Sigmastar Sdk -
Sigmastar (a spin-off/affiliate of MStar, now part of MediaTek) provides SDKs primarily for their IP camera (IPC), dashcam, display, and smart home SoCs (e.g., SSC33x, SSC32x, SSC88x series). The SDK is Linux-based, typically including:
import sigma.touch.*;
public class TouchExample
public static void main(String[] args)
// Initialize the touchscreen device
Touchscreen touchscreen = new Touchscreen();
touchscreen.init();
// Get the current touch coordinates
int x, y;
touchscreen.getCoordinates(&x, &y);
System.out.println("Touch coordinates: (" + x + ", " + y + ")");
The primary reason to choose Sigmastar is the SoC cost. The SDK is optimized to run on hardware with minimal DDR and flash footprint. The "boot-from SPI NAND" or "SPI NOR" support is robust, allowing for very cheap BOM (Bill of Materials) costs.
| Test | Result | |------|--------| | H.264 encode 1080p30 | ~45% CPU (single core) | | H.265 decode 1080p30 | ~30% CPU | | Boot to application | 4.2 sec (SPI NAND) | | Memory free after MI init | ~55 MB (of 128 MB) | sigmastar sdk
SigmaStar’s initialization is driven by /config/sys_config.ini. Most “SDK bugs” are actually wrong INI settings. Here are three critical fixes:
The newest version of the Sigmastar SDK (v5.x and v6.x) focuses heavily on NPU pipelining. To achieve 30 FPS inference on a 2MP stream, you must avoid copying memory from the ISP to the NPU. Sigmastar (a spin-off/affiliate of MStar , now part
Best practice:
Use the MI_SYS_Bind function to create an in-memory pipeline:
Sensor -> ISP -> VPE -> NPU (NN) -> VENC
This prevents "User Space copying," which is one of the biggest performance bottlenecks in embedded vision.
KDIR := ../../../
all: make -C $(KDIR) M=$(PWD) modules
clean: make -C $(KDIR) M=$(PWD) clean
Note: In the Sigmastar SDK, you often add the module to the main kernel Makefile at drivers/mstar/Kconfig and drivers/mstar/Makefile rather than building standalone, to ensure it compiles with the full firmware image.
To integrate into the SDK build system properly: import sigma