The Detailed Explanation of GPU Turbo

Under the hood, Huawei uses TensorFlow neural network models that are pre-trained by the company on a title-by-title basis. By examining the title in detail, over many thousands of hours (real or simulated), the neural network can build its own internal model of how the game runs and its power/performance requirements. The end result can be put into one dense sentence:

Optimized Per-Device Per-Game DVFS Control using Neural Networks

In the training phase, the network analyzes and adjusts the SoC’s DVFS parameters in order to achieve the best possible performance while minimizing power consumption. This entails trying its best to hit the nearest DVFS states on the CPUs, GPU, and memory controllers that still allow for hitting 60fps, yet without going to any higher state than is necessary (in other words, minimizing performance headroom). The end result is that for every unit of work that the CPU/GPU/DRAM has to do or manage, the corresponding hardware block has the perfectly optimized amount of power needed. This has a knock-on effect for both performance and power consumption, but mostly in the latter.

The resulting model is then included in the firmware for devices that support GPU Turbo. Each title has a specific network model for each smartphone, as the workload varies with the title and the resources available vary with the phone model. As far as we understand the technology, on the device itself there appears to be an interception layer between the application and GPU driver which monitors render calls. These serve as inputs to the neural network model.  Because the network model was trained to output the DVFS settings that would be most optimal for a given scene, the GPU Turbo mechanism can apply this immediately to the hardware and adjust the DVFS accordingly.

For SoCs that have them, the inferencing (execution) of the network model is accelerated by the SoC’s own NPU. Where GPU Turbo is introduced in SoCs that don’t sport an NPU, a CPU software fall-back is used. This allows for extremely fast prediction. One thing that I do have to wonder is just how much rendering latency this induces, however it can’t be that much and Huawei says they focus a lot on this area of the implementation. Huawei confirmed that these models are all 16-bit floating point (FP16), which means that for future devices like the Kirin 980, further optimization might occur through using INT8 models based on the new NPU support.

Essentially, because GPU Turbo is in effect a DVFS mechanism that works in conjunction with the rendering pipeline and with a much finer granularity, it’s able to predict the hardware requirements for the coming frame and adjust accordingly. This is how GPU Turbo in particular is able to make claims of much reduced performance jitter versus more conventional "reactive" DVFS drivers, which just monitor GPU utilization rate via hardware counters and adapt after-the-fact.

Thoughts After A More Detailed Explanation

What Huawei has done here is certainly an interesting approach with the clear potential for real-world benefits. We can see how distributing resources optimally across available hardware within a limited power budget will help the performance, the efficiency, and the power consumption, all of which is already a careful balancing act in smartphones. So the detailed explanation makes a lot of technical sense, and we have no issues with this at all. It’s a very impressive feat that could have ramifications in a much wider technology space, eventually including PCs.

The downside to the technology is the per-device & per-game nature of it. Huawei did not go into detail about long it took to train a single game: the first version of GPU Turbo supports PUBG and a Chinese game called Mobile Legends: Bang Bang. The second version, coming with the Mate 20, includes NBA 2K18, Rules of Survival, Arena of Valor, and Vainglory.

Technically the granularity is per-SoC rather than per-device, although different devices will have different limits in thermal performance or memory performance. But it is obvious that while Huawei is very proud of the technology, it is a slow per-game roll out. There is no silver bullet here – while an ideal goal would be a single optimized network to deal with every game in the market, we have to rely on default mechanisms to get the job done.

Huawei is going after its core gaming market first with GPU Turbo, which means plenty of Battle Royale and MOBA action, like PUBG and Arena of Valor, as well as tie-ins with companies like EA/Tencent for NBA 2K18. I suspect on the back of this realization, some companies will want to get in contact with Huawei to add their title to the list of games to be optimized. Our only request is that you also include tools so we can benchmark the game and output frame-time data, please!

On the next page, we go into our analysis on GPU Turbo with devices on hand. We also come across an issue with how Arm’s Mali GPU (used in Huawei Kirin SoCs) renders games differently to Huawei’s competitor devices.

The Claimed Benefits of GPU Turbo: Huawei’s Figures The Difficulty in Analyzing GPU Turbo
POST A COMMENT

64 Comments

View All Comments

  • Ian Cutress - Tuesday, September 4, 2018 - link

    In the past, those 'cheats' were often from not rendering parts of the scene. This is still doing the full render that any Mali GPU does, but in a more power efficient way. The key to benchmarking is to test across several titles regardless, which is going to be important moving forward. Reply
  • Manch - Wednesday, September 5, 2018 - link

    Does Mali or any mobile GPUs do culling of unseen objects? If not, can that be implemented to further reduce load? Reply
  • The Hardcard - Tuesday, September 4, 2018 - link

    That isn’t a quandry, it solves the problem. The problem before is that the makers showed benchmark performance that they didn’t feel the device could handle in normal user apps. If this pans out and users can have it everyday apps means no harm, no foul.

    Having it be a special mode for apps that can use it, while turning it off when it is not necessary is exactly what is needed and what everyone is trying to do and should do.

    If they do it properly, then it is on the developers to use it. Sure, older, unupdated apps will be left behind. That is the nature of advancing technology.
    Reply
  • melgross - Tuesday, September 4, 2018 - link

    A benchmark cheat is just for benchmarks. There’s a reason for that, and it has to do with the fact that the SoC, and the device, as a whole, can’t perform at that level commercially, otherwise something negative will happen, such as overheating, and battery failure.

    So, no, they can’t extend cheating to regular apps, and that’s the entire point to the cheat. If they could, then they would, and it wouldn’t be a cheat. This cheating is different from the turbo mode the article is about.
    Reply
  • s.yu - Monday, September 10, 2018 - link

    The only way this is working is the apparent popularity of MMO games. They only plan on catering to low end customer who only play whatever "everybody else" plays. I for one avoid them like the plague, IAP rigged games are cheap stimulation, too cheap. Reply
  • tipoo - Tuesday, September 4, 2018 - link

    Reminds me of the good old ATI vs Nvidia days when there were notable differences in render quality, usually with the edge to ATI. That all but went away at least as far back as the 8800, maybe before. Now for mobile to repeat that process. Reply
  • Ian Cutress - Tuesday, September 4, 2018 - link

    Just to make sure you're aware, that's kind of orthogonal to GPU Turbo. It's Mali behaviour right now, which explains some of the perf differences, but GPU Turbo is something separate. Reply
  • Lord of the Bored - Wednesday, September 5, 2018 - link

    Not ALWAYS to ATI, though. Sometimes they got a little aggressive in their "optimizations" too.
    QUAFF3 NEVER FORGET!

    https://techreport.com/review/3089/how-ati-drivers... ffor the kiddos that never saw this one. Back when men were men, and PC gaming was the exclusive domain of nerds that knew what IRQ and DMA meant(but probably not PCMCIA. No one could remember PCMCIA).
    Reply
  • Holliday75 - Friday, September 7, 2018 - link

    I recently found a PCMCIA 10mb NIC in one of my file cabinets and a 28.8k modem. I looked at them a second like wtf then remembered what they were. Reply
  • nils_ - Friday, September 7, 2018 - link

    People Can't Memorize Computer Industry Acronyms Reply

Log in

Don't have an account? Sign up now