Quick Answer
The Google Tensor G2 is the successor to the original Google Tensor chip. While both are custom-designed system-on-chips (SoCs) for Pixel devices, the Tensor G2 generally offers improved CPU and GPU performance, more efficient machine learning capabilities, and enhanced image processing for computational photography.
Google Tensor G2 vs Google Tensor: Full Comparison
Introduction
When comparing generations of smartphone technology, the processor is a central point of analysis. This comparison examines the differences between Google’s first custom mobile chip, the Tensor, and its successor, the Tensor G2. Understanding the evolution between these two processors provides insight into performance improvements, efficiency gains, and the direction of Google’s hardware-software integration, which is a key factor for users considering devices powered by these chips.
Architecture and Manufacturing Process
The fundamental design and construction of a chipset influence its performance and power efficiency.
- Google Tensor: Manufactured using Samsung’s 5-nanometer (5LPE) process. Its CPU configuration typically featured two high-performance Cortex-X1 cores, two mid-range Cortex-A76 cores, and four efficiency-focused Cortex-A55 cores.
- Google Tensor G2: Also built on a 5nm process (Samsung’s 4LPE node, which is an enhanced version). It upgraded the CPU cores, moving to newer Cortex-X1, Cortex-A78, and Cortex-A55 cores, offering a balance of improved performance and better power management.
The move to a refined manufacturing node with the G2 typically allows for slightly better power efficiency and thermal management under load compared to the first-generation Tensor.
Performance and GPU
This area covers general computational speed and graphics rendering capabilities.
- CPU Performance: The Tensor G2’s updated CPU cores provide a moderate boost in processing speed for everyday tasks and multi-threaded applications. The difference is often most noticeable in sustained performance, where thermal throttling may be less pronounced.
- GPU Performance: The original Tensor used a Mali-G78 MP20 graphics processor. The Tensor G2 upgraded to the Mali-G710 MP7. While the core count is lower, the G710 architecture is more modern and efficient, generally delivering better graphics performance for gaming and visual effects with improved power efficiency.
AI, Machine Learning, and Photography
A primary focus of Google’s Tensor chips is accelerating on-device AI and machine learning (ML) tasks, which directly powers many camera and software features.
- Tensor Processing Unit (TPU): Both chips include a custom TPU. The Tensor G2 features a next-generation TPU that is reportedly 60% faster and 20% more power-efficient for specific ML workloads. This enables more complex on-device processing.
- Computational Photography: The enhanced TPU and image signal processor (ISP) in the Tensor G2 enable more advanced camera features. This includes improvements to existing modes like Night Sight and Face Unblur, and the introduction of new capabilities such as Photo Unblur and more responsive Magic Eraser in Google Photos.
- Speech & Language: Both chips excel at on-device speech recognition and language translation, with the G2 offering incremental improvements in speed and accuracy for features like Live Translate and Recorder app summaries.
Connectivity and Other Features
These specifications affect data speeds, call quality, and peripheral support.
- Modem: Both chips integrate a Samsung Exynos modem. The Tensor G2 uses a slightly newer iteration (Exynos 5300 vs. 5123 in the original), which may offer marginal improvements in 5G connectivity and power consumption in most regions.
- Security: Both processors include Google’s Titan M2 security coprocessor for hardware-level protection of sensitive data like passwords and transaction information.
Comparison Table: Google Tensor vs Tensor G2
| Feature | Google Tensor | Google Tensor G2 |
|---|---|---|
| Manufacturing Process | Samsung 5nm LPE | Samsung 4nm LPE (Enhanced 5nm) |
| CPU Cores | 2x Cortex-X1 @ 2.80 GHz 2x Cortex-A76 @ 2.25 GHz 4x Cortex-A55 @ 1.80 GHz |
2x Cortex-X1 @ 2.85 GHz 2x Cortex-A78 @ 2.35 GHz 4x Cortex-A55 @ 1.80 GHz |
| GPU | ARM Mali-G78 MP20 | ARM Mali-G710 MP7 |
| Machine Learning TPU | Custom Google TPU | Next-gen Google TPU (Faster & more efficient) |
| Modem | Exynos Modem 5123 (Integrated) | Exynos Modem 5300 (Integrated) |
| Notable Camera Features | Face Unblur, Magic Eraser, Night Sight | Enhanced Face Unblur, Photo Unblur, Faster Magic Eraser, Cinematic Blur |
| Security | Titan M2 Security Chip | Titan M2 Security Chip |
| First Featured In | Pixel 6 & Pixel 6 Pro | Pixel 7 & Pixel 7 Pro |
Frequently Asked Questions (FAQ)
What is the main difference between Google Tensor and Tensor G2?
The main differences lie in the updated CPU/GPU architectures and a significantly more powerful and efficient Tensor Processing Unit (TPU) in the G2. This translates to better overall performance, improved power management, and more advanced on-device AI and photography features.
Is the Tensor G2 much faster for gaming than the original Tensor?
The Tensor G2’s Mali-G710 GPU generally provides better graphics performance and efficiency than the Tensor’s Mali-G78. This can result in smoother gameplay at higher settings and potentially longer sustained performance before thermal throttling occurs.
Do both chips support the same camera software features?
While many core computational photography features are available on both, the Tensor G2’s enhanced ISP and TPU enable newer and improved capabilities, such as Photo Unblur for old photos and a more responsive Magic Eraser. Features like Face Unblur are also more effective on the G2.
Which chip is more power-efficient?
The Tensor G2, built on a refined manufacturing process and with more efficient CPU/GPU cores, typically offers better power efficiency. This can contribute to longer battery life under similar usage conditions compared to a device with the original Tensor chip.
Final Thoughts
The Google Tensor G2 represents a clear, iterative evolution over the first-generation Google Tensor. It builds upon the original’s foundation—strong AI acceleration and deep software integration—by delivering measurable improvements in general performance, graphics, and power efficiency. The most significant advancements are often seen in the realm of machine learning, which unlocks more sophisticated camera and language features. The choice between devices using these chips often comes down to valuing the latest performance and feature refinements offered by the G2 versus the capable, foundational experience provided by the original Tensor.