What Is Hardware Acceleration in Embedded GUI?

Hardware acceleration in embedded GUI means offloading pixel-level graphics operations — fill, blit, alpha blend, rotate, and scale — from the main CPU to a dedicated hardware unit on the MCU or MPU. The CPU issues commands to the accelerator, which operates autonomously on the framebuffer while the CPU continues executing application code. The result is higher frame rates, lower CPU utilisation, and smoother animations on the same silicon — without moving to a faster or more expensive processor.

Sparklet's GDI (Graphics Device Interface) layer abstracts all hardware accelerators behind a uniform API. Widget code never contains hardware-specific calls. The platform-specific acceleration is implemented inside the GDI HAL layer, which Sparklet ships pre-ported for each supported platform. When Sparklet detects a hardware accelerator, it automatically delegates pixel operations to it; on platforms without a hardware unit, the same GDI API falls back to Sparklet's optimised software renderer — no code changes required in the application or widget layer.

Sparklet currently supports four hardware acceleration units: DMA2D (STM32 Chrom-ART), D/AVE2D (Renesas RH850 and RA series), Mali-Limav GPU (Renesas RA8D1 and NXP i.MX 8), and PXP (NXP i.MX RT series). Each is detailed below. For platform-specific frame-rate data, visit the Performance and Memory Footprint page.

Hardware Accelerator Support by Platform

DMA2D — STM32 Chrom-ART

Alpha blend, fill, format convert, and blit in hardware on STM32 F4, H7, and U5 with zero CPU involvement. DMA2D (Chrom-ART) is STMicroelectronics' dedicated 2D DMA unit that understands pixel formats and compositing operations natively. Sparklet's STM32 GDI HAL configures DMA2D for four operation types: memory fill (background clear), pixel format conversion (ARGB8888 to RGB565 on-the-fly during blit), alpha blending (widget overlay and dialog fade), and memory-to-display copy at full DMA bandwidth. On STM32H7 at 480 MHz, DMA2D reduces average CPU load from 60–80% to under 20% for a typical 480×272 RGB565 UI with animated transitions. The CPU submits a DMA2D descriptor and immediately proceeds — pixel operations execute asynchronously.

D/AVE2D — Renesas RH850 & RA

D/AVE2D (Dave2D) is a full programmable 2D vector and raster GPU integrated into Renesas RH850 automotive MCUs and several RA series devices. Unlike DMA2D, D/AVE2D executes drawing commands rather than simple DMA transfers — making it capable of anti-aliased vector shapes. Sparklet drives D/AVE2D for: anti-aliased line, arc, and polygon rendering (smooth gauge needles and chart lines); gradient fills (linear and radial backgrounds without pre-baked image assets); hardware-accelerated glyph rendering with sub-pixel anti-aliasing; and texture mapping with affine transformation (scale and rotate). On an RH850 D1M2, a 60fps 1280×480 dual-gauge cluster display is achievable with CPU utilisation below 30%. Sparklet abstracts D/AVE2D completely — the application calls standard widget APIs; the GDI HAL translates to D/AVE2D display list commands. See Sparklet for Renesas for platform details.

Mali GPU & PXP — RA8D1 / i.MX

The Mali-Limav GPU (Renesas RA8D1 and NXP i.MX 8) provides a full OpenGL ES 2.0 pipeline — GPU-composited multi-layer 2D GUI, hardware-accelerated 3D rendering, and high-bandwidth texture operations at 60fps ARGB8888 on MCU-class silicon. Sparklet uses Mali in two modes: GPU-accelerated 2D compositing (widget layers as OpenGL ES draw calls, GPU writes result to the display framebuffer) and full 3D mode for 3DWidget and 3DView components. The PXP (Pixel Pipeline) on NXP i.MX RT1170 is a 2D graphics accelerator optimised for the i.MX RT framebuffer architecture. PXP delivers: hardware scale and rotate, pixel format conversion (RGB565, ARGB8888, YCbCr — critical for camera feed integration), alpha blending, and Porter-Duff compositing modes. On i.MX RT1170, PXP enables 60fps at 1280×720 RGB565 with CPU utilisation under 25%. See Sparklet for NXP for details.

Sparklet GDI — One API, All Accelerators

Sparklet's architecture ensures that hardware acceleration is completely transparent to the application layer. This is the property that lets the same UI code run unchanged on a Windows PC simulator, an STM32H7 with DMA2D, and an NXP i.MX RT1170 with PXP — with the appropriate accelerator selected at build time, not hardcoded in application logic.

The dispatch stack:

APP layer: Application code calls Sparklet widget APIs (set_text, set_image, set_position). No knowledge of hardware required.
WID layer (Widgets): Widget rendering logic calls standard GDI draw functions — draw_rect, draw_image, draw_text, blend_layers.
GDI layer: Receives draw calls and dispatches to the platform HAL. On STM32: DMA2D driver. On Renesas RH850/RA: D/AVE2D driver. On RA8D1 and i.MX 8: Mali OpenGL ES driver. On i.MX RT: PXP driver. On platforms without a hardware unit: optimised software renderer.
HAL layer: Platform-specific driver code that configures and drives the hardware accelerator. Embien supplies pre-ported HAL implementations for all supported platforms.

This architecture means a UI developed on the Windows PC simulator (software renderer) runs without application code changes on any hardware target. When porting to a new platform, only the HAL layer changes — the widget tree, screen layouts, state machines, and application logic are identical. This is why teams using Sparklet report significant reductions in hardware bring-up time: the UI is fully validated on the simulator before the target hardware is even available.

For a complete view of Sparklet's embedded GUI feature set across all rendering modes.

Hardware Accelerator Comparison by Platform

Platform	Accelerator	Type	Key Operations	Rendering Mode
STM32 F4 / H7 / U5	DMA2D (Chrom-ART)	2D DMA Unit	Fill, blit, alpha blend, format convert	2D Software + DMA offload
Renesas RH850	D/AVE2D GPU	2D Vector GPU	Anti-aliased vector, gradient fill, glyph render	2D Hardware-Accelerated
Renesas RA6M3 / RA4 series	D/AVE2D GPU	2D Vector GPU	Anti-aliased shapes, blit, texture map	2D Hardware-Accelerated
Renesas RA8D1	Mali-Limav GPU	OpenGL ES 2.0 GPU	GPU-composited 2D, full 3D widget rendering	2.5D / 3D GPU
NXP i.MX RT1170	PXP	2D Pixel Pipeline	Scale, rotate, format convert, alpha blend, Duff	2D Hardware-Accelerated
NXP i.MX 8	Mali GPU	OpenGL ES / Vulkan	Full 3D, GPU-composited layers, shader pipeline	3D High-Performance
STM32F4 / no GPU	None	Software Renderer	Optimised software fill, blit, blend (fallback)	2D Software

Why Hardware Acceleration Matters for Your Product

Four ways Sparklet's hardware acceleration integration directly improves your embedded GUI development outcome.

Smooth 60fps on Cost-Effective MCUs

DMA2D on STM32H7, D/AVE2D on RH850, PXP on i.MX RT — hardware acceleration delivers 60fps at production-grade resolutions without moving to a more expensive processor tier.

CPU Freed for Application Logic

Pixel operations execute asynchronously on the hardware unit. Your CPU continues executing RTOS tasks, CAN/Ethernet communication, sensor processing, and safety monitors — in parallel with GUI rendering.

Automatic Accelerator Selection

The GDI layer detects the available accelerator at build time and activates it automatically. No conditional compilation in application code. Swap MCU families, swap the HAL — widget code is unchanged.

Software Fallback on Any Platform

No accelerator on your custom hardware? Sparklet's optimised software renderer runs on any 32-bit MCU with no HAL configuration. Your UI works from day one — acceleration is additive, not a dependency.

Frequently Asked Questions

How does hardware acceleration work in embedded GUI?

In an embedded GUI with hardware acceleration, the CPU issues high-level draw commands (fill rectangle, blit image, blend layers). A dedicated hardware unit — such as DMA2D on STM32 or D/AVE2D on Renesas — executes those pixel operations autonomously on the framebuffer memory while the CPU continues executing other tasks. The result is more frames per second with lower CPU utilisation than a purely software-rendered GUI.

Which embedded GUI libraries support DMA2D?

How much faster is hardware-accelerated embedded GUI?

Does Sparklet fall back to software rendering if no hardware accelerator is present?

What hardware accelerators does Sparklet support?

Can Sparklet use PXP for camera video feed integration on i.MX RT?

Hardware Acceleration for Embedded GUI