

Raspberry Pi Compute Module 4 (CM4) vs Banana Pi CM4
Benchmarks, specifications, and real-world performance data
Tested by Bret Weber Last updated
Fastest
Banana Pi CM4
1,293 GB6 multi-core
Best value
Raspberry Pi Compute Module 4 (CM4)
9.9 GB6 pts/$ at $62
Most efficient
Raspberry Pi Compute Module 4 (CM4)
4.2W under load
Comparison Summary
The Raspberry Pi Compute Module 4 (CM4) and the Banana Pi CM4 go head to head below, with every figure taken from our own hands-on testing. In raw CPU performance the Banana Pi CM4 leads, its 1,293 Geekbench 6 multi-core score coming in 112% ahead of the Raspberry Pi Compute Module 4 (CM4) at 609. Factor in price and the Raspberry Pi Compute Module 4 (CM4) makes the strongest case, working out to 9.9 Geekbench 6 points per dollar at $61.62. The Raspberry Pi Compute Module 4 (CM4) is the most frugal under load, drawing 4.2W.
Read our individual reviews:
| Specification | Banana Pi CM4 4GB RAM | |
|---|---|---|
| Basic Specifications | ||
| Price | ||
| RAM | 2GB | 4GB |
| SoC | ||
| Architecture | ARM | ARM |
| CPU Cores | 4 cores 4× Cortex-A72 @ 1.5GHz | 6 cores 4× Cortex-A73 @ 2.2GHz 2× Cortex-A53 @ 2.0GHz |
| Hardware Video Decode | H.264, HEVC, HEVC 10-bit Up to 4K60 | None (CPU only) MPEG-2 (partial) Drivers installed by the test |
Every figure above comes from hardware run in-house by Bret Weber, on the exact OS build, kernel, firmware and boot device listed here. Boards are tested to a fixed procedure so results stay comparable between them: see our testing methodology for how each benchmark is run.
CPU Performance
Processor performance across single and multi-core workloads
Geekbench 6
Higher Scores are Better
7-Zip Benchmark
Higher Scores are Better
High Performance Linpack
Higher Scores are Better
UnixBench
Higher Scores are Better
PassMark CPU
Higher Scores are Better
CPU Mining (cpuminer-multi)
Higher Hashrates are Better
MAME: Street Fighter II: The World Warrior
Higher Speed is Better
Geekbench 7
Higher Scores are Better
Only this board has been tested so far, so there is nothing to compare against yet.
Memory Performance
Memory bandwidth and throughput benchmarks
TinyMemBench
Higher Bandwidth is Better
GPU Performance
Graphics rendering with OpenGL and Vulkan, plus raw Vulkan compute throughput
AI & LLM Performance
Large Language Model inference performance
Ollama LLM Benchmark
Higher Scores are Better
Storage Performance
Disk I/O throughput measured with FIO
fio Storage Performance - eMMC
Higher Speeds are Better • Raspberry Pi CM4 (16GB eMMC), Banana Pi CM4 (16GB eMMC)
fio Storage Performance - microSD
Higher Speeds are Better • Raspberry Pi 64GB (64GB microSD)
Only this board has been tested so far, so there is nothing to compare against yet.
fio Storage Performance - M.2 NVMe
Higher Speeds are Better • Crucial P5 Plus (1024GB M.2 NVMe)
Only this board has been tested so far, so there is nothing to compare against yet.
Network Performance
Ethernet and WiFi throughput measured with iPerf3
iPerf3 Network Performance
Higher Speeds are Better
Media & Video
Jellyfin transcoding and FFmpeg video decode and convert speed
Jellyfin Transcoding
Simultaneous real-time streams, more is better
2 more scenarios and the CPU and hardware stream counts are in the benchmark table · every board ranked
Video Decoding
Software decoding on the CPU, higher fps is better
8 more workloads and the hardware decode speeds are in the benchmark table · every board ranked
Power & Thermal
Power consumption at idle and under various workloads
Power Consumption
Lower is "Better"
Cooling & Thermal Testing
Frequently Asked Questions
Which is faster, the Banana Pi CM4 or the Raspberry Pi Compute Module 4 (CM4)?
The Banana Pi CM4 is faster in our testing, scoring 1,293 in Geekbench 6 multi-core versus 609 for the Raspberry Pi Compute Module 4 (CM4), a 112% advantage.
Which board is the best value?
Of the boards in this comparison, the Raspberry Pi Compute Module 4 (CM4) offers the best performance per dollar: 9.9 Geekbench 6 multi-core points per dollar at $61.62.
Add One More Board
You can compare up to 3 boards at once. Add one more to get the complete comparison.