At Arm it’s always a challenge topping previous year’s premium IP. In 2018, we launched the Arm Mali-G76 GPU which, at the time, boasted our highest ever GPU performance. Already Mali-G76 has seen great success, delivering on the performance and energy efficiency numbers, see our blog from this time last year, and being the leading GPU for the latest content on Premium mobile devices. Our partners have also incorporated Mali-G76 into their leading premium mobile devices, including the Samsung Galaxy S10.
Despite all this success with Mali-G76, we’ve yet again managed to boost performance and energy efficiency levels with our new Arm Mali-G77 GPU, which is our first premium GPU based on the brand new Valhall architecture. After three years with Bifrost, we thought it would be the perfect opportunity to introduce the new Valhall architecture – a new architecture for the next generation IP.
Before delving more deeply into the features of the new Valhall architecture, it’s worth summarizing the improvements that Mali-G77 brings to the table. Consumers are doing far more with their mobile devices ‘on-the-go’, such as high-performance mobile gaming, augmented reality (AR) and tasks with complex machine learning (ML) workloads. They therefore need high performing and efficient mobile devices that can manage these compute-intensive tasks while not impacting battery life.
Mali-G77 aims to tackle these challenges head-on through a number of improvements from Mali-G76. The first Valhall GPU delivers a 30 percent increase in performance density, 30 percent energy efficiency improvements, and 60 percent improvement for ML. All of this means that we expect Mali-G77-based devices to deliver a 40 percent better peak graphics performance when it arrives in mobile devices.
As with Mali-G76, the new Mali-G77 helps to bring more high-end gaming to mobile devices. According to the market intelligence agency Newzoo, 2018 was the year when revenues from mobile gaming overtook revenues on console and PC-based gaming for the very first time, contributing to more than half of gaming revenues worldwide ($70.3 billion). Current mobile gaming trends show that users are moving towards very popular high-fidelity, multi-user games, such as Fortnite and PUBG. Mali-G77 caters to the mobile gaming evolution through helping to deliver high-end graphics for mobile gaming and mobile devices that have a longer battery life. This means users can play their favorite mobile games for far longer.
More consumers are looking for new experiences on their mobile devices, with AR, in particular, being one such experience. AR is increasingly being adopted into mobile devices, allowing users to transform their surroundings, themselves and their interactions between physical and virtual worlds. The recent Arm-commissioned report from the market intelligence agency Newzoo highlighted the ‘boundless potential’ of AR in mobile gaming. The high-end graphics support of Mali-G77 provides mobile devices with the capabilities to increasingly adopt AR features that will transform mobile gaming and other mobile user experiences.
ML has been happening for some time on devices for a variety of everyday tasks, from predictive text to face unlock. However, as our devices become smarter, there may be a need for additional compute to create an intelligent, personalized and secure premium experience. Mali-G77 provides devices with the capabilities to perform increasingly complex ML tasks faster on the device with a 60 percent performance density improvement. Being able to perform tasks on the device rather than sending them to the cloud for processing means improved performance, less latency and fewer security concerns.
The Valhall architecture is the basis of Mali-G77 and future Mali GPUs. This makes all the high-end and complex use cases that I outlined above possible on mobile devices. The following features make Valhall such a novel architecture:
While there are many different advancements and new features, the two key ones are the execution engine and texture mapper in Mali-G77.
The wide execution engines of Mali-G77 improve performance density through sharing control over a wider number of lanes. Looking at Mali-G76, this has 8-wide warps and a total of 24 FMA lanes per shader core. However, Mali-G77 has 16 wide-warps, 32 lanes (two clusters of 16 FMA per execution engine) and one engine per shader core. This means 33 percent more compute in the same area when compared to Mali-G76.
The improved gaming performance of Mali-G77 is linked to the quad texture mapper, which provides four texels/ cycle. This is 2x greater throughput than Mali-G76 and 4x greater than Mali-G72. The quad texture mapper provides improvements across the board of high-fidelity and casual gaming, but has an especially large impact on texture heavy games. In addition, because we increased the compute capability in Mali-G77, we also need to increase the texture capability to keep the machine balanced. This enables Mali-G77 to deliver more performance per square millimetre than ever before.
Furthermore, Mali-G77 has been optimized to match the new 16-wide execution engines and quad texture mapper. This includes a re-design of the LSC and attribute pipe with a focus on performance density and energy efficiency.
Arm has a significant focus on improving energy efficiency which impacts all areas of the design. This means that Mali-G77 can do the same work in 50 percent of the energy of Mali-G72 from only two years ago. The Valhall architecture and Mali-G77 boost energy efficiency across all workloads, leading to an improvement of 1.3x across a wide range of content. This delivers longer battery life for premium products.
Although Bifrost remains a good fit for current content, the ever-evolving complex nature of content in the future is likely to require a new approach. Dynamic instruction scheduling is now handled in hardware, enabling better performance. The dynamic scheduler in the Valhall architecture decides which instructions to execute from which warps. The work is then issued to independent parallel ALUs in superscalar style.
Finally, the Valhall architecture continues the evolution of the Arm Frame Buffer Compression through AFBC 1.3. The new features in Valhall through AFBC 1.3. include support for 2-plane YUV, improved support for front-buffer rendering, support for separate depth/ stencil encoding (a better match for Vulkan) and opacity/ transparency hints.
The new Valhall architecture is at the heart of the key performance improvements of Mali-G77. The uncompromising performance and energy efficiency of the GPU is enabled by the 16 wide warps and two clusters of 16 FMA per execution engine per core, which leads to the 33 percent increase in compute performance compared to Mali-G76. Meanwhile, mobile gaming experiences are enhanced with up to 16 shader cores and the quad texture mapper.
Mali-G77 is yet another step change in Arm’s high-performance GPU roadmap, providing uncompromising graphics performance and increased efficiency. It also brings significant performance improvements to complex AR and ML, which will help to drive future use cases. We look forward to the seeing the development of these use cases on mobile devices, which will be made possible by Mali-G77!
Read our newsroom blog about the launch of the Premium IP suite on Arm.com.
Learn about Mali-G77