matlab-deploy-embedded-code

Name: matlab-deploy-embedded-code
Author: matlab

815stars105forks

How to install matlab-deploy-embedded-code?

matlab/matlab-agentic-toolkit/matlab-deploy-embedded-code

$npx -y skills add matlab/matlab-agentic-toolkit --skill matlab-deploy-embedded-code

Installs into the current project.

Summary

Deploy MATLAB-generated code to embedded hardware using Embedded Coder. Use when configuring code generation for microcontrollers (STM32, Raspberry Pi, ARM Cortex), setting up PIL/SIL verification, disabling dynamic memory allocation, or configuring hardware-specific code generat

Files · 1

View on GitHub

SKILL.md

1	# Deploy Embedded Code
2
3	Configure MATLAB Coder with Embedded Coder for production-quality code generation
4	targeting embedded hardware, and verify correctness with processor-in-the-loop (PIL)
5	testing.
6
7	## When to Use
8
9	- Generating C/C++ code for a microcontroller or embedded Linux board
10	- Setting up PIL or SIL verification for generated code
11	- Configuring code generation with no dynamic memory allocation
12	- Deploying deep learning models to resource-constrained hardware
13	- Selecting a hardware target board (STM32, Raspberry Pi)
14
15	## When NOT to Use
16
17	- Generating MEX or desktop libraries — use standard `codegen` workflows
18	- Simulink-based deployment — use Simulink Coder / Embedded Coder workflows directly
19	- GPU code generation (CUDA) — use GPU Coder
20
21	## Workflow
22
23	### 1. Create an ERT-Based Configuration
24
25	```matlab
26	cfg = coder.config("lib", "ecoder", true);
27	```
28
29	The `"ecoder", true` flag creates an ERT-based (Embedded Real-Time) configuration
30	that generates production-quality code with no OS dependencies.
31
32	### 2. Select Target Hardware
33
34	With `coder.hardware`:
35
36	```matlab
37	cfg.Hardware = coder.hardware("STM32F746G-Discovery");
38	```
39
40	Without the support package, configure hardware manually:
41
42	```matlab
43	cfg.HardwareImplementation.ProdHWDeviceType = 'ARM Compatible->ARM Cortex-M';
44	cfg.HardwareImplementation.ProdBitPerFloat = 32;
45	cfg.HardwareImplementation.ProdBitPerDouble = 64;
46	```
47
48	See `references/supported-hardware.md` for the full list of supported boards and
49	their constraints.
50
51	### 3. Configure Memory for Bare-Metal Targets
52
53	```matlab
54	cfg.EnableDynamicMemoryAllocation = false;
55	cfg.StackUsageMax = 512;
56	```
57
58	- `EnableDynamicMemoryAllocation = false` — disables `malloc`/`free` for targets
59	where heap is unavailable or non-deterministic. All arrays must be bounded at
60	compile time.
61	- `StackUsageMax` — set based on target SRAM. The code generation report shows
62	actual usage after compilation.
63
64	For entry-points that use deep learning inference (`invoke`, `predict`):
65
66	```matlab
67	cfg.DeepLearningConfig = coder.DeepLearningConfig('none');
68	cfg.LargeConstantGeneration = "KeepInSourceFiles";
69	```
70
71	- `DeepLearningConfig('none')` — generates C with no external DL library dependencies
72	(MKL-DNN, cuDNN, TensorRT). Required for bare-metal targets. Without this, codegen
73	may attempt to link an unavailable library and fail.
74	- `LargeConstantGeneration = "KeepInSourceFiles"` — keeps weight constants in source
75	files rather than separate data files. Needed for bare-metal targets where external
76	data file linking is unsupported.
77
78	### 4. Configure Performance (SIMD and OpenMP)
79
80	SIMD vectorization — generates vectorized code for the target ISA:
81
82	```matlab
83	cfg.InstructionSetExtensions = 'Neon v7'; % ARM Cortex-A (128-bit, 4x float32)
84	```
85
86	\| Target \| Value \| Notes \|
87	\|--------\|-------\|-------\|
88	\| ARM Cortex-A (Raspberry Pi) \| `'Neon v7'` \| 128-bit SIMD \|
89	\| Intel x86-64 \| `'SSE'`, `'SSE4.1'`, `'AVX'`, `'AVX2'`, `'AVX512F'` \| Match target CPU \|
90	\| ARM Cortex-M \| Do not set — use `CodeReplacementLibrary` instead \| Different mechanism \|
91
92	OpenMP multi-threading — enables parallel loops in generated code:
93
94	```matlab
95	cfg.EnableOpenMP = true; % multi-core targets (Cortex-A, x86)
96	cfg.EnableOpenMP = false; % single-core targets (Cortex-M) — no OS/threading support, won't compile
97	```
98
99	### 5. Set Up PIL Verification
100
101	PIL compiles the generated code, deploys it to the physical board, sends test
102	vectors, and compares outputs against MATLAB. This catches precision differences,
103	stack overflows, and memory issues that SIL cannot detect.
104
105	Cortex-M (serial transport):
106
107	```matlab
108	cfg.VerificationMode = "PIL";
109	cfg.Hardware.PILInterface = "Serial";
110	cfg.Hardware.PILCOMPort = "COM4"; % adjust to your system
111	```
112
113	Cortex-A / Raspberry Pi (SSH transport):
114
115	```matlab
116	cfg.VerificationMode = "PIL";
117	cfg.Hardware = coder.hardware("Raspberry Pi");
118	cfg.Hardware.DeviceAddress = "192.168.1.10";
119	cfg.Hardware.Username = "<your-pi-username>";
120	cfg.Hardware.Password = "<your-pi-password>";
121	cfg.Hardware.BuildDir = "/home/pi/mymodel"; % optional: defaults to /home/pi/MATLAB_ws/<release>
122	```
123
124	Pi PIL runs over SSH (not serial). The support package uses `DeviceAddress`,
125	`Username`, and `Password` to establish the SSH connection. `BuildDir` specifies
126	where the compiled binary is deployed on the target; if omitted, defaults to
127	`/home/pi/MATLAB_ws/<release>/`. Do not set `PILInterface` or `PILCOMPort` — th

People who liked this also viewed

6 picks

Type

matlab-deploy-embedded-code

815stars105forks

View on GitHub

How to install matlab-deploy-embedded-code?

matlab/matlab-agentic-toolkit/matlab-deploy-embedded-code

$npx -y skills add matlab/matlab-agentic-toolkit --skill matlab-deploy-embedded-code

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Deploy Embedded Code
2
3	Configure MATLAB Coder with Embedded Coder for production-quality code generation
4	targeting embedded hardware, and verify correctness with processor-in-the-loop (PIL)
5	testing.
6
7	## When to Use
8
9	- Generating C/C++ code for a microcontroller or embedded Linux board
10	- Setting up PIL or SIL verification for generated code
11	- Configuring code generation with no dynamic memory allocation
12	- Deploying deep learning models to resource-constrained hardware
13	- Selecting a hardware target board (STM32, Raspberry Pi)
14
15	## When NOT to Use
16
17	- Generating MEX or desktop libraries — use standard `codegen` workflows
18	- Simulink-based deployment — use Simulink Coder / Embedded Coder workflows directly
19	- GPU code generation (CUDA) — use GPU Coder
20
21	## Workflow
22
23	### 1. Create an ERT-Based Configuration
24
25	```matlab
26	cfg = coder.config("lib", "ecoder", true);
27	```
28
29	The `"ecoder", true` flag creates an ERT-based (Embedded Real-Time) configuration
30	that generates production-quality code with no OS dependencies.
31
32	### 2. Select Target Hardware
33
34	With `coder.hardware`:
35
36	```matlab
37	cfg.Hardware = coder.hardware("STM32F746G-Discovery");
38	```
39
40	Without the support package, configure hardware manually:
41
42	```matlab
43	cfg.HardwareImplementation.ProdHWDeviceType = 'ARM Compatible->ARM Cortex-M';
44	cfg.HardwareImplementation.ProdBitPerFloat = 32;
45	cfg.HardwareImplementation.ProdBitPerDouble = 64;
46	```
47
48	See `references/supported-hardware.md` for the full list of supported boards and
49	their constraints.
50
51	### 3. Configure Memory for Bare-Metal Targets
52
53	```matlab
54	cfg.EnableDynamicMemoryAllocation = false;
55	cfg.StackUsageMax = 512;
56	```
57
58	- `EnableDynamicMemoryAllocation = false` — disables `malloc`/`free` for targets
59	where heap is unavailable or non-deterministic. All arrays must be bounded at
60	compile time.
61	- `StackUsageMax` — set based on target SRAM. The code generation report shows
62	actual usage after compilation.
63
64	For entry-points that use deep learning inference (`invoke`, `predict`):
65
66	```matlab
67	cfg.DeepLearningConfig = coder.DeepLearningConfig('none');
68	cfg.LargeConstantGeneration = "KeepInSourceFiles";
69	```
70
71	- `DeepLearningConfig('none')` — generates C with no external DL library dependencies
72	(MKL-DNN, cuDNN, TensorRT). Required for bare-metal targets. Without this, codegen
73	may attempt to link an unavailable library and fail.
74	- `LargeConstantGeneration = "KeepInSourceFiles"` — keeps weight constants in source
75	files rather than separate data files. Needed for bare-metal targets where external
76	data file linking is unsupported.
77
78	### 4. Configure Performance (SIMD and OpenMP)
79
80	SIMD vectorization — generates vectorized code for the target ISA:
81
82	```matlab
83	cfg.InstructionSetExtensions = 'Neon v7'; % ARM Cortex-A (128-bit, 4x float32)
84	```
85
86	\| Target \| Value \| Notes \|
87	\|--------\|-------\|-------\|
88	\| ARM Cortex-A (Raspberry Pi) \| `'Neon v7'` \| 128-bit SIMD \|
89	\| Intel x86-64 \| `'SSE'`, `'SSE4.1'`, `'AVX'`, `'AVX2'`, `'AVX512F'` \| Match target CPU \|
90	\| ARM Cortex-M \| Do not set — use `CodeReplacementLibrary` instead \| Different mechanism \|
91
92	OpenMP multi-threading — enables parallel loops in generated code:
93
94	```matlab
95	cfg.EnableOpenMP = true; % multi-core targets (Cortex-A, x86)
96	cfg.EnableOpenMP = false; % single-core targets (Cortex-M) — no OS/threading support, won't compile
97	```
98
99	### 5. Set Up PIL Verification
100
101	PIL compiles the generated code, deploys it to the physical board, sends test
102	vectors, and compares outputs against MATLAB. This catches precision differences,
103	stack overflows, and memory issues that SIL cannot detect.
104
105	Cortex-M (serial transport):
106
107	```matlab
108	cfg.VerificationMode = "PIL";
109	cfg.Hardware.PILInterface = "Serial";
110	cfg.Hardware.PILCOMPort = "COM4"; % adjust to your system
111	```
112
113	Cortex-A / Raspberry Pi (SSH transport):
114
115	```matlab
116	cfg.VerificationMode = "PIL";
117	cfg.Hardware = coder.hardware("Raspberry Pi");
118	cfg.Hardware.DeviceAddress = "192.168.1.10";
119	cfg.Hardware.Username = "<your-pi-username>";
120	cfg.Hardware.Password = "<your-pi-password>";
121	cfg.Hardware.BuildDir = "/home/pi/mymodel"; % optional: defaults to /home/pi/MATLAB_ws/<release>
122	```
123
124	Pi PIL runs over SSH (not serial). The support package uses `DeviceAddress`,
125	`Username`, and `Password` to establish the SSH connection. `BuildDir` specifies
126	where the compiled binary is deployed on the target; if omitted, defaults to
127	`/home/pi/MATLAB_ws/<release>/`. Do not set `PILInterface` or `PILCOMPort` — th

People who liked this also viewed

6 picks

Type