Google Tensor G4
Google Tensor G4: A New Stage in the Evolution of Mobile Processors
April 2025
Since the launch of the first Tensor in 2021, Google has continued to improve its chips, focusing on AI integration and optimization for the Android ecosystem. The Tensor G4, introduced at the end of 2024, was the company's response to the growing demands for mobile performance. In this article, we will explore what the new processor hides under the hood, how it performs in real tasks, and who should pay attention to it.
1. Architecture and Process Technology: Innovations in Detail
CPU: 8 Cores with a Focus on AI
The Tensor G4 is built on a hybrid architecture with 8 cores divided into three clusters:
- 1× Cortex-X4 at 3.1 GHz for resource-intensive tasks.
- 4× Cortex-A720 (2.8 GHz) — a balance between performance and power consumption.
- 3× Cortex-A520 (2.0 GHz) for background processes.
This configuration ensures smooth multitasking and reduces heat generation. The 4 nm process from Samsung (SF4P) has improved energy efficiency by 15% compared to Tensor G3.
GPU: Mali-G715 MP7 — Gaming Without Compromises
The Mali-G715 MP7 graphics accelerator with 7 cores supports hardware ray tracing and VRS (Variable Rate Shading) technology, enhancing detail in games while reducing the load on the GPU. The GPU clock speed reaches 900 MHz, and its performance in the 3DMark Wild Life Extreme test is 4200 points, which is 25% higher than its predecessor.
NPU: AI as the Foundation
The specialized Titan Edge NPU delivers 73 TOPS (trillions of operations per second) for machine learning tasks. This allows for voice assistant requests to be processed in 0.3 seconds and real-time photo enhancement.
2. Real-World Performance: From Gaming to AI
Gaming
The Tensor G4 handles AAA titles like Genshin Impact at Ultra settings (60 FPS) without drops. During a 40-minute session, the smartphone's chassis temperature does not exceed 42°C thanks to a cooling system with a vapor chamber. However, under peak loads (e.g., in Honkai: Star Rail), throttling can occur down to 85% of the maximum frequency.
Multimedia
The chip supports 8K@60fps decoding and AV1 encoding, which is crucial for streamers. Displays with a refresh rate of 1-120 Hz dynamically adapt to content, saving power.
AI Applications
- Photography:
- Night shots are processed 30% faster than on Tensor G3.
- The "Cinematic Video" mode automatically adjusts stabilization and depth of field.
- Voice Input:
- The offline translation mode operates with 98% accuracy, even without internet access.
Power Consumption
With moderate use (social media, music, 2 hours of video), a smartphone with Tensor G4 (with a 5000 mAh battery) lasts up to 28 hours. In intensive mode (gaming, 4K recording) — around 6 hours.
3. Built-in Modules: The Connectivity of the Future
- Exynos 5400 Modem:
Supports 5G mmWave and Sub-6, frequency aggregation for speeds up to 10 Gbps. In areas with poor signal, the AI connection optimizer is automatically activated.
- Wi-Fi 7 and Bluetooth 5.4:
Data transfer speeds up to 40 Gbps, with latency under 3 ms in gaming. Bluetooth utilizes LC3+ codecs for high-resolution audio.
- Navigation:
Supports GPS, GLONASS, Galileo, BeiDou, and QZSS. In emergencies, communication through Iridium satellites is available (subscription required).
4. Comparison with Competitors: Who are the Leaders?
- Tensor G4 vs Snapdragon 8 Gen 4:
Qualcomm leads in GPU performance (Adreno 760 vs Mali-G715 MP7: 4800 vs 4200 points in 3DMark), but Tensor G4 excels in AI tasks (73 vs 65 TOPS).
- Tensor G4 vs Apple A18 Pro:
Apple's chip is stronger in single-core performance (Geekbench 6 SC: 2450 vs 2045), but Tensor G4 offers a better price-to-camera capabilities ratio.
- Tensor G4 vs Dimensity 9300:
MediaTek is more power-efficient in gaming (by 10%), but falls short in update support (3 years compared to 5 for Google).
5. Use Cases: Who is Tensor G4 Designed For?
- Gamers:
Suitable for gaming at QHD+ (1440p) with HDR, but for 4K gaming, flagship devices with Snapdragon are better.
- Amateur Photographers:
Magic Eraser Pro and Night Sight Video modes turn the smartphone into a professional tool.
- AI Enthusiasts:
Local processing of requests in Gemini Ultra and image generation in seconds.
6. Pros and Cons
Strengths:
- Best-in-class integration with Google services.
- Long-term update support (5 years).
- Energy efficiency in everyday tasks.
Weaknesses:
- Heating during prolonged gaming.
- Lack of low-latency stylus support (unlike Snapdragon).
7. Device Selection Tips
- What to Look For:
- Cooling: Devices with vapor chambers (e.g., Pixel 9 Pro, $999) run cooler.
- RAM: At least 12 GB for AI tasks.
- Display: LTPO 3.0 with a refresh rate of 1-120 Hz for battery life.
- Typical Devices:
- Premium Segment: Google Pixel 9 Pro ($999), Nothing Phone 3 ($799).
- Mid-Tier: Expected in 2026.
8. Final Conclusion: Who is Tensor G4 For?
The processor is designed for those who value:
- A DSLR-level camera in their pocket.
- Long-term update support.
- A seamless Google ecosystem without lag.
With a budget of $800 or more, Tensor G4 is an optimal choice. However, hardcore gamers or users needing maximum power should consider Snapdragon 8 Gen 4 or Apple A18 Pro.
Tensor G4 confirms that Google is focusing not on "numbers in benchmarks," but on the harmony of hardware and software. And therein lies its main strength.
