To help customers develop more efficiently and conveniently based on CIM chips, we have created a complete set of chip development tools and IDE environment. At the Developer Center, we will continuously share and update software, compilation toolchains, algorithm resources, and development tutorials for CIM chips, providing tools and inspiration for engineers, researchers, and technology enthusiasts. We also warmly welcome your sharing, suggestions, new ideas and even development achievements. Have Fun!
Software

Computer software

SDK and real-time operating system
Witmem Studio IDE
A full-featured integrated development environment, which includes the SDK push function for customer recognition, the automatic update, download and installation function of SDK packages, the kernel automatic recognition syntax highlighting editor, the personalized project creation function for different functions, as well as the project management, file editing, compilation, debugging and other functions that regular ides have.
Hearing aid console software
A Witmem-developed PC-end software platform specifically for generating different hearing aid solutions. Through the graphical control platform, users can generate various algorithm parameter configuration schemes more conveniently and quickly.
6FDSP Automated assembly instruction generator...
This is an upper computer software developed by Witmem, which is used to generate assembly instructions quickly. Users can form an instruction flow diagram by dragging the instruction box on the desktop, and then generate the corresponding assembly instructions for the instruction flow diagram with one click.
SPI_Moniter
An auxiliary research tool for real-time monitoring of on-chip data, including audio sample points, frequency domain vectors, key scalar values, etc. It can assist algorithm engineers and system engineers in analyzing algorithm performance, variable behavior, debugging, etc.
2302 Firmware download tool
This is an upper computer software that connects to the upper computer via a USB interface. The upper computer automatically selects the protocol of the lower computer by configuring the parameters of the I2C/SPI/UART protocol and burns the code according to the selected protocol. After the upper computer is connected to the PC via USB, only operations need to be performed on the upper computer. The physical connection of the lower computer does not need to be changed to switch among the three protocols....
FDSP Parameter adjustment software
This is an upper computer software developed by us, which can set the parameters of the self-developed FDSP chip and configure the parameters of BQ, WFIR and Limter for the lower device through the serial port.
WTM2101 Tool Chain
witin_mapper is a self-developed compilation software stack for neural network mapping by Witmem. It can map the quantized neural network model to the WTM2101 MPU accelerator. It is a complete solution including RISC-V and MPU, and can complete the transformation and optimization at the operator and graph levels. The pre-training weights are arranged into the computing-memory array, and the computing-memory optimization scheme is given for the network structure and operators. Meanwhile, the operators that are not suitable for MPU operations are scheduled to be operated on the CPU to achieve the scheduling of the entire network, enabling neural network developers to run the trained algorithms on the WTM2101 chip efficiently and quickly. Greatly shorten the development cycle of model transplantation and improve the efficiency of algorithm development.
View Details

Develop Board
WTMDK2101-ZT1

Designed for WTM2101 AI SOC
Can be used for the WTM2101-S1
core board test. And the development of
a general solution for hearing AIDS.
View Details
core board test. And the development of
a general solution for hearing AIDS.
WTMDK2101-X3

Designed for WTM2101 AI SOC.
Burn the demonstration program (DEMO)
and power on to run the demonstration.
View Details
Burn the demonstration program (DEMO)
and power on to run the demonstration.
mnist demo

Classification based on the mnist dataset
A full-process tutorial for the model from training to deployment.
Please refer to the development documentation
View details.
View Details