MediaTek Dimensity 9500 Shows Google’s Tensor G5 The Way

When MediaTek's Dimensity 9500 chip is able to handily beat Google's new Tensor G5 SoC in one benchmark test after another, and that too by simply leveraging ARM's generic technology. It is clear that Google's silicon design process has left quite a lot to be desired.

Google is now loudly proclaiming its disdain for benchmark tests as the yardstick to measure the success of its silicon design process. Agreed. These tests, after all, do not tell the whole story on their own.

But, when you are charging Apple-level prices for a silicon that shows a proclivity for heating up at the first inkling of a compute-intensive task, there is an evident need for introspection. And, MediaTek's Dimensity 9500 appears willing to show Google the way.

The Architecture Of MediaTek's Dimensity 9500 Vs. Google's Tensor G5

1. CPU
   • Dimensity 9500 sports an eight-core CPU, consisting of one ARM C1-Ultra core clocked at 4.21 GHz with an L2 cache of 2MB, three ARM C1-Premium cores clocked at 3.50 GHz with a L2 cache of 1MB, and four ARM C1-Pro cores clocked at 2.70 GHz with a L2 cache of 512KB.
   • Tensor G5 also sports an eight core CPU, made up of one ARM Cortex-X4 core clocked at 3.78 GHz, five ARM Cortex-A725 cores clocked at 3.05 GHz, and two ARM Cortex-A520 cores clocked at 2.25 GHz. Please note that Google has not publicly disclosed the L2 cache sizes of Tensor G5's CPU cores.
2. GPU
   • Dimensity 9500 features the new ARM Mali-G1 Ultra MC12 GPU that offers superior ray-tracing performance and can deliver fluid gameplay at 120fps.
   • Google's Tensor G5 sports an Imagination IMG DXT-48-1536 GPU sans ray-tracing support.
3. AI
   • Dimensity 9500 sports the new MediaTek NPU 990 to handle AI and machine learning workloads.
   • Tensor G5 features a bespoke TPU to handle AI workloads.  

MediaTek's Dimensity 9500 Shows Google How To Produce A Near-Perfect SoC At Reduced Costs And Without Cutting Corners, As Is The Case With Tensor G5 In Base Pixel 10

We noted in a recent post that Google's Tensor G5 suffers from a piecemeal approach to the chip design process. Even so, there are additional nuances to this story.

Consider the fact that ARM Cortex-X4, the high-performance core within the Tensor G5, is now around 2 and a half years old. Google chose to stick with ARM's older CPU cores to save on costs, while MediaTek's Dimensity 9500 leveraged ARM's latest technology.

2 of 9

And, the difference is obvious, with Dimensity 9500 managing to absolutely smoke the Tensor G5 in the GeekBench 6 test.

What's more, Google deliberately axed its benchmark test scores by refusing to bring the Imagination IMG DXT-48-1536 GPU's ray-tracing support to the Tensor G5.

It All Boils Down To Cost

MediaTek's Dimensity 9500 costs roughly between $180 and $200, which constitutes a ~50 percent discount to Qualcomm's Snapdragon 8 Elite Gen 5. In contrast, we know from previous reports that Google was targeting a $65 per unit price cap for the Tensor G5.

My contention is simple: Google's Pixel 10 starts at $799, which is the exact same price charged by Apple for the base iPhone 17. When Google chooses to charge iPhone 17-level prices, its smartphones should deliver the same all-around performance.

Yet, here too, the otherwise excellent Pixel 10 smartphones fall victim to Google's penny-pinching ways. How can the tech giant justify the lack of a dedicated vapor chamber in the base Pixel 10, especially when it would have known about the Tensor G5's proclivity to heat up and throttle under intense workloads?

Compare this relatively haphazard approach with the one adopted by MediaTek, which managed to cut Dimensity 9500's cost without compromising on performance, or resorting to band-aid measures such as eschewing ray-tracing support.

And, before anyone accuses me of nitpicking, ask yourself: Shouldn't Google have discounted its base Pixel 10 relative to the Apple iPhone 17 when it knew that it had deliberately hobbled the phone?

No one would have complained had the base Pixel 10 launched at a $100 discount to the base iPhone 17. This is not what occurred though. Google chose to increase its own margins at the expense of its consumers.

To sum up, MediaTek's Dimensity 9500 can teach Google the following important lessons for the next generations of its Tensor chips:

1. Stick with generic but latest-gen ARM CPU cores if the alternative is weirdly applied cost-cutting.
2. Do not go for bespoke GPUs if you are only going to hobble them by eschewing critical features such as ray-tracing support.
3. Do not try to opt for extreme penny-pinching measures, especially in relation to the Tensor SoC, while thinking that adding AI capabilities would make up for a hobbled SoC.
4. And, most importantly, do not charge Apple-level prices if your finished product does not match iPhone-level performance.

Follow Wccftech on Google to get more of our news coverage in your feeds.