Skip to content

Releases: SweiryDev/MicroCNN-TangNano20k

Release list

Bias ROM Fix and MIT License

Bias ROM Fix and MIT License Latest
Latest

Choose a tag to compare

View all tags
@SweiryDev SweiryDev released this 08 Oct 08:22
V1.0.1
2711cdd
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

This release contains the pre-compiled bitstream (NPU.fs) for the MicroCNN Accelerator, ready to be flashed directly to the Sipeed Tang Nano 20K.

How to Flash using Gowin Programmer:

Connect your Tang Nano 20K to your PC via USB.
Open the Gowin Programmer tool.
Click the Scan Device icon to detect the FPGA (it should identify as GW2AR-18C).
Double-click the blank cell under the File Name column for your detected device.
In the Device configuration window:
Set Access Mode to SRAM Program (for temporary testing) or Embedded Flash Mode (for persistent storage).
In the File name field, browse and select the downloaded NPU.fs file.
Click Save.
Click the Program/Configure (Play button) icon in the top toolbar to flash the board.
Once the progress bar completes successfully, the accelerator is running and waiting for data. You can now use the uart_medmnist_loader.py script provided in the main repository to stream images and test the inference.

Version Changes:

  • Added MIT License to the repository.
  • Merged PR from @philtomson fixing an off-by-one scale issue in the bias ROMs.
  • Updated .fs bitstream to reflect the mathematically accurate bias values.

Contributors

philtomson
Assets 3
Loading

Initial Release - MicroCNN Bitstream

Initial Release - MicroCNN Bitstream

Choose a tag to compare

View all tags
@SweiryDev SweiryDev released this 10 Aug 12:40
v1.0.0
92019ac

This release contains the pre-compiled bitstream (NPU.fs) for the MicroCNN Accelerator, ready to be flashed directly to the Sipeed Tang Nano 20K.

How to Flash using Gowin Programmer:

  1. Connect your Tang Nano 20K to your PC via USB.
  2. Open the Gowin Programmer tool.
  3. Click the Scan Device icon to detect the FPGA (it should identify as GW2AR-18C).
  4. Double-click the blank cell under the File Name column for your detected device.
  5. In the Device configuration window:
  • Set Access Mode to SRAM Program (for temporary testing) or Embedded Flash Mode (for persistent storage).
  • In the File name field, browse and select the downloaded NPU.fs file.
  • Click Save.
  1. Click the Program/Configure (Play button) icon in the top toolbar to flash the board.

Once the progress bar completes successfully, the accelerator is running and waiting for data. You can now use the uart_medmnist_loader.py script provided in the main repository to stream images and test the inference.

Loading