MediaTek Dimensity 9500 Outperforms Google Tensor G5 in Latest Benchmark

The recent benchmarks show MediaTek’s Dimensity 9500 chip outshining Google’s Tensor G5 SoC on multiple fronts by effectively utilizing ARM’s latest technologies. This raises critical questions about the effectiveness of Google’s approach to silicon design, indicating potential shortcomings in their strategy.

In light of these developments, Google has begun to publicly reject benchmark tests as definitive measures of chip performance. While it’s true these assessments don’t capture the full picture, they remain a vital representation of a chip’s capabilities—especially when a product is priced similarly to offerings from Apple, yet struggles with overheating and performance under load.

Comparative Architecture: MediaTek Dimensity 9500 vs. Google Tensor G5

CPU Configuration
- The Dimensity 9500 is equipped with an eight-core architecture that includes one ARM C1-Ultra core running at 4.21 GHz with a 2MB L2 cache, three ARM C1-Premium cores at 3.50 GHz with a 1MB L2 cache, and four ARM C1-Pro cores clocked at 2.70 GHz with a 512KB L2 cache.
- In comparison, the Tensor G5 also features an eight-core CPU composed of one ARM Cortex-X4 core at 3.78 GHz, five ARM Cortex-A725 cores clocked at 3.05 GHz, and two ARM Cortex-A520 cores at 2.25 GHz, although Google has not disclosed the L2 cache specifications for these cores.
GPU Performance
- Dimensity 9500 integrates the newly launched ARM Mali-G1 Ultra MC12 GPU, which delivers enhanced ray-tracing capabilities and a smooth gaming experience at 120fps.
- Conversely, the Tensor G5 is powered by an Imagination IMG DXT-48-1536 GPU, lacking support for ray tracing entirely.
AI Processing
- The Dimensity 9500 includes MediaTek’s advanced NPU 990 for AI and machine learning tasks.
- The Tensor G5 employs a custom TPU for handling similar AI workloads.

MediaTek’s Dimensity 9500: A Lesson in Efficiency for Google’s Tensor G5

Recent discussions have highlighted Google’s fragmented approach to chip design with the Tensor G5, but it goes deeper. The ARM Cortex-X4 core within the Tensor G5 has been around for nearly two and a half years—pointing to a decision by Google to prioritize cost savings by using older technology, while MediaTek opted for ARM’s latest offerings in their Dimensity 9500.

GeekBench 6 benchmark screen shows scores: Single-Core 2301, Multi-Core 6987. — Google Tensor G5 GeekBench 6 Results

GeekBench 6 scores on benchmark screen: Single-Core Score 3460, Multi-Core Score 10192, Compute Score (GPU) 24891. — MediaTek Dimensity 9500 GeekBench 6 Results

The disparity is stark, with the Dimensity 9500 defeating the Tensor G5 in the GeekBench 6 tests. Furthermore, Google’s decision to limit the GPU capabilities by excluding ray tracing from its offering further undercuts performance metrics.

A Financial Perspective

The Dimensity 9500 comes at a competitive price range of $180 to $200, nearly a 50% reduction compared to Qualcomm’s Snapdragon 8 Elite Gen 5. Conversely, Google’s target pricing strategy for the Tensor G5 is around $65 per unit.

With Google’s Pixel 10 starting at $799, paralleling the cost of the base iPhone 17, one could argue that the performance should reflect that premium price. However, the Pixel 10 seems to falter due to Google’s restrictive cost strategies. Notably, the absence of a dedicated vapor chamber raises concerns, especially given the Tensor G5’s tendency to overheat under demanding tasks.

In contrast, MediaTek has successfully produced the Dimensity 9500 by minimizing expenses without sacrificing quality or resorting to compromises like the omission of advanced features. This strategy points to a more consumer-focused approach.

Finally, with appropriate reflection, Google should consider a more honest pricing strategy. Offering the Pixel 10 at a reduced price compared to the iPhone 17, given its limitations, could align perceptions of value better. It would be reasonable to expect a $100 discount for the Pixel 10 to match consumer expectations more closely.

To summarize, MediaTek’s Dimensity 9500 provides Google with significant lessons for future Tensor chip designs:

Utilize the latest ARM CPU technology instead of aging cores to avoid unintentional cost-cutting.
Avoid bespoke GPUs unless they can be fully featured, including essential capabilities like ray tracing.
Avoid extreme cost reductions that jeopardize product integrity in the Tensor lineup.
Ensure pricing aligns with performance to meet customer expectations comparable to that of leading competitors.

Source&Images