ST Edge AI Core v2.2.0-20266 2adc00962
Created date          : 2025-11-07 09:55:20
Parameters            : generate --target stm32f4 --name har_github -m E:/YAL/Projet/Formation_AI_Embarquee/Training_STM32_AI_J3/har_github.h5 --compression medium --verbosity 1 --workspace C:/Users/AIXSON~1/AppData/Local/Temp/mxAI_workspace3785047036485007944985187664727254 --output C:/Users/Aix Sonic/.stm32cubemx/network_output

Exec/report summary (generate)
---------------------------------------------------------------------------------------------------------------
model file         :   E:\YAL\Projet\Formation_AI_Embarquee\Training_STM32_AI_J3\har_github.h5                 
type               :   keras                                                                                   
c_name             :   har_github                                                                              
compression        :   medium                                                                                  
options            :   allocate-inputs, allocate-outputs                                                       
optimization       :   balanced                                                                                
target/series      :   stm32f4                                                                                 
workspace dir      :   C:\Users\AIXSON~1\AppData\Local\Temp\mxAI_workspace3785047036485007944985187664727254   
output dir         :   C:\Users\Aix Sonic\.stm32cubemx\network_output                                          
model_fmt          :   float                                                                                   
model_name         :   har_github                                                                              
model_hash         :   0xdc582ba86ee8a11fd06b698b258a4310                                                      
params #           :   738,950 items (2.82 MiB)                                                                
---------------------------------------------------------------------------------------------------------------
input 1/1          :   'input_0', f32(1x90x3x1), 1.05 KBytes, activations                                      
output 1/1         :   'dense_3', f32(1x6), 24 Bytes, activations                                              
macc               :   875,232                                                                                 
weights (ro)       :   375,448 B (366.65 KiB) (1 segment) / -2,580,352(-87.3%) vs float model                  
activations (rw)   :   25,616 B (25.02 KiB) (1 segment) *                                                      
ram (total)        :   25,616 B (25.02 KiB) = 25,616 + 0 + 0                                                   
---------------------------------------------------------------------------------------------------------------
(*) 'input'/'output' buffers are allocated in the activations buffer

Model name - har_github
------ -------------------------------------- ---------------------- ------------------- --------- ----------------- --- -------------------- ------------------- ------------------------------- 
m_id   layer (type,original)                  oshape                 param/size               macc      connected to   | c_size               c_macc              c_type                          
------ -------------------------------------- ---------------------- ------------------- --------- ----------------- --- -------------------- ------------------- ------------------------------- 
0      input_0 (Input, None)                  [b:1,h:90,w:3,c:1]                                                       |                                          
       conv2d_1_conv2d (Conv2D, Conv2D)       [b:1,h:89,w:2,c:128]   640/2,560              91,264           input_0   | -2,560(-100.0%)      -91,264(-100.0%)    
       conv2d_1 (Nonlinearity, Conv2D)        [b:1,h:89,w:2,c:128]                          22,784   conv2d_1_conv2d   |                      -22,784(-100.0%)    
------ -------------------------------------- ---------------------- ------------------- --------- ----------------- --- -------------------- ------------------- ------------------------------- 
1      max_pooling2d_1 (Pool, MaxPooling2D)   [b:1,h:44,w:1,c:128]                          22,528          conv2d_1   | +2,560(+100.0%)      +114,048(+506.2%)   Conv2D_[0]                      
------ -------------------------------------- ---------------------- ------------------- --------- ----------------- --- -------------------- ------------------- ------------------------------- 
3      flatten_1 (Reshape, Flatten)           [b:1,c:5632]                                           max_pooling2d_1   |                                          
------ -------------------------------------- ---------------------- ------------------- --------- ----------------- --- -------------------- ------------------- ------------------------------- 
4      dense_1_dense (Dense, Dense)           [b:1,c:128]            721,024/2,884,096     721,024         flatten_1   | -2,523,072(-87.5%)   +128(+0.0%)         Dense_/Nonlinearity_[1, 2]      
       dense_1 (Nonlinearity, Dense)          [b:1,c:128]                                      128     dense_1_dense   |                      -128(-100.0%)       
------ -------------------------------------- ---------------------- ------------------- --------- ----------------- --- -------------------- ------------------- ------------------------------- 
5      dense_2_dense (Dense, Dense)           [b:1,c:128]            16,512/66,048          16,512           dense_1   | -57,280(-86.7%)      +128(+0.8%)         Dense_/Nonlinearity_[3, 4]      
       dense_2 (Nonlinearity, Dense)          [b:1,c:128]                                      128     dense_2_dense   |                      -128(-100.0%)       
------ -------------------------------------- ---------------------- ------------------- --------- ----------------- --- -------------------- ------------------- ------------------------------- 
6      dense_3_dense (Dense, Dense)           [b:1,c:6]              774/3,096                 774           dense_2   |                      +90(+11.6%)         Dense_/Nonlinearity_[o][5, 6]   
       dense_3 (Nonlinearity, Dense)          [b:1,c:6]                                         90     dense_3_dense   |                      -90(-100.0%)        
------ -------------------------------------- ---------------------- ------------------- --------- ----------------- --- -------------------- ------------------- ------------------------------- 
model/c-model: macc=875,232/875,232  weights=2,955,800/375,448 -2,580,352(-87.3%) activations=--/25,616 io=--/0



Generated C-graph summary
------------------------------------------------------------------------------------------------------------------------
model name            : har_github
c-name                : har_github
c-node #              : 7
c-array #             : 18
activations size      : 25616 (1 segment)
weights size          : 375448 (1 segment)
macc                  : 875232
inputs                : ['input_0_output']
outputs               : ['dense_3_output']

C-Arrays (18)
------ -------------------------- --------------- ------------------------- ------------------ --------- 
c_id   name (*_array)             item/size       domain/mem-pool           c-type             comment   
------ -------------------------- --------------- ------------------------- ------------------ --------- 
0      conv2d_1_conv2d_bias       128/512         weights/weights           const float                  
1      conv2d_1_conv2d_output     5632/22528      activations/**default**   float                        
2      conv2d_1_conv2d_scratch0   4/16            activations/**default**   float                        
3      conv2d_1_conv2d_scratch1   512/2048        activations/**default**   float                        
4      conv2d_1_conv2d_weights    512/2048        weights/weights           const float                  
5      dense_1_dense_bias         128/512         weights/weights           const float                  
6      dense_1_dense_output       128/512         activations/**default**   float                        
7      dense_1_dense_weights      720896/360512   weights/weights           const lut4_float             
8      dense_1_output             128/512         activations/**default**   float                        
9      dense_2_dense_bias         128/512         weights/weights           const float                  
10     dense_2_dense_output       128/512         activations/**default**   float                        
11     dense_2_dense_weights      16384/8256      weights/weights           const lut4_float             
12     dense_2_output             128/512         activations/**default**   float                        
13     dense_3_dense_bias         6/24            weights/weights           const float                  
14     dense_3_dense_output       6/24            activations/**default**   float                        
15     dense_3_dense_weights      768/3072        weights/weights           const float                  
16     dense_3_output             6/24            activations/**default**   float              /output   
17     input_0_output             270/1080        activations/**default**   float              /input    
------ -------------------------- --------------- ------------------------- ------------------ --------- 

C-Layers (7)
------ ----------------- ---- --------------- -------- -------- ----------------------------- --------------------- 
c_id   name (*_layer)    id   layer_type      macc     rom      tensors                       shape (array id)      
------ ----------------- ---- --------------- -------- -------- ----------------------------- --------------------- 
0      conv2d_1_conv2d   1    Conv2D          136576   2560     I: input_0_output             f32(1x90x3x1) (17)    
                                                                S: conv2d_1_conv2d_scratch0                         
                                                                S: conv2d_1_conv2d_scratch1                         
                                                                W: conv2d_1_conv2d_weights    f32(128x2x2x1) (4)    
                                                                W: conv2d_1_conv2d_bias       f32(128) (0)          
                                                                O: conv2d_1_conv2d_output     f32(1x44x1x128) (1)   
------ ----------------- ---- --------------- -------- -------- ----------------------------- --------------------- 
1      dense_1_dense     4    Dense           721024   361024   I: conv2d_1_conv2d_output     f32(1x44x1x128) (1)   
                                                                W: dense_1_dense_weights      c4(128x5632) (7)      
                                                                W: dense_1_dense_bias         f32(128) (5)          
                                                                O: dense_1_dense_output       f32(1x128) (6)        
------ ----------------- ---- --------------- -------- -------- ----------------------------- --------------------- 
2      dense_1           4    Nonlinearity    128      0        I: dense_1_dense_output       f32(1x128) (6)        
                                                                O: dense_1_output             f32(1x128) (8)        
------ ----------------- ---- --------------- -------- -------- ----------------------------- --------------------- 
3      dense_2_dense     5    Dense           16512    8768     I: dense_1_output             f32(1x128) (8)        
                                                                W: dense_2_dense_weights      c4(128x128) (11)      
                                                                W: dense_2_dense_bias         f32(128) (9)          
                                                                O: dense_2_dense_output       f32(1x128) (10)       
------ ----------------- ---- --------------- -------- -------- ----------------------------- --------------------- 
4      dense_2           5    Nonlinearity    128      0        I: dense_2_dense_output       f32(1x128) (10)       
                                                                O: dense_2_output             f32(1x128) (12)       
------ ----------------- ---- --------------- -------- -------- ----------------------------- --------------------- 
5      dense_3_dense     6    Dense           774      3096     I: dense_2_output             f32(1x128) (12)       
                                                                W: dense_3_dense_weights      f32(6x128) (15)       
                                                                W: dense_3_dense_bias         f32(6) (13)           
                                                                O: dense_3_dense_output       f32(1x6) (14)         
------ ----------------- ---- --------------- -------- -------- ----------------------------- --------------------- 
6      dense_3           6    Nonlinearity    90       0        I: dense_3_dense_output       f32(1x6) (14)         
                                                                O: dense_3_output             f32(1x6) (16)         
------ ----------------- ---- --------------- -------- -------- ----------------------------- --------------------- 



Number of operations per c-layer
------- ------ -------------------------- --------- -------------- 
c_id    m_id   name (type)                      #op           type 
------- ------ -------------------------- --------- -------------- 
0       1      conv2d_1_conv2d (Conv2D)     136,576   smul_f32_f32 
1       4      dense_1_dense (Dense)        721,024    smul_f32_f4 
2       4      dense_1 (Nonlinearity)           128     op_f32_f32 
3       5      dense_2_dense (Dense)         16,512    smul_f32_f4 
4       5      dense_2 (Nonlinearity)           128     op_f32_f32 
5       6      dense_3_dense (Dense)            774   smul_f32_f32 
6       6      dense_3 (Nonlinearity)            90     op_f32_f32 
------- ------ -------------------------- --------- -------------- 
total                                       875,232 

Number of operation types
---------------- --------- ----------- 
operation type           #           % 
---------------- --------- ----------- 
smul_f32_f32       137,350       15.7% 
smul_f32_f4        737,536       84.3% 
op_f32_f32             346        0.0% 

Complexity report (model)
------ ----------------- ------------------------- ------------------------- -------- 
m_id   name              c_macc                    c_rom                     c_id     
------ ----------------- ------------------------- ------------------------- -------- 
1      max_pooling2d_1   |||               15.6%   |                  0.7%   [0]      
4      dense_1_dense     ||||||||||||||||  82.4%   ||||||||||||||||  96.2%   [1, 2]   
5      dense_2_dense     |                  1.9%   |                  2.3%   [3, 4]   
6      dense_3_dense     |                  0.1%   |                  0.8%   [5, 6]   
------ ----------------- ------------------------- ------------------------- -------- 
macc=875,232 weights=375,448 act=25,616 ram_io=0
 
 Requested memory size by section - "stm32f4" target
 ------------------------------ -------- --------- ------- -------- 
 module                             text    rodata    data      bss 
 ------------------------------ -------- --------- ------- -------- 
 NetworkRuntime1020_CM4_GCC.a     11,000         0       0        0 
 har_github.o                        648        56   2,364      180 
 har_github_data.o                    48        16      88        0 
 lib (toolchain)*                    614        24       0        0 
 ------------------------------ -------- --------- ------- -------- 
 RT total**                       12,310        96   2,452      180 
 ------------------------------ -------- --------- ------- -------- 
 weights                               0   375,448       0        0 
 activations                           0         0       0   25,616 
 io                                    0         0       0        0 
 ------------------------------ -------- --------- ------- -------- 
 TOTAL                            12,310   375,544   2,452   25,796 
 ------------------------------ -------- --------- ------- -------- 
 *  toolchain objects (libm/libgcc*)
 ** RT AI runtime objects (kernels+infrastructure)
  
  Summary - "stm32f4" target
  -------------------------------------------------
               FLASH (ro)     %*   RAM (rw)      % 
  -------------------------------------------------
  RT total         14,858   3.8%      2,632   9.3% 
  -------------------------------------------------
  TOTAL           390,306            28,248        
  -------------------------------------------------
  *  rt/total


Generated files (7)
------------------------------------------------------------------------- 
C:\Users\Aix Sonic\.stm32cubemx\network_output\har_github_data_params.h   
C:\Users\Aix Sonic\.stm32cubemx\network_output\har_github_data_params.c   
C:\Users\Aix Sonic\.stm32cubemx\network_output\har_github_data.h          
C:\Users\Aix Sonic\.stm32cubemx\network_output\har_github_data.c          
C:\Users\Aix Sonic\.stm32cubemx\network_output\har_github_config.h        
C:\Users\Aix Sonic\.stm32cubemx\network_output\har_github.h               
C:\Users\Aix Sonic\.stm32cubemx\network_output\har_github.c               
