This thread has been locked.
If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.
Tool/software: Code Composer Studio
Hi,
I am doing light-weight Machine Learning with the MSP430FR5994 launchpad. I trained the model off-line and I need to upload the trained model data to the launchpad. I am using CCS to program the board. The trained model data should be saved in a float_point array which resides in FRAM. I used PERSISTENT to declare the variable.
#pragma LOCATION(model, 0x14800); #pragma PERSISTENT(model); float model[width][length] = {0};
How can I write pre-trained model data into this variable while I am programming the board with CCS?
Thanks,
Yubo
Hi Yubo,
If the data won't be modified, then using const to declare the variable is more suitable. There are two suggestions to upload the model data.
1. Declare the float
model[width][length]
variable in a C file with it's initial value as below. These can be performed by a little PC program which reads the model data file in PC and generates the C file automatically.
float model[width][length] = { 0.1, 0.2, 0.3, ...};
2. Using BSL to upload the model data to it's memory space. This can update the model date without re-programming the code. Please see MSP430 BSL page for more information.
Hi Winter,
Sorry, I did not explain well. My system does on-device training too, which means the float model array needs to update on the fly and I can not use const. Anyway, I will try your suggestions.
Thanks,
Yubo
OK. I'll close this thread. You can still reply this post if any questions or create a new post.
**Attention** This is a public forum