Jim Keller's AI chip company 'Tenstorrent' releases PCIe expansion cards 'Grayskull e75' and 'Grayskull e150' specializing in AI inference & also announces cooperation with Japan's LSTC and Rapidus
On March 5, 2024, Tenstorrent , an AI startup whose CEO is
Cards - Tenstorrent
https://tenstorrent.com/cards/
Today we are officially launching our Grayskull Dev Kit, available for purchase on our website. This is our 1st gen AI PCIe card - an inference-only hardware kit we are releasing alongside TT-Metallium - our open source software stack. https:// t.co/gXzm2S8xWz
— Tenstorrent (@tenstorrent) March 4, 2024
Grayskull e75 and Grayskull e150 can perform AI inference processing through Tenstorrent's AI execution software platforms ' TT-Buda ' and ' TT-Metalium '. Of both models, the Grayskull e75 is positioned as an entry model, equipped with 96 AI processing cores 'Tensix' developed by Tenstorrent, and consumes 75W of power. The price is $ 599 (about 88,000 yen).
Grayskull e150 is equipped with 120 Tensix cores and consumes 200W of power. The price is $ 799 (about 117,000 yen).
TT-Buda is a software platform that allows you to run AI models relatively easily, and AI models such as GPT-2 and Stable Diffusion are already available. You can check the list of AI models that can be used with TT-Buda on the following page.
tt-buda-demos/model_demos at main · tenstorrent/tt-buda-demos · GitHub
https://github.com/tenstorrent/tt-buda-demos/tree/main/model_demos
Tenstorrent is also implementing a plan to pay rewards to developers who make their AI models compatible with TT-Buda. At the time of article creation, ' Gemma 2B ', ' undefined ', and ' Qwen 1.5 (0.5B) ' are eligible for the bounty.
Alongside the launch of our Grayskull DevKit we are excited to announce the launch of the Tenstorrent AI Bounty Program.
— Tenstorrent (@tenstorrent) March 6, 2024
Join us in adding AI models into TT-Buda demos. Opportunities include working on Qwen-1.5 (0.5B), Phi-2 (2.7B), Gemma 2B
Find out how to get started on our… pic.twitter.com/WAXU9kFJNA
In addition, when Jim Keller announced the Grayskull e75 and Grayskull e150, he announced that he would ``announce an AI network product soon'' and ``announce the second generation (AI inference card) in April 2024.'' Masu.
Tenstorrent plan update
— Jim Keller (@jimkxa) March 4, 2024
1. open source software (Buda and Metalium) - January
2. sell Gen 1 Greyskull devkits, enable
online purchase - today
3. coming soon Gen 1 networked AI
4. Power on Gen 2 in April
5. design best possible Gen 3 low cost AI - next…
◆ Also announced cooperation with Japanese semiconductor development company
Tenstorrent announced a partnership agreement with the technology research association Advanced Semiconductor Technology Center (LSTC) on February 27, 2024. This partnership will enable LSTC to develop semiconductors utilizing Tenstorrent's IP. Tenstorrent has also announced cooperation with Rapidus , which aims to produce 2nm semiconductors.
Tenstorrent RISC-V and Chiplet Technology Selected to Build the Future of AI in Japan - Tenstorrent
https://tenstorrent.com/research/tenstorrent-risc-v-and-chiplet-technology-selected-to-build-the-future-of-ai-in-japan/
Related Posts: