Creator: STMicroelectronics

ST is asserting at this time the NeoChromVG GPU, the primary vector graphics accelerator for general-purpose microcontrollers. Out there on the upcoming STM32U5G9 and STM32U5F9, it builds on the Neochrom GPU, launched in 2023, and takes options like vector fonts, accessible on TouchGFX 4.23, to the subsequent degree by enabling a complete new kind of optimization on embedded techniques. Avid readers of the ST Weblog will discover how NeoChromVG GPU is the fruits of our push into vector graphics within the hope of serving to makers scale back their reminiscence footprint. We even present growth kits and demos to assist engineers know if this resolution is true for his or her challenge.

Desk of Contents

What’s new? NeoChromVG GPU!

The problem of making UIs with much less reminiscence

The explanation ST determined to launch a selected {hardware} IP accelerating vector operations is as a result of it was an more and more well-liked demand from prospects trying to scale back their dependency on exterior flash modules. Reminiscence markets have all the time been unstable, however many analysts, in line with Forbes, anticipate sturdy demand in flash throughout 2024, bolstered partially by new AI functions attempting to create “all-flash object storage techniques”. Consequently, smaller firms with much less buying energy may encounter severe points sourcing modules or just going through costs too excessive to make their merchandise viable. Therefore, many wish to both scale back their exterior NAND wants or obliterate them altogether, which led us to supply NeoChromVG GPU in new STM32U5F9 and STM32U5G9 MCUs.

The NeoChromVG GPU resolution

In essence, NeoChromVG GPU is a conventional Neochrom GPU with a {hardware} unit accelerating vector graphics computations, corresponding to quadratic Bézier curves, that are quite common in vector fonts, Bézier tessellation to render surfaces, and vector calculations for vertex transformation, amongst different issues, to quickly draw vector graphics on display. It nonetheless depends on the identical texture mapper because the common Neochrom GPU and has related options like L8 compression. Vector graphic acceleration merely turns into a further instrument in a developer’s arsenal to optimize their UI. The function additionally comes on the heels of TouchGFX 4.23, which simply gained help for vector fonts. Therefore, engineers see the overarching technique ST is utilizing to assist with reminiscence dependency.

NeoChromVG GPU might be accessible on the brand new STM32U5G9 and STM32U5F9, the previous being the model housing a crypto core. Each will function 3 MB of RAM and 4 MB of dual-bank flash. This {hardware} configuration has enabled a few of our prospects to maneuver to a single-chip design that didn’t require exterior reminiscence in any respect, whereas the rise in inside reminiscence and vector graphics help helped others use smaller NAND gadgets. In each circumstances, it helps simplify the PCB design to maneuver to solely 4 layers and decrease the invoice of supplies. Equally, as embedded techniques are likely to map their NOR reminiscence for higher optimization, not having to map something above 32 MB turns into a big time saver.

One of the simplest ways to start out utilizing vector graphics on STM32

Vector graphics aren’t for each system. Even with {hardware} acceleration, the efficiency penalty incurred by vector rendering could also be an excessive amount of in sure UIs. Nonetheless, we’re making certain that builders get speedy prototyping instruments to check NeoChromVG GPU in minutes. We’re releasing the STM32U5G9J-DK, with a 480 x 480 show, and the STM32U5G9J-DK2 with an 800 x 480 display, in addition to a UI demo exhibiting what builders can accomplish because of vector graphics. Even the ST Companion Riverdi, well-known for its STM32 Embedded Shows, will quickly supply a mannequin with the STM32U5G9 to assist develop and make the most of the vector graphics accelerator.

From Chrom-ART to NeoChrom GPU

What Are Chrom-ART and NeoChrom GPU?

Readers conversant in ST applied sciences can guess that NeoChrom GPU is the subsequent evolution of Chrom-ART. The latter is a 2D accelerator (DMA2D) embedded inside a number of STM32 MCUs that optimizes reminiscence entry to govern photos sooner. Many instruments and merchandise from ST companions use it, corresponding to Storyboard from Crank, Microvisor from Twilio, or interfaces from PsiControl, a TouchGFX Implementer, to solely identify three. Comparatively, NeoChrom GPU attracts from the identical philosophy whereas providing a higher scope of optimizations. Due to texture mapping with bilinear and level sampling, the brand new know-how offers new drawing acceleration and helps animations. Therefore, what used to require a Desktop PC GPU card can now run on a microcontroller.

Evaluating specs between Chrom-ART and NeoChrom

How Does NeoChrom GPU Combine Into the TouchGFX Ecosystem?

Traditionally, Chrom-ART represented one in every of ST’s first forays into graphical optimizations. Since then, we acquired TouchGFX and constructed a whole ecosystem with TouchGFX Designer, Simulator, and Generator. In consequence, ST is integrating NeoChrom GPU into these instruments proper from the beginning. TouchGFX Designer accommodates demos exhibiting a easy animation working at 60 FPS on a round show. Extra importantly, it reveals the affect on the body price when NeoChrom GPU is off to assist decision-makers grasp its significance. Equally, the feel mapping widget will robotically use the brand new GPU if the MCU helps it. Therefore, the rapid integration of the brand new IP inside TouchGFX serves as a witness to ST’s cohesive ecosystem as an alternative of a easy function checklist.

From NeoChrom GPU to Actual-World Efficiency

What Are the Challenges to Graphical Performances?

The STM32U5

There exist two major challenges to graphical performances: an instructions-per-clock limitation and a reminiscence challenge. The primary one typically follows architectural improvements. Microcontrollers have ALU and FPU restrictions as a consequence of their measurement and low energy consumption, however enhancements in manufacturing processes permit for extra transistors able to dealing with extra operations. NeoChrom GPU follows such tendencies with a {hardware} IP accelerating extra sorts of graphical calculations. The second problem is trickier however all of the extra important as a result of reminiscence will all the time stay a scarce useful resource. Whereas the STM32U599 will include 2.5 MB of inside RAM able to storing two body buffers, many embedded techniques want exterior RAM. The issue is that it comes with vital efficiency penalties.

What Are Distinctive STM32 Implementations?

To unravel this second problem, NeoChrom GPU comes with {hardware} optimizations particular to STM32 MCUs. As an example, the STM32U5 depends on both an 8-bit Octal Serial Peripheral Interface (OSPI) or a 16-bit high-speed interface (HSPI) at 200 MHz to exterior reminiscence modules to hasten fetching operations to the L2 cache. Equally, the L1 cache can immediately entry two 832 KB body buffers within the inside RAM. In consequence, tessellation is considerably sooner because of fast entry to all the information essential to render the picture in real-time. Put merely, our engineers optimized efficiency by engaged on decreasing the time it takes to entry textures and different property, finally accelerating rendering by as much as six instances.

What’s Subsequent?

The primary business suggestions round NeoChrom GPU has been very constructive. Whereas many MCU makers have related IPs to speed up graphical computations, the particular STM32 optimizations and the TouchGFX ecosystem led many early alpha testers to undertake NeoChrom GPU as an alternative of competing options. The brand new know-how is already supported inside TouchGFX 4.19, and some alpha prospects have already labored on production-ready UIs utilizing an upcoming STM32U599 growth equipment. ST can even embody NeoChrom GPU in different upcoming microcontrollers and we’ll replace this weblog put up as they change into public.