Table of Contents
Hardware Specifications
TetherBox runs on Intel, AMD, and ARM-based Linux systems. Start by choosing hardware for the number of cameras you will record, then size its CPU, RAM, storage, and - if needed - AI acceleration using the guidance below.
Choose Hardware for Your Deployment
| Deployment | Recommended hardware | Typical recording capacity |
|---|---|---|
| Small or space-constrained | TetherBox: Raspberry Pi 5 with cooling and external SSD or M.2 storage | Up to 10 cameras |
| Small to medium | TetherBox Pico: Intel N100 or N150 mini PC with 8 GB RAM and SSD storage | Around 16-20 cameras |
| Larger or expandable | TetherBox Giga: Compact desktop, tower or rackmount system, sized for the site | More than 20 cameras |
These capacities assume the recommended per-stream settings in camera configuration. See CPU recommendations for how different camera settings affect them.
TetherBox: Raspberry Pi
A Raspberry Pi 5 is a good fit for low-power, mobile, concealed, or space-constrained installations. It has a benchmark capacity of about 11 cameras; use it for up to 10 to leave headroom. A Raspberry Pi 4 handles up to 4 cameras.
Use a cooled enclosure, the official USB-C power supply, and a suitable USB SSD or M.2 drive. SD cards and USB memory sticks are not suitable for continuous recording. Choose an enclosure with passive or active cooling and enough clearance for an M.2 or AI HAT where required.
For complete Raspberry Pi setup guidance, including on-device AI analytics, see Raspberry Pi TetherBox.
TetherBox Pico: Mini PC
An Intel N100 or N150 mini PC is a practical choice for approximately 16 to 20 cameras. Choose a system with 8 GB RAM, reliable SSD storage, and enough USB or internal storage connectivity for the required retention. The Shuttle XPC DL30N, Gigabyte Mini PCs, and ASUS Mini PCs are suitable examples. Select the enclosure and storage configuration for the installation environment. On Intel N100 and N150 systems, the hardware watchdog works only once it is enabled in the BIOS; see Hardware Watchdog.
For sustained AI workloads, prefer active cooling and check clearance for the accelerator. Confirm the exact M.2 key, PCIe lanes and storage connections before choosing a Hailo module; an available M.2 socket alone does not establish compatibility. See Interface Requirements.
TetherBox Giga: Rackmount or Tower
For larger deployments, build or source a system to the site's CPU, memory, storage, and resilience requirements. Standard motherboard and CPU components, configured rackmount systems, and corporate-vendor servers are all suitable. A dedicated GPU does not require a full-size tower: Mini-ITX can suit a compact single-GPU unit, while micro-ATX can reduce cost or add expansion space.
The following configurations illustrate the required specification without tying the choice to a particular supplier:
| Target deployment | Compute and memory | Recording storage |
|---|---|---|
| Around 64 cameras | AMD Ryzen 7 7700 with 16 GB RAM. Check the CPU reference table and leave headroom above the target count. | Four 18 TB enterprise drives in Linux software RAID5 provide about 54 TB usable, against an estimated 48 TB requirement for 30 days of 4K recording. A 1U chassis with four 3.5-inch and four 2.5-inch hot-swap bays, eight SATA connections, and M.2 operating-system storage is suitable. |
| Around 128 cameras | AMD Ryzen 9 9950X with 32 GB ECC RAM. Check the CPU reference table and leave headroom above the target count. | Eight 18 TB enterprise drives in RAID6 provide about 108 TB usable, against an estimated 96 TB requirement. Use a chassis with at least eight 3.5-inch bays and a suitable RAID controller where required. |
Compact AI Trial Example
See Compact TetherBox Demo Build for the compact NVIDIA reference build, assembly checks, recording-capacity assumptions and separate Australian, Canadian and UK buying examples. Size the recording count and AI workload independently using the guidance below.
Sizing at a Glance
Most TetherBox deployments need enough CPU for the camera count, enough RAM for analytics, and reliable storage for recordings. AI analytics may also require an accelerator or GPU.
| Component | Rule of thumb | Details |
|---|---|---|
| CPU | Choose from the CPU reference tables and leave workload headroom | See CPU Requirements and CPU Recommendations |
| RAM | About 1.5 GB, plus 1 GB per 10 cameras, 4 GB minimum | See RAM Requirements |
| Storage | About 750 GB per 4K camera for 30 days (50% motion, 4096 kbit average) | See Storage Requirements and Storage Hardware Recommendations |
| AI acceleration | Only required for TetherX AI analytics | See Graphics / GPU |
CPU Requirements
Use CPU Recommendations for the current purchasing and older/deployed reference tables, GeekBench 6 sizing rule, benchmark assumptions and camera configuration effects. Keep CPU estimates separate from GPU AI capacity and recording retention.
RAM Requirements
Formula: about 1.5 GB for the system, plus 1 GB per 10 cameras, rounded up to the next standard memory size, with 4 GB as the minimum. These figures come from memory use measured on TetherBoxes in service: 60-camera TetherBoxes with 16 GB use between 5 and 8 GB.
| Cameras | RAM | Example systems |
|---|---|---|
| 1-16 | 4 GB | Raspberry Pi, mini PC, N100 fanless |
| 17-40 | 8 GB | Mini PC, rack units |
| 41-100 | 16 GB | High-density rack |
| 101+ | 32 GB | Large installations |
These are recording/system memory baselines. Allow additional RAM for AI workloads and model development; 16 GB is a sensible starting point for the compact NVIDIA trial example above even though its recording count falls in the 8 GB tier.
Warning: Systems sustaining more than 85% memory usage can experience recording failures and instability. After adding cameras, check memory usage on the TetherBox page, and choose the next size up if it stays above 85%.
Storage Requirements
Storage depends on resolution, motion activity, and retention period. These estimates are per camera for 30 days of retention, including video recording and a 720p snapshot every five seconds, 24/7.
| Resolution | Average bitrate | Motion | Storage per camera (30 days) |
|---|---|---|---|
| 720p | 1024 kbit | 50% | 250 GB |
| 1080p | 2048 kbit | 50% | 400 GB |
| 4K | 4096 kbit | 50% | 750 GB |
| 4K | 4096 kbit | 100% | 1.4 TB |
Tip: Multiply the figure by the camera count. For local and cloud storage capacity planning, see TetherBox Recording Capacity.
Size recording capacity and annual drive workload separately. Shorter retention reduces required capacity, but does not reduce how much data is written each year. Allow for snapshots, filesystem overhead and free space; do not count a separate boot SSD towards recording retention. See Capacity and Annual Workload.
Recommended Storage Hardware
Continuous 24/7 recording requires drives with sustained-write endurance: verified CMR NAS, surveillance or enterprise HDDs, or high-TBW SSDs. Choose the exact model and capacity from the professional drive ranges and its SSD examples; a family name alone does not establish suitability. Never use USB memory sticks, SD cards or SMR drives for continuous recording. Connect external drives using USB 3.0 or better.
Above about 50 cameras, plan for multiple drives and consider RAID5 or RAID6 to spread the load and add redundancy. Camera count alone is not a universal single-drive limit: check bitrate, mixed I/O, annual workload and retention. See Drive Type by Camera Count.
For drive selection by camera count, endurance comparisons, USB enclosure guidance, and storage performance troubleshooting, see Storage Hardware Recommendations.
Graphics / GPU
Standard deployments do not need a dedicated GPU: TetherBox uses onboard graphics or the CPU for processing. It automatically uses available Nvidia, AMD, or Intel dedicated graphics for hardware acceleration.
Out of the box, TetherBox uses the analytics built into your Cameras or recorder. For on-device TetherX AI analytics, add one of the following accelerators:
| Accelerator | TetherX AI capacity | Suitable for |
|---|---|---|
| Hailo-8L | Up to 4 cameras | Mini PC or Raspberry Pi 5. See Hailo AI Accelerators |
| Hailo-8 or Hailo-10H | Up to 8 cameras | Mini PC, Raspberry Pi 5, or rackmount TetherBox. See Hailo AI Accelerators |
| NVIDIA Jetson Orin Nano 8GB | Up to 12 cameras | Compact, low-power installations, for example the fanless MSI MS-C9ZA (opens in a new tab) |
| NVIDIA GeForce RTX 5060 | Up to 16 cameras | Compact desktop, tower or rackmount TetherBox with a suitable PCIe slot, PSU and cooling |
| NVIDIA L4 | Up to 64 cameras | Rackmount server with a low-profile PCIe slot |
Note: NVIDIA capacities assume AI analytics run at 5 megapixels. Hailo capacities assume AI analytics run at 1 megapixel, such as a 720p sub-stream. For custom requirements, contact TetherX Support.
Next Steps
- Build a TetherBox on Your Own Hardware for operating-system installation, partitioning, and configuration.
- BIOS Settings for the settings an Intel or AMD TetherBox needs to run unattended, including the hardware watchdog.
- TetherBox Recording Capacity for detailed retention planning.
- Cloud Backup for off-site protection against theft, damage, or local disasters.
References
- CPU Monkey (opens in a new tab) - compare CPU performance across generations
- Geekbench Browser (opens in a new tab) - search specific CPU benchmarks
Referenced in:
- People Counting
- Frequently Asked Questions
- Build a TetherBox on Your Own Hardware
- Marketing PDFs
- Pre-Flight Checklist
- Quote Tool
- TetherBox Deployment Guide
- BIOS Settings
- CPU Recommendations
- External Storage Required
- Hailo AI Accelerators
- Raspberry Pi TetherBox
- Replacing a TetherBox
- Replacing a TetherBox Drive
- Storage Hardware Recommendations
- Talkdown (Push to Talk)
- TetherBox AI Facial Recognition
- TetherBox Recording Capacity
- TetherBox Troubleshooting
- TetherX AI
- Where to Run the TetherBox Software