The Koyeb server cloud platform allows developers to turn Tenstorrent’s AI accelerators

Just a few weeks after the funding of Chipmaker’s Tenstorrent, Chipmaker Tenstorrent brought nearly $ 700 million, developers can now try Tenstorrent’s AI accelerators Kind. Tenstorrent sells AI processors built around the RISC-V Set architecture and has developed its own Library Library, TT-NN and low-level low-level programming model, TT-metalium.

Tenstorrent is part of a group of companies trying to create alternatives to the NVIDIA GPU and the company’s Cuda library. He fights with Axelera, Getched, Groq and more.

Koyeb was founded by former Scaleway executives and focuses on developing a cloud -free platform for developers looking for a layer of abstraction at the cloud infrastructure level. Competes with compatriots Fly.; Railway and Make.

Koyeb allows developers to develop applications in various virtual engines using a command line interface or a git push after integrating with the code repository. It supports docker containers and many popular languages.

One of the main features of Koyeb is that it can automatically scale an application to hundreds of servers, if necessary, and when there is less movement, it can automatically reduce the server infrastructure.

In recent months, Koyeb has specifically focused on AI applications. Due to the nature without a platform server, it can provide a low delay experience for the AI ​​workload.

On the front of the material, Koyeb has developed the Tenstorrent’s PCIE tables in his data centers. Developers can access TT-metalium’s low-level TT-metalium SDK to write host and core programs.

Developers will find two new types of cases in the documentation and Koyeb administrator tables:

• The presence of TT-N300s has 24GB of SRAM GDDR6, 192MB, and provides up to 466 FP8 TFLOPS. Combined with 64GB of RAM and 4 VCPUS.
• The presence of TT-Loudbox has four N300s. Developers receive 96GB GDDR6, 768MB of SRAM and up to 1,864 FP8 TFLOPS. It has 256GB of RAM and 16 VCPUS.

With this version, Koyeb tries to be positioned as a material-consensual cloud platform. “This reminds us of the ARM debut in the high performance server market,” said co -founder and CEO of Koyeb Yann Leger at Techcrunch.

“Since we imported the ARM to the market with scaleway back in the days, offering fully customized servers in 2013-2014, we have the experience of developing various architecture and the operation of different material,” he added.

As for Tenstorrent, Ai Chipmaker is looking for partners to build an ecosystem of developer around him Model Planning Open Code. It will take a village to offer an alternative to Nvidia’s stack.