ARM’s new mobile GPU will bring hardware ray tracing to smartphone gaming


Not everyone is aware of the complex technical lineage involved in the parts nestled inside their smartphone, even if they are familiar with names like Qualcomm or MediaTek and the chips those companies make. Basically, all mainstream Android devices use a processor design and architecture created by a company called ARM, tweaking and customizing bits as they see fit. Today ARM announced its new processors and GPUs, which over the next couple of years will trickle down to actual hardware released by chipset makers and eventually end up in a phone you see online, on a store shelf or maybe in your pocket. And by far the biggest announcement is ARM’s new GPU, which supports hardware ray tracing – a feature mostly limited to recent game consoles and high-end desktop GPUs, although Samsung has beaten ARM. .


Immortalis-G715 and new GPUs

The new GPU, Immortalis-G715, is based on the same Mali architecture that many ARM GPUs use. This is the part where some people groan because Malian designs are sometimes considered to be a few steps behind in performance compared to other companies in the industry, but ARM argues that there are big performance gains to get here.

In a purely functional sense, the Immortalis supports hardware-based ray tracing (compatible with the Vulkan API), variable rate shading, support for matrix multiplication instructions, and support and increased development tools. Those numbers aren’t particularly helpful, but ARM claims a 15% performance improvement over previous solutions and a 2x increase in machine learning performance. The company claims it’s also 15% more energy efficient, which could equate to reduced power consumption or operating temperatures.

Ray tracing means more sophisticated and better reflections and lighting in rendered scenes, and it’s worth pointing out here that ARM isn’t the first to announce a mobile chipset with this feature. The Samsung Exynos 2200 revealed in January has an AMD-based GPU that also claims to offer hardware-accelerated ray tracing.

Many of the details presented by ARM were then cautioned that the details and benefits will vary depending on the implementation chosen by the chipset partners – i.e. ARM has these designs, but that they can evolve upwards or downwards according to the requirements of the manufacturer. Ultimately, they can use and modify it within those limits as they see fit, adjusting things like the number of cores or other metrics to suit their design. The G715 can be configured with 10 or more cores (we weren’t given an upper limit).

A ray tracing demo provided by ARM.

ARM hasn’t directly mentioned any partners that will use the technology, but MediaTek praised the company for its announcement, which could be tops. MediaTek has also used Mali GPUs in its products before.

ARM had fewer details to share regarding smaller GPUs, but the Immortalis has a pair of younger siblings it shares its architecture with. The G715 replaces last year’s G710 and inherits some of the features of the G715-Immortalis, such as variable rate shading. Based on the shared nomenclature, we have to assume that the two designs share a lot of guts, but the G715 won’t have hardware-based ray tracing. It will be available in 7-9 core designs.

The slightly lower-end G615 replaces last year’s G610 and offers the same features but comes in designs with 6 cores or less.

Cortex-X3 and new (and old) processors

The new Cortex-X3 is the easiest core to get excited about as it is the largest and latest high performance ARMv9 processor design. ARM again claims double-digit IPC improvements. The company doesn’t even mince words, bluntly stating that it targets things like benchmarks, but can deliver high performance for other workloads as well. Cumulatively, this means we have a big and fast new core to look forward to in the most high-end flagship chipsets.

Technical improvements include 50% larger L1/L2 branch target buffer capacity and 10x larger L0 capacity.

Even outside of smartphones, the X3 can be used in configurations of up to 12 cores for things like laptops or desktops, and ARM claims the new design delivers 34% improved single-threaded performance over to the latest consumer laptops.

While many of us have been waiting for Qualcomm’s Nuvia plans for ARM desktop/laptop computing outside of Apple to finally take off, it sounds as ARM’s generic designs come to a similar point. We’ll have to wait and see how they end up being used.

The Cortex-A715 offers improvements over the A710 in things like branch prediction accuracy and data prefetch, while still offering “constant IPC gains”. In fact, ARM claims the A715 matches the performance of the Cortex-X1, the first of ARM’s still relatively new “super-big” cores that landed in 2020 as part of its custom CPU program. Compared to the older A710, it is 20% more energy efficient when operating at the same performance level and 5% faster at peak.

There really isn’t a new small ARMv9 kernel this year. Instead, ARM increased the efficiency of last year’s Cortex-A510 by slightly improving its efficiency at the same performance level, saving about 5% power, according to ARM. The entire second-generation ARMv9 line also includes a handful of security architecture perks – acronyms for specific Linux kernel features and other tweaks that you won’t understand, but that mean cores have slightly stronger security.

It’s all part of ARM’s “Total Compute Solution 2022” it’s what it calls its relatively recent end-to-end designs for chipsets – think of it as something closer to an instant fix that a chipset maker can rely on for GPU, CPU, machine learning, and pretty much everything. It has designs ranging from high-end phones and laptops, mid-range phones, Chromebooks and XRs, to budget devices, AR and even set-top boxes. Here, ARM is making further improvements to its machine learning solutions (claimed double-digit improvement in some tasks such as object detection and classification) and further tightening security.

ARM also pointed out in the Q&A portion of a recent briefing that its second-generation ARMv9 solutions are really meant to push 64-bit home and encourage the end of 32-bit support, mocking the apparent insistence. from the Chinese market that 32-bit legacy support will be maintained, which has held back the advancement of the hardware to some extent. Note also that A510 still has optional 32-bit support. If ARM puts its foot down, we could see 32-bit support end really, really soon.

The company’s future hardware roadmap was also shown, with names such as Titan and Krake corresponding to future GPUs, the “CXC23” and “CXC24” designs planned as successors to the Cortex X series (paired with a new cluster Hayden DSU), and a new Hayes small/efficient core design planned for 2023 and 2024. Other seemingly more fundamental architectural changes also appear to be planned for 2024, with what appear to be interconnect changes to come.

Ultimately, this all means that ARM has new designs – for CPU cores, GPUs, and even complete chipsets – that you may soon see land in devices. ARM has hinted that at least one vendor using the Cortex X3 will be announced later this year, with actual hardware expected to land early next year.


About Author

Comments are closed.