 01:00

As technologies change and adapt, we’re often left with seemingly useless junk that has nowhere to go. Certainly anyone still sitting on a pile of floppy disks feels this way sometimes, but odds are anyone who owns a mining ASIC or an NFT can attest to that as well. The trillions of dollars flowing into GPU-based data centers will likely become the next victim of this trend, so if you want to capitalize on the losses of some venture capitalist you’ll want to figure out a way to get GPUs meant for a server into your desktop doing useful work.

Of course, calling these devices GPUs is a bit of a stretch compared to the Radeon and GeForce cards many of us are used to using for gaming. These don’t even have a PCIe slot or video output, after all. But, as [Oscar] notes, the VRAM and GPU cores are very real and can still do useful work. An adapter board is able to mate a Tesla V100 SXM2 16 GB GPU to a standard PCIe slot, which solves the first problem, but the major downside from there is that the cooling fan for this unit was literally deafeningly loud. At 82 dB it was about as loud as a lawnmower, which is fine in a server rack but not great in a bedroom. [Oscar] found a way to tamp down the fan speed, making it usable in a home.

Without video output, the utility of these cards mainly comes from adding VRAM and compute for tasks that benefit from parallel computing. Using tensor splitting, [Oscar] is running a local LLM with this card alongside his RTX 4080, providing 32 GB of VRAM on his NixOS system. With his benchmarking tests, the LLM sports impressive stats for a self-hosted model, ranking somewhere around Claude Sonnet 4.6. What’s even more impressive is that this is all done for around £200, and with the rate the various LLM companies are ratcheting up pricing could pay itself back very quickly. If trading off performance for cost is acceptable, though, it’s possible to run local models on much less powerful hardware as well.

Yesterday — 22 July 2026Main stream
Before yesterdayMain stream

UK data center startup Nscale bets big on Bellevue for U.S. engineering hub amid AI boom

✇GeekWire
By: John Cook
20 July 2026 at 12:37
Nscale’s Nidhi Chappell. Photo via Nscale.

Fresh off a $2 billion fundraising and $900 million line of credit, London-based data center startup Nscale is planning a big expansion at a new engineering office in Bellevue, Wash.

Nscale, one of the fastest-growing companies building AI computing infrastructure, recently inked a deal for nearly 24,000 square feet of space at The Eight office tower in downtown Bellevue.

The office is slated to open in January 2027. It will serve as Nscale’s primary engineering hub in the United States, a company spokesperson said. The company currently employs about 50 people in the Seattle area, and the new office will be able to accommodate up to 250 people.

The company earlier this year hired Nidhi Chappell, the former Microsoft corporate vice president who led Azure AI and high-performance computing infrastructure, including the supercomputers that power ChatGPT. As Nscale’s new president of AI infrastructure, based in the Seattle area, Chappell will oversee the company’s global engineering and data center operations.

“I’ve had a front-row seat to some of the biggest moments in AI over the past several years, but one thing has always stood out: the world remembers the breakthroughs, but it’s the people building the infrastructure behind the scenes who make them possible,” Chappell wrote in a LinkedIn post last week announcing the company’s first “onboarding” event in Seattle.

Nscale, which is also preparing to open an office in New York, said it selected Bellevue because of the Seattle region’s concentration of AI infrastructure talent and its proximity to major customers.

Microsoft is one example. Earlier this year, the companies announced an expanded collaboration to deploy Microsoft’s next-generation AI infrastructure across Europe, including large-scale installations of NVIDIA Vera Rubin GPUs in Norway, Portugal and other locations. Nscale said it would be among the first providers outside of Microsoft to deploy the Vera Rubin platform, supporting Microsoft’s growing AI cloud infrastructure.

The new office is the latest sign of Bellevue’s growing role in the AI economy. The Eastside has become a magnet for companies building AI applications and infrastructure, with xAI, OpenAI, Databricks, CoreWeave, Armada, Anduril and others establishing and expanding offices.

AI companies have been giving a boost to the regional office market overall. Claude maker Anthropic, for example, recently announced an expansion of its offices in Dexter Yard in Seattle.

Nscale was founded in 2024. Its $2 billion funding round earlier this year valued the company at $14.6 billion, believed to be the largest Series C financing ever raised by a European technology company. The capital is being used to expand Nscale’s AI cloud platform, GPU infrastructure and data center footprint across North America and Europe.

Its backers include Astra Capital Management, Citadel, Dell, Jane Street, Lenovo, Linden Advisors, Nokia, NVIDIA and Point72.

News of the Nscale office in Bellevue was first reported by the Puget Sound Business Journal.

Porting the Nvidia GPU Driver to Haiku for 3D Acceleration

✇Hackaday
By: Maya Posch
12 July 2026 at 16:00

As good as a desktop OS may be, at some point it has to feature accelerated 3D graphics. This has been a bit of a sticking point for Haiku OS, as none of the big names in GPU cards are likely to start putting out drivers for this OS any time soon. Fortunately there is the Linux open source driver code from Nvidia that can be used as a jumping-off point for a port, which is what [X512] and the community over at the Haiku forums did over the course of more than a year.

In a recent video [Action Retro] takes a poke at the fruits of these efforts, trying out the driver with an RTX2070 Super GPU. Of note is that this driver requires the GSP (GPU System Processor) controller that got added by Nvidia with the Turing series of GPUs, meaning that you need at least a GTX16 or RTX20 series card.

You can get an installation package from the GitHub repository, such as for the v0.0.2 pre-release that was created in January of 2026. In this pre-release state quite a few things are working, with the ability to play 3D games at a reasonable FPS being the biggest improvement over plain VESA mode. Features like CUDA are not available as they’re not in the open sourced section, of course.

In the [Action Retro] video the whole installation process is demonstrated, starting with a fresh nightly Haiku build. First the gaming performance in software-rendered VESA mode is demonstrated before the GPU driver is installed. This shows a marked improvement in performance, although Minecraft needs to be updated for the newest Mesa library that omits OSMesa, so that couldn’t be tested. Overall it shows that Haiku has made another massive leap forward in becoming a viable daily driver OS.

Meanwhile, over on the ReactOS side of things we just saw a Half-Life 2 playthrough by [Aotori Hibiki], on an Intel Sandy Bridge PC with GeForce 8400GS graphics. Here ReactOS has the advantage of being Windows NT-compatible, including WDDM-style GPU drivers, allowing it to use the same drivers as Windows. Simultaneously, ReactOS is now implementing its first NT6 kernel API calls to make it compatible with modern  (Vista+) Windows.

The upshot here is that for people who want to daily drive an open source OS with all the creature comforts imaginable, things have never seemed more promising. Especially for people who don’t want Yet Another Linux Distro but just an utterly boring desktop-centric, single-user focused OS that Just Works™ these are great tidings.

❌
❌