# Benchmarks

## RTX 4090 24 GB, Intel Core i7-13700

Hardware specification: 

* Intel Core i7-13700
* G.SKILL Flare X5 64 GB RAM (2x32)
* MSI MAG B760 TOMAHAWK WIFI DDR5 Mainboard, Intel
* Asus GeForce RTX 4090 24 GB
* 2 TB Samsung SSD 980 PRO, PCIe 4.0 NVMe M.2

### 1 - 8 MP Images (GPU)

Processing time of **100 images** using GPU. Includes software ramp-up time (end-to-end). 

<table><thead><tr><th width="141">Image Resolution</th><th width="212.66668701171875">Processing Time (100 images)</th><th width="197">Memory Utilization</th><th>Comparison with Celantur SDK</th></tr></thead><tbody><tr><td>1 MP (1 Tile)</td><td><strong>12 s</strong></td><td><p><strong>2.6 GB VRAM</strong></p><p><strong>4 GB RAM</strong></p></td><td>4 s<br>686 MB VRAM<br>&#x3C; 500 MB RAM</td></tr><tr><td>2 MP (1 Tile)</td><td><strong>17 s</strong></td><td><p><strong>2.7 GB VRAM</strong></p><p><strong>4.5 GB RAM</strong></p></td><td><p>4.6 s </p><p>686 MB VRAM</p><p>&#x3C;500 MB RAM</p></td></tr><tr><td>8 MP (2 Tiles)</td><td><strong>91 s</strong> </td><td><p><strong>10 GB VRAM</strong></p><p><strong>5 GB RAM</strong></p></td><td><p>10s </p><p>686 MB VRAM</p><p>&#x3C;500 MB RAM</p></td></tr></tbody></table>

### Images (CPU)

Processing time of **100 images** using CPU only. Includes software ramp-up time (end-to-end).&#x20;

<table><thead><tr><th width="161">Image Resolution</th><th width="222">Processing Time (100 images)</th><th width="152">Memory Utilization</th><th>Comparison with Celantur SDK</th></tr></thead><tbody><tr><td>1 MP (1 Tile)</td><td><strong>40 s</strong></td><td><strong>&#x3C; 1.5GB RAM</strong></td><td><p>15 s</p><p>&#x3C; 800 MB RAM</p></td></tr><tr><td>2 MP (1 Tile)</td><td><strong>44 s</strong></td><td><strong>&#x3C; 1.5 GB RAM</strong></td><td><p>15 s</p><p>&#x3C; 900 MB RAM</p></td></tr><tr><td>8 MP (2 Tiles)</td><td><strong>201 s</strong> </td><td><strong>&#x3C; 3 GB RAM</strong></td><td><p>30 s</p><p>&#x3C; 1.4 GB RAM</p></td></tr></tbody></table>

### 32 MP Images (GPU)

Processing time of **1 image** using GPU. Includes software ramp-up time (end-to-end). Celantur Container Version 25.04.2.

<table><thead><tr><th width="181">Image Resolution</th><th>Processing Time (1 image)</th><th>Memory Utilization</th></tr></thead><tbody><tr><td>32 MP (8000x4000)</td><td><strong>1.04 s</strong></td><td><p>up to <strong>11.8 GB VRAM</strong></p><p><strong>5 GB RAM</strong></p></td></tr></tbody></table>

{% code title="Command" overflow="wrap" %}

```bash
-a person -a face -a license-plate --format pano:8000
```

{% endcode %}

### 72 MP Images (GPU)

Processing time of **1 image** by **one instance** of Celantur Container using GPU. Includes software ramp-up time (end-to-end). Celantur Container Version 25.09.1.

<table><thead><tr><th width="181">Image Resolution</th><th>Processing Time (1 image)</th><th>Memory Utilization</th></tr></thead><tbody><tr><td>72 MP</td><td><strong>0.27 s</strong></td><td><p>up to <strong>4 GB VRAM</strong></p><p><strong>4.5 GB RAM</strong></p></td></tr></tbody></table>

Processing time and throughput with **8 instances of Celantur Container** using GPU. Includes software ramp-up time (end-to-end). Celantur Container v25.11.1.

<table><thead><tr><th width="181">Image Resolution</th><th>Processing Time per Image</th><th>Throughput</th></tr></thead><tbody><tr><td>72 MP</td><td><strong>0.13 s</strong></td><td>7.5 images / s</td></tr></tbody></table>

{% code title="Command" overflow="wrap" %}

```bash
for i in $(seq 1 8); do PROCESS_DIR=$HOME/test$i ./celantur.sh -a face -a license-plate --format '{"number":[4,1],"overlap":[300,0],"section":[0,1870,12288,4950]}' & ; done
```

{% endcode %}

### 91 MP Images (GPU)

Processing time of **1 image** (Mosaic X panorama) using GPU. Includes software ramp-up time (end-to-end). Celantur Container Version 25.04.2.

<table><thead><tr><th width="181">Image Resolution</th><th>Processing Time (1 image)</th><th>Memory Utilization</th></tr></thead><tbody><tr><td>91 MP (13504x6752)</td><td><strong>2.9 s</strong></td><td><p>up to <strong>15 GB VRAM</strong></p><p><strong>6 GB RAM</strong></p></td></tr></tbody></table>

{% code title="Command" overflow="wrap" %}

```bash
--model segmentatio-omnium -a face -a license-plate --license-plate-threshold 0.1 --face-threshold 0.1 -f '{"number":[5,1],"overlap":[400,0],"section":[0,2200,13504,5300]}'
```

{% endcode %}

### FullHD Video (GPU)

Processing time of **1 video** (GoPro/mp4) using GPU. Includes software ramp-up time (end-to-end). Celantur Container v25.11.1.

<table><thead><tr><th width="181">Video Resolution / Framerate</th><th>Processing Time </th><th>Video Duration</th></tr></thead><tbody><tr><td>FullHD / 30 fps</td><td>7 m 6 s</td><td>7 m</td></tr></tbody></table>

{% code title="Command" overflow="wrap" %}

```bash
--video --keep-bit-rate -a face -a license-plate
```

{% endcode %}


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://doc.celantur.com/container/benchmarks.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.