Share link|

Submit benchmark

Comparison Search Benchmarks CPUs Leaderboards CPU News & Articles GPU comparison Apple product comparison

Deutsch English Español Français Italiano Japanese

Google Tensor G4 vs Apple M3 Ultra (32-CPU 80-GPU)

Last updated: 2026-06-04

Table of Contents

Summary
Benchmarks
Specifications
User Reviews
Popular Comparisons

Summary

Benchmarks

Performance Overview
Average performance across multiple benchmarks

Single-core performance

Rank 2 - 60 %

Google Tensor G4

Rank 1 - 100 %

Apple M3 Ultra (32-CPU 80-GPU)

Multi-core performance

Rank 2 - 16 %

Google Tensor G4

Rank 1 - 100 %

Apple M3 Ultra (32-CPU 80-GPU)

Geekbench 6 Single-Core
Single-core performance

8C / 8T · 3.10 GHz

1,927

32C / 32T · 4.06 GHz

3,240

Geekbench 6 Multi-Core
Multi-core performance

8C / 8T · 3.10 GHz

4,550

32C / 32T · 3.60 GHz

28,170

iGPU - FP32 Performance (Single-precision GFLOPS)

ARM Immortalis-G715 MP10

Apple M3 Ultra (80 Core)

28,400

More benchmarks

Cinebench 2026 (ST)
218 entries

Cinebench 2026 (MC)
219 entries

Cinebench 2024 (SC)
427 entries

Cinebench 2024 (MC)
422 entries

Cinebench R23 (SC)
690 entries

Cinebench R23 (MC)
665 entries

Geekbench 6 (SC)
1,989 entries

Geekbench 6 (MC)
1,927 entries

Geekbench 5 (SC)
2,512 entries

FP32 SP (iGPU)
2,219 entries

AI performance (NPU)
218 entries

Specifications

At a glance

Wischen	Google Tensor G4	Apple M3 Ultra (32-CPU 80-GPU)
Family	Google Tensor (5)	Apple M series (38)
CPU group	Google Tensor G4 (1)	Apple M3 Ultra (2)
Architecture	G4	M3
Technology	4 nm	3 nm
Segment	Smartphone / Tablet	Desktop
Socket
Predecessor	Google Tensor	Apple M2 Ultra (76-GPU)
Successor	Google Tensor G5

CPU Cores and Base Frequency

Wischen	Google Tensor G4	Apple M3 Ultra (32-CPU 80-GPU)
CPU Cores / Threads	8 / 8	32 / 32
Hyperthreading / SMT	✗	✗
Core architecture	hybrid (Prime / big.LITTLE)	hybrid (big.LITTLE)
Core Cluster 1:	1x Cortex-X4 3.10 GHz	24x Palma P-Core 0.70 - 4.06 GHz
Core Cluster 2:	3x Cortex-A720 2.60 GHz	8x Palma E-Core 0.74 - 2.75 GHz
Core Cluster 3:	4x Cortex-A520 1.95 GHz
L2-Cache		80.00 MB
L3-Cache
Overclocking	✗	✗

Integrated graphics (iGPU)

Wischen	Google Tensor G4	Apple M3 Ultra (32-CPU 80-GPU)
GPU name	ARM Immortalis-G715 MP10	Apple M3 Ultra (80 Core)
GPU frequency	0.89 GHz	0.39 - 1.40 GHz
CUs / Shader	10 /	1280 / 10240
Raytracing	✗	✓
Max. displays	2	8
Max. GPU Memory		512 GB
Technology	4 nm	3 nm
Release date	Q4/2022	Q1/2025

NPU AI performance

Wischen	Google Tensor G4	Apple M3 Ultra (32-CPU 80-GPU)
AI hardware	Google Tensor AI	Apple Neural Engine
AI specifications	Google Edge TPU	32 Neural cores @ 70 TOPS
NPU + CPU + iGPU

Memory & PCIe

Wischen	Google Tensor G4	Apple M3 Ultra (32-CPU 80-GPU)
Memory	LPDDR5-6400 (51.2 GB/s)	LPDDR5-6400 (818.7 GB/s)
Max. Memory	16 GB	512 GB
Memory channels	4	8
ECC	No	No
PCIe		5.0
PCIe Bandwidth

Thermal Management

Wischen	Google Tensor G4	Apple M3 Ultra (32-CPU 80-GPU)
TDP	10 W	180 W
TDP (PL2)
TDP up
TDP down
T. junction max.		100 °C

Technical details

Wischen	Google Tensor G4	Apple M3 Ultra (32-CPU 80-GPU)
Chip design	Chiplet	Monolithic
AES-NI	✗	✓
Operating systems	Android	macOS, iPadOS
Instruction set	Armv8.7-A (64 bit)	Armv8-A (64 bit)
ISA extensions		Rosetta 2 x86-Emulation
Release date	Q4/2024	Q1/2025
Release price
Documents

User Reviews

Your rating

Here you can rate one or both of the processors. The average rating of Google Tensor G4 is 3.8 stars (175 ratings), while Apple M3 Ultra (32-CPU 80-GPU) has received 4.9 stars (452 ratings).

Click on the stars to rate one or both processors.

Google Tensor G4
8C / 8T · 3.10 GHz

★★★★★

Apple M3 Ultra (32-CPU 80-GPU)
32C / 32T · 0.70 - 4.06 GHz

★★★★★