Maxim

https://github.com/MaximIntegratedAI/MaximAI_Documentation/tree/master/MAX78000_Feather
https://github.com/MaximIntegratedAI/MaximAI_Documentation/tree/master/MAX78000_Evaluation_Kit

Getting Started with the MAX78000EVKIT
https://www.youtube.com/watch?v=2rUFPAfajP8

https://analog-devices-msdk.github.io/msdk/USERGUIDE/

## FAIL.txt
## File sent out of order by PC. Target might not be programmed correctly.

• Power shortage on the connected USB port would be a possible cause of this failure; disconnect all other USB peripherals (if possible) and try again

• Baud rate = 115200
• Data bits = 8
• Stop bits = 1
• Flow control = None
• Parity = None

https://github.com/Analog-Devices-MSDK/msdk/tree/main/Examples/MAX78000/Hello_World

Hello World!
count : 0
count : 1
count : 2
count : 3
count : 4
count : 5
...

https://github.com/Analog-Devices-MSDK/msdk/tree/main/Examples/MAX78000/CNN/mnist

Waiting...

*** CNN Inference Test ***

*** PASS ***

Approximate inference time: 1409 us

Classification results:
[ -77992] -> Class 0: 0.0%
[  93377] -> Class 1: 0.0%
[  45484] -> Class 2: 0.0%
[  29850] -> Class 3: 0.0%
[ -36248] -> Class 4: 0.0%
[-130974] -> Class 5: 0.0%
[-272399] -> Class 6: 0.0%
[ 416616] -> Class 7: 100.0%
[-148783] -> Class 8: 0.0%
[  25073] -> Class 9: 0.0%

https://docs.nvidia.com/deeplearning/performance/dl-performance-convolutional/index.html
https://gist.github.com/mingfeima/595f63e5dd2ac6f87fdb47df4ffe4772
https://stackoverflow.com/questions/24563786/conversion-from-hex-to-signed-dec-in-python
https://qiita.com/filu_/items/09c1d897b4d470c10336

• Defined in mnist/sampledata.h as NCHW memory layout
• 1-channel 28×28 pixel data (784 bytes) represented by 196×32-bit little-endian hex words consisting of 4(pixels)x8-bit signed integers (0x80-0x7f)
  • e.g., 0x3ad48080 > [-128, -128, -44, 58]

maxim_mnist_sampledata.py.gz

def s08(value) :
    return -(int(value, 16) & 0x80) | (int(value, 16) & 0x7f)

vf = np.vectorize(s08)
img_dec = vf(img)

plt.imshow(img_dec, cmap='gray')
plt.colorbar()
plt.show()

https://github.com/Analog-Devices-MSDK/msdk/tree/main/Examples/MAX78000/CNN/cifar-10

Waiting...

*** CNN Inference Test ***

*** PASS ***

Approximate inference time: 4570 us

Classification results:
[-248970] -> Class 0: 0.0%
[-233651] -> Class 1: 0.0%
[-180131] -> Class 2: 0.0%
[ 199318] -> Class 3: 100.0%
[-197772] -> Class 4: 0.0%
[   4497] -> Class 5: 0.0%
[-144072] -> Class 6: 0.0%
[-223746] -> Class 7: 0.0%
[-249736] -> Class 8: 0.0%
[-259639] -> Class 9: 0.0%

https://docs.nvidia.com/deeplearning/performance/dl-performance-convolutional/index.html
https://gist.github.com/mingfeima/595f63e5dd2ac6f87fdb47df4ffe4772
https://stackoverflow.com/questions/24563786/conversion-from-hex-to-signed-dec-in-python
https://qiita.com/filu_/items/09c1d897b4d470c10336

• Defined in cifar-10/sampledata.h as NHWC memory layout
• 3-channel 32×32 pixel data (3072 bytes) represented by 1024×32-bit little-endian hex words consisting of 4(rgb+dummy)x8-bit signed integers (0x80-0x7f)
  • e.g., 0x00b1f01f > [R=31, G=-16, B=-79]

maxim_cfiar-10_sampledata.py.gz

def s08(value) :
    return -(int(value, 16) & 0x80) | (int(value, 16) & 0x7f)

vf = np.vectorize(s08)
img_dec = vf(img) + 128

plt.imshow(img_dec)
plt.colorbar()
plt.show()

https://analog-devices-msdk.github.io/msdk/