File size: 1,708 Bytes
0e83daa
 
 
 
 
875cad3
0e83daa
 
 
fea9424
 
875cad3
fea9424
4f01886
 
875cad3
 
65d7347
 
 
fea9424
 
a0fdb46
10a11e5
a0fdb46
10a11e5
4f01886
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
---
title: README
emoji: 😻
colorFrom: gray
colorTo: purple
sdk: gradio
pinned: false
---

Tenstorrent is a next-generation computing company building AI accelerator and RISC-V CPU designs. Tenstorrent is bringing to market highly programmable chips at an accessible price point as a means of driving innovation. 
Tenstorrent has developed a proprietary processor core called the Tensix which it uses to make its own AI chips, but also sells its IP and other technology to customers looking to make their own AI chips.

Tenstorrent has two open source software stacks:
TT-Forge (https://github.com/tenstorrent/tt-forge), and 
TT-Metalium (https://github.com/tenstorrent/tt-metal/)

To contact us, please reach to our Discord channel (https://discord.gg/tenstorrent)

-------

TT-Metalium’s bare-metal programming stack defines a low-level programming model to allow users to write custom kernels for deployment on Tenstorrent hardware.
The software stack comes with debug and performance profiling tools as well as Tenstorrent's Python-based TT-NN library, which can run AI models out of the box without needing to deal with the complexities of the hardware. This negates theneed to write custom kernels unless the application requires it.

TT-Metalium will be used to fine-tune performance for production deployments, where small efficiencies can be multiplied by the scale of the deployment, and to develop custom kernels for scientific computing, where operations are not well-supported by existing kernel libraries.

For more details on the software, please visit https://tenstorrent.com/software/tt-metalium/
and the documentation https://docs.tenstorrent.com/tt-metal/latest/tt-metalium/index.html