Downloads · 30 days
0
sd-speed-list
mrpopsalot/sd-speed-list
sd-speed-list is a text-to-image model from mrpopsalot. Use it when you need an image from a text prompt. The card lists the license as gfdl.
<style .main-div { background-color: bb99bb; padding: 20px; border-radius: 10px; margin-bottom: 20px; } h1 { color: 24292e; margin-top: 0; font-size: 25px; } h2 { color: 24292e; margin-top: 0; font-size: 22px; } h3 {…
Access
Public
Updated Nov 30, 2025
Repo size
4.9 MB
Likes
1
Public
Repo makeup
Click a slice to open those files.
.png1.5 MB · 95%
At a glance
- Task
- Text-to-Image
- License
- gfdl
- Access
- Public
- Created
- Jun 26, 2025
- Updated
- Nov 30, 2025
- SHA
- 7f31fb3a
Model card
From the Hugging Face model README
<style>
.main-div {
background-color: #bb99bb;
padding: 20px;
border-radius: 10px;
margin-bottom: 20px;
}
h1 {
color: #24292e;
margin-top: 0;
font-size: 25px;
}
h2 {
color: #24292e;
margin-top: 0;
font-size: 22px;
}
h3 {
color: #24292e;
margin-top: 0;
font-size: 20px;
}
h4 {
color: #24292e;
margin-top: 0;
font-size: 15px;
}
p {
color:#24292e;
}
a {
color: #0366d6;
text-decoration: none;
}
.section-div {
background-color: white;
padding: 15px;
border-radius: 8px;
margin: 15px 0;
box-shadow: 0 2px 4px rgba(0,0,0,0.1);
}
.subsection-div {
background-color: #bbb;
padding: 15px;
border-radius: 8px;
margin: 15px 0;
box-shadow: 0 2px 4px rgba(0,0,0,0.1);
}
ul {
margin-bottom: 0;
color: #24292e;
}
td {
margin-bottom: 0;
color: #24292e;
}
code {
background-color: #333;
padding: 3px 6px;
border-radius: 4px;
}
table {
color: #24292e;
}
</style>
<div class="main-div">
<h1>Pops' Stable Diffusion Speed List</h1>
<div class="section-div">
<img src="./sd-speeds-v003.png">
<p>A hand curated list of generation speeds for various hardware and models.</p>
<p><b>LAST UPDATED:</b> 2025.11.29</p>
<p><b>Use the ComfyUI workflow above to start testing.</b></p>
</div>
<div class="section-div">
<h1>Contributor List</h1>
<ul>
<li><b>Mr Pops Alot (your's truely):</b> RTX 3090, GTX 1080 Ti, GTX 980, GTX 1050 Ti, GTX 1650M, MX 330, Pro W5500, Pro Wx 4100, HD 7790, HD 7750, RG 512SP Mobile, 5800X, 10700K, 1165G7, Q9300, T9300, 4790K, 6300U, 3770, 4300U, 5257U, SD8 G1</li>
<li><b>Lopi99:</b> PRO 6000 B, RTX 5060 Ti 16GB, A4000, 4060 Ti 16GB, 3060 Ti, EPYC 9654</li>
<li><b>Hugs288:</b> RTX 4090</li>
<li><b>Disty:</b> RX 7900 XTX, A770, 5800X3D, Dimensity 1080</li>
<li><b>Panchovix:</b> 5090 (Fedora 42)</li>
</ul>
</div>
<div class="section-div">
<h1>Methodology</h1>
<p>All GPUs tested use the following settings:</p>
<ul>
<li>Euler sampler</li>
<li>Normal scheduler</li>
<li>CFG 8</li>
<li>Square aspect ratio (1:1)</li>
<li>Steps can be changed. 20 is fine for slow systems but for systems with high end modern GPUs you may have to increase the step count to 100 or more for accurate readings.</li>
<li>OS if known will be stated. The OS used for initial testing is Arch Linux (Unstated OSes in submissions will be considered as "Unknown"). DirectML stuff is all done on Windows 11 22H2</li>
<li>Attention is default (unset), unless stated otherwise.</li>
<li>The workflow used is the provided workflow in the repo.</li>
<li>App versions are listed. If no version is listed then the commit version will be "Unknown".</li>
</li>
<p>Raw gen times are not recorded due to variance due to steps being variable. Instead iterations per second (and the inverse of it) are given since they are independent of steps.</p>
<p>The given speed value (it/s or s/it) is used, and then extrapolated using the formula <code>1/speed</code> to get the other value. If its under 0.1 then it will be expanded to four digits compared to the usual 2</p>
<p>If you can contribute to the list, do so as well. Lets make the most comprehensive, curated list of local Image Gen speeds!</p>
<div class="subsection-div">
<h3>Models Used</h3>
<p>The following are the models used for testing. The models you use can be the same architecture as the tested models</p>
<ul>
<li>SD1.5: <a href="https://huggingface.co/jzli/Hassaku-1.3">jzli/Hassaku-1.3</a></li>
<li>SDXL: <a href="https://huggingface.co/jzli/Hassaku-1.3">OnomaAIResearch/Illustrious-XL-v2.0</a></li>
<li>Lumina 2: <a href="https://huggingface.co/neta-art/NetaLumina_Alpha">neta-art/NetaLumina_Alpha</a> Round NNNN EP6 S127716</li>
<li>Z-Image: <a href="https://huggingface.co/Tongyi-MAI/Z-Image-Turbo">Tongyi-MAI/Z-Image-Turbo</a></li>
</li>
</div>
</div>
<div class="section-div">
<h1>Benchmarks</h1>
<div class="subsection-div">
<h2>Z-Image</h2>
<h3>1536px</h3>
<table>
<thead>
<th>Chip</th>
<th>it/s</th>
<th>s/it</th>
<th>Backend</th>
<th>App (Commit)</th>
<th>OS</th>
<th>Notes</th>
</thead>
<tbody>
<tr>
<td>NVIDIA RTX 3090</td>
<td>0.20it/s</td>
<td>4.90s/it</td>
<td>CUDA 12.6</td>
<td>ComfyUI (5151cff)</td>
<td>Arch Linux</td>
<td></td>
</tr>
</tbody>
</table>
<h3>1024px</h3>
<table>
<thead>
<th>Chip</th>
<th>it/s</th>
<th>s/it</th>
<th>Backend</th>
<th>App (Commit)</th>
<th>OS</th>
<th>Notes</th>
</thead>
<tbody>
<tr>
<td>NVIDIA RTX 3090</td>
<td>0.50it/s</td>
<td>2.01s/it</td>
<td>CUDA 12.6</td>
<td>ComfyUI (5151cff)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>NVIDIA GTX 980</td>
<td>0.0226it/s</td>
<td>44.26s/it</td>
<td>CUDA 12.6</td>
<td>ComfyUI (5151cff)</td>
<td>Arch Linux</td>
<td>FP8 DiT, CPU TE, PCIe 2.0x1</td>
</tr>
<tr>
<td>NVIDIA GTX 980</td>
<td>0.0215it/s</td>
<td>46.51s/it</td>
<td>CUDA 12.6</td>
<td>ComfyUI (5151cff)</td>
<td>Arch Linux</td>
<td>Q3_K_S C96 GGUF DiT, CPU TE, PCIe 2.0x1</td>
</tr>
<tr>
<td>NVIDIA GTX 1650 Mobile</td>
<td>0.0147it/s</td>
<td>67.85s/it</td>
<td>CUDA 12.8</td>
<td>ComfyUI (5151cff)</td>
<td>Arch Linux</td>
<td>FP8 DiT and TE</td>
</tr>
<tr>
<td>NVIDIA GTX 1050 Ti</td>
<td>0.0115it/s</td>
<td>86.22s/it</td>
<td>CUDA 12.6</td>
<td>ComfyUI (5151cff)</td>
<td>Arch Linux</td>
<td>Q3_K_S C96 GGUF DiT, CPU TE, PCIe 2.0x1</td>
</tr>
</tbody>
</table>
<h3>512px</h3>
<table>
<thead>
<th>Chip</th>
<th>it/s</th>
<th>s/it</th>
<th>Backend</th>
<th>App (Commit)</th>
<th>OS</th>
<th>Notes</th>
</thead>
<tbody>
<tr>
<td>NVIDIA RTX 3090</td>
<td>1.76it/s</td>
<td>2.01s/it</td>
<td>CUDA 12.6</td>
<td>ComfyUI (5151cff)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>NVIDIA GTX 980</td>
<td>0.082it/s</td>
<td>12.24s/it</td>
<td>CUDA 12.6</td>
<td>ComfyUI (5151cff)</td>
<td>Arch Linux</td>
<td>Q3_K_S C96 GGUF DiT, CPU TE, PCIe 2.0x1</td>
</tr>
<tr>
<td>NVIDIA GTX 1650 Mobile</td>
<td>0.0536t/s</td>
<td>18.69s/it</td>
<td>CUDA 12.8</td>
<td>ComfyUI (5151cff)</td>
<td>Arch Linux</td>
<td>FP8 DiT and TE</td>
</tr>
<tr>
<td>NVIDIA GTX 980</td>
<td>0.0474it/s</td>
<td>21.06s/it</td>
<td>CUDA 12.6</td>
<td>ComfyUI (5151cff)</td>
<td>Arch Linux</td>
<td>FP8 DiT, CPU TE, PCIe 2.0x1</td>
</tr>
<tr>
<td>NVIDIA GTX 1050 Ti</td>
<td>0.0457it/s</td>
<td>21.87s/it</td>
<td>CUDA 12.6</td>
<td>ComfyUI (5151cff)</td>
<td>Arch Linux</td>
<td>Q3_K_S C96 GGUF DiT, CPU TE, PCIe 2.0x1</td>
</tr>
</tbody>
</table>
<h3>256px</h3>
<table>
<thead>
<th>Chip</th>
<th>it/s</th>
<th>s/it</th>
<th>Backend</th>
<th>App (Commit)</th>
<th>OS</th>
<th>Notes</th>
</thead>
<tbody>
<tr>
<td>NVIDIA RTX 3090</td>
<td>4.06it/s</td>
<td>0.24s/it</td>
<td>CUDA 12.6</td>
<td>ComfyUI (5151cff)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>NVIDIA GTX 1650 Mobile</td>
<td>0.16it/s</td>
<td>6.02s/it</td>
<td>CUDA 12.8</td>
<td>ComfyUI (5151cff)</td>
<td>Arch Linux</td>
<td>FP8 DiT and TE</td>
</tr>
<tr>
<td>NVIDIA GTX 980</td>
<td>0.15it/s</td>
<td>6.71s/it</td>
<td>CUDA 12.6</td>
<td>ComfyUI (5151cff)</td>
<td>Arch Linux</td>
<td>Q3_K_S C96 GGUF DiT, CPU TE, PCIe 2.0x1</td>
</tr>
<tr>
<td>NVIDIA GTX 1050 Ti</td>
<td>0.0935it/s</td>
<td>10.69s/it</td>
<td>CUDA 12.6</td>
<td>ComfyUI (5151cff)</td>
<td>Arch Linux</td>
<td>Q3_K_S C96 GGUF DiT, CPU TE, PCIe 2.0x1</td>
</tr>
<tr>
<td>NVIDIA GTX 980</td>
<td>0.0523it/s</td>
<td>19.11s/it</td>
<td>CUDA 12.6</td>
<td>ComfyUI (5151cff)</td>
<td>Arch Linux</td>
<td>FP8 DiT, CPU TE, PCIe 2.0x1</td>
</tr>
</tbody>
</table>
</div>
<div class="subsection-div">
<h2>Lumina 2</h2>
<h3>1536px</h3>
<table>
<thead>
<th>Chip</th>
<th>it/s</th>
<th>s/it</th>
<th>Backend</th>
<th>App (Commit)</th>
<th>OS</th>
<th>Notes</th>
</thead>
<tbody>
<tr>
<td>NVIDIA RTX PRO 6000 B</td>
<td>1.88it/s</td>
<td>0.53s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 5090</td>
<td>1.31it/s</td>
<td>0.76s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (37d620a)</td>
<td>Fedora 42</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 4090</td>
<td>1.29it/s</td>
<td>0.78s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Windows 11 24H2</td>
</tr>
<tr>
<td>NVIDIA RTX 3090</td>
<td>0.41it/s</td>
<td>2.40s/it</td>
<td>CUDA 12.6</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>AMD RX 7900 XTX</td>
<td>0.34it/s</td>
<td>2.98s/it</td>
<td>ROCm 6.4.1</td>
<td>SDNext (aa0652ca)</td>
<td>Arch Linux</td>
<td>INT8 Matmul, Flash Attention, OC</td>
</tr>
<tr>
<td>AMD RX 7900 XTX</td>
<td>0.33it/s</td>
<td>3.06s/it</td>
<td>ROCm 6.4.1</td>
<td>SDNext (aa0652ca)</td>
<td>Arch Linux</td>
<td>Flash Attention, OC</td>
</tr>
<tr>
<td>NVIDIA RTX A4000</td>
<td>0.32it/s</td>
<td>3.11s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 5060 Ti</td>
<td>0.26it/s</td>
<td>3.83s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 3070</td>
<td>0.25it/s</td>
<td>3.95s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Unknown</td>
<td>CPU TE</td>
</tr>
<tr>
<td>NVIDIA RTX 3060 Ti</td>
<td>0.22it/s</td>
<td>4.57s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td>CPU TE</td>
</tr>
<tr>
<td>NVIDIA RTX 3060</td>
<td>0.18it/s</td>
<td>5.47s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Unknown</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 3060</td>
<td>0.18it/s</td>
<td>5.52s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Unknown</td>
<td>CPU TE</td>
</tr>
<tr>
<td>Intel A770</td>
<td>0.11it/s</td>
<td>9.38s/it</td>
<td>PyTorch XPU 2.7.1</td>
<td>SDNext (a9c65c0e)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>NVIDIA GTX 1080 Ti</td>
<td>0.0421it/s</td>
<td>23.77s/it</td>
<td>CUDA 12.6</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
</tbody>
</table>
<h3>1024px</h3>
<table>
<thead>
<th>Chip</th>
<th>it/s</th>
<th>s/it</th>
<th>Backend</th>
<th>App (Commit)</th>
<th>OS</th>
<th>Notes</th>
</thead>
<tbody>
<tr>
<td>NVIDIA RTX PRO 6000 B</td>
<td>5.12it/s</td>
<td>0.20s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 5090</td>
<td>3.70it/s</td>
<td>0.27s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (37d620a)</td>
<td>Fedora 42</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 5090</td>
<td>2.58it/s</td>
<td>0.39s/it</td>
<td>CUDA 12.8</td>
<td>ComfyUI (Unknown)</td>
<td>Windows 11 24H2</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 4090</td>
<td>2.22it/s</td>
<td>0.45s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Windows 11 24H2</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 3090</td>
<td>1.14it/s</td>
<td>0.87s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td>Sage Attention</td>
</tr>
<tr>
<td>AMD RX 7900 XTX</td>
<td>1.08it/s</td>
<td>0.96s/it</td>
<td>ROCm 6.4.1</td>
<td>SDNext (aa0652ca)</td>
<td>Arch Linux</td>
<td>INT8 Matmul, Flash Attention, OC</td>
</tr>
<tr>
<td>AMD RX 7900 XTX</td>
<td>1.04it/s</td>
<td>0.96s/it</td>
<td>ROCm 6.4.1</td>
<td>SDNext (aa0652ca)</td>
<td>Arch Linux</td>
<td>Flash Attention, OC</td>
</tr>
<tr>
<td>NVIDIA RTX 3090</td>
<td>1.00it/s</td>
<td>1.00s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX A4000</td>
<td>0.82it/s</td>
<td>1.22s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 5060 Ti</td>
<td>0.68it/s</td>
<td>1.47s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 3070</td>
<td>0.66it/s</td>
<td>1.52s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Unknown</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 3060 Ti</td>
<td>0.58it/s</td>
<td>1.72s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 3060</td>
<td>0.44it/s</td>
<td>2.25s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Unknown</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 3060</td>
<td>0.44it/s</td>
<td>2.28s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Unknown</td>
<td>CPU TE</td>
</tr>
<tr>
<td>Intel A770</td>
<td>0.24it/s</td>
<td>4.24s/it</td>
<td>PyTorch XPU 2.7.1</td>
<td>SDNext (a9c65c0e)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>NVIDIA GTX 1080 Ti</td>
<td>0.15it/s</td>
<td>6.77s/it</td>
<td>CUDA 12.6</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>NVIDIA GTX 980</td>
<td>0.0599it/s</td>
<td>16.69s/it</td>
<td>CUDA 12.4</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td>FP32 CPU TE</td>
</tr>
<tr>
<td>AMD Ryzen 5800X</td>
<td>0.0102it/s</td>
<td>97.86s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>Intel i7-10700K</td>
<td>0.0090it/s</td>
<td>107.25s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
</tr>
</tbody>
</table>
<h3>512px</h3>
<table>
<tr>
<th>Chip</th>
<th>it/s</th>
<th>s/it</th>
<th>Backend</th>
<th>App (Commit)</th>
<th>OS</th>
<th>Notes</th>
</tr>
<tr>
<td>NVIDIA RTX PRO 6000 B</td>
<td>17.43it/s</td>
<td>0.0574s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 5090</td>
<td>12.17it/s</td>
<td>0.0821s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (37d620a)</td>
<td>Fedora 42</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 5090</td>
<td>8.85it/s</td>
<td>0.11s/it</td>
<td>CUDA 12.8</td>
<td>ComfyUI (Unknown)</td>
<td>Windows 11 24H2</td>
</tr>
<tr>
<td>NVIDIA RTX 5090</td>
<td>8.04it/s</td>
<td>0.12s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Windows 11 24H2</td>
</tr>
<tr>
<td>AMD RX 7900 XTX</td>
<td>4.65it/s</td>
<td>0.22s/it</td>
<td>ROCm 6.4.1</td>
<td>SDNext (aa0652ca)</td>
<td>Arch Linux</td>
<td>INT8 Matmul, Flash Attention, OC</td>
</tr>
<tr>
<td>NVIDIA RTX 3090</td>
<td>4.35it/s</td>
<td>0.23s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td>Sage Attention</td>
</tr>
<tr>
<td>AMD RX 7900 XTX</td>
<td>4.18it/s</td>
<td>0.24s/it</td>
<td>ROCm 6.4.1</td>
<td>SDNext (aa0652ca)</td>
<td>Arch Linux</td>
<td>Flash Attention, OC</td>
</tr>
<tr>
<td>NVIDIA RTX A4000</td>
<td>3.29it/s</td>
<td>0.30s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Unknown</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 5060 Ti</td>
<td>3.22it/s</td>
<td>0.31s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 3070</td>
<td>2.62it/s</td>
<td>0.38s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Unknown</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 3060 Ti</td>
<td>2.28it/s</td>
<td>0.44s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 3060</td>
<td>1.65it/s</td>
<td>0.61s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Unknown</td>
<td>CPU TE</td>
</tr>
<tr>
<td>NVIDIA RTX 3060</td>
<td>1.44it/s</td>
<td>0.69s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Unknown</td>
<td></td>
</tr>
<tr>
<td>NVIDIA GTX 1080 Ti</td>
<td>0.73it/s</td>
<td>1.37s/it</td>
<td>CUDA 12.6</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>Intel A770</td>
<td>0.58it/s</td>
<td>1.75s/it</td>
<td>PyTorch XPU 2.7.1</td>
<td>SDNext (a9c65c0e)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>NVIDIA GTX 980</td>
<td>0.28it/s</td>
<td>3.57s/it</td>
<td>CUDA 12.4</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td>FP8 CPU TE</td>
</tr>
<tr>
<td>NVIDIA GTX 980</td>
<td>0.25it/s</td>
<td>3.99s/it</td>
<td>CUDA 12.4</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td>FP32 CPU TE</td>
</tr>
<tr>
<td>AMD Ryzen 5800X</td>
<td>0.0649it/s</td>
<td>15.42s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>Intel i7-10700K</td>
<td>0.0513it/s</td>
<td>19.49s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
</table>
<h3>256px</h3>
<table>
<tr>
<th>Chip</th>
<th>it/s</th>
<th>s/it</th>
<th>Backend</th>
<th>App (Commit)</th>
<th>OS</th>
<th>Notes</th>
</tr>
<tr>
<td>NVIDIA RTX PRO 6000 B</td>
<td>18.78it/s</td>
<td>0.0532s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 5090</td>
<td>18.29it/s</td>
<td>0.0547s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (37d620a)</td>
<td>Fedora 42</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 4090</td>
<td>14.92it/s</td>
<td>0.067s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Windows 11 24H2</td>
</tr>
<tr>
<td>NVIDIA RTX 5090</td>
<td>13.42it/s</td>
<td>0.0745s/it</td>
<td>CUDA 12.8</td>
<td>ComfyUI (Unknown)</td>
<td>Windows 11 24H2</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 3090</td>
<td>11.37it/s</td>
<td>0.0880s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td>Sage Attention</td>
</tr>
<tr>
<td>NVIDIA RTX A4000</td>
<td>10.22it/s</td>
<td>0.10s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>AMD RX 7900 XTX</td>
<td>9.71it/s</td>
<td>0.10s/it</td>
<td>ROCm 6.4.1</td>
<td>SDNext (aa0652ca)</td>
<td>Arch Linux</td>
<td>Flash Attention, OC</td>
</tr>
<tr>
<td>NVIDIA RTX 3070</td>
<td>9.14it/s</td>
<td>0.11s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Unknown</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 5060 Ti</td>
<td>8.76it/s</td>
<td>0.11s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>AMD RX 7900 XTX</td>
<td>7.93it/s</td>
<td>0.13s/it</td>
<td>ROCm 6.4.1</td>
<td>SDNext (aa0652ca)</td>
<td>Arch Linux</td>
<td>INT8 Matmul, Flash Attention, OC</td>
</tr>
<tr>
<td>NVIDIA RTX 3060 Ti</td>
<td>7.21it/s</td>
<td>0.14s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>NVIDIA GTX 1080 Ti</td>
<td>1.74it/s</td>
<td>0.57s/it</td>
<td>CUDA 12.6</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>NVIDIA GTX 980</td>
<td>0.78it/s</td>
<td>1.27s/it</td>
<td>CUDA 12.4</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td>FP8 CPU TE</td>
</tr>
<tr>
<td>Intel A770</td>
<td>0.71it/s</td>
<td>1.40s/it</td>
<td>PyTorch XPU 2.7.1</td>
<td>SDNext (a9c65c0e)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>NVIDIA GTX 980</td>
<td>0.59it/s</td>
<td>1.68s/it</td>
<td>CUDA 12.4</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td>FP32 CPU TE</td>
</tr>
<tr>
<td>Intel i7-10700K</td>
<td>0.26it/s</td>
<td>3.78s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>AMD Ryzen 5800X</td>
<td>0.25it/s</td>
<td>3.98s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
</table>
</div>
<div class="subsection-div">
<h2>SDXL</h2>
<h3>1536px</h3>
<table>
<tr>
<th>Chip</th>
<th>it/s</th>
<th>s/it</th>
<th>Backend</th>
<th>App (Commit)</th>
<th>OS</th>
<th>Notes</th>
</tr>
<tr>
<td>NVIDIA RTX PRO 6000 B</td>
<td>6.61it/s</td>
<td>0.15s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 5090</td>
<td>6.12it/s</td>
<td>0.16s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (37d620a)</td>
<td>Fedora 42</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 5090</td>
<td>3.38it/s</td>
<td>0.29s/it</td>
<td>CUDA 12.8</td>
<td>ComfyUI (Unknown)</td>
<td>Windows 11 24H2</td>
</tr>
<tr>
<td>NVIDIA RTX 4090</td>
<td>3.11it/s</td>
<td>0.32s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Windows 11 24H2</td>
</tr>
<tr>
<td>AMD RX 7900 XTX</td>
<td>1.96it/s</td>
<td>0.51s/it</td>
<td>ROCm 6.4.1</td>
<td>SDNext (aa0652ca)</td>
<td>Arch Linux</td>
<td>Flash Attention, OC</td>
</tr>
<tr>
<td>AMD RX 7900 XTX</td>
<td>1.93it/s</td>
<td>0.52s/it</td>
<td>ROCm 6.4.1</td>
<td>SDNext (aa0652ca)</td>
<td>Arch Linux</td>
<td>Flash Attention, OC</td>
</tr>
<tr>
<td>NVIDIA RTX 3090</td>
<td>1.63it/s</td>
<td>0.61s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX A4000</td>
<td>1.16it/s</td>
<td>0.86s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>Intel A770</td>
<td>1.12it/s</td>
<td>0.89s/it</td>
<td>OpenVINO 2025.2</td>
<td>SDNext (a9c65c0e)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>NVIDIA RTX 5060 Ti</td>
<td>1.10it/s</td>
<td>0.91s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 3070</td>
<td>1.04it/s</td>
<td>0.96s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Unknown</td>
<td></td>
</tr>
<tr>
<td>Intel A770</td>
<td>1.02it/s</td>
<td>0.98s/it</td>
<td>PyTorch XPU 2.7.1</td>
<td>SDNext (a9c65c0e)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>NVIDIA RTX 3060 Ti</td>
<td>0.85it/s</td>
<td>1.18s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 3060</td>
<td>0.67it/s</td>
<td>1.49s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Unknown</td>
<td></td>
</tr>
<tr>
<td>NVIDIA GTX 1080 Ti</td>
<td>0.15it/s</td>
<td>6.43s/it</td>
<td>CUDA 12.6</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
</table>
<h3>1024px</h3>
<p>Runs on 2GB of VRAM with tiled VAE. 4GB can run the regular variant</p>
<table>
<tr>
<th>Chip</th>
<th>it/s</th>
<th>s/it</th>
<th>Backend</th>
<th>App (Commit)</th>
<th>OS</th>
<th>Notes</th>
</tr>
<tr>
<td>NVIDIA RTX PRO 6000 B</td>
<td>15.02it/s</td>
<td>0.0665s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 5090</td>
<td>14.15it/s</td>
<td>0.0707s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (37d620a)</td>
<td>Fedora 42</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 5090</td>
<td>8.95it/s</td>
<td>0.11s/it</td>
<td>CUDA 12.8</td>
<td>ComfyUI (Unknown)</td>
<td>Windows 11 24H2</td>
</tr>
<tr>
<td>NVIDIA RTX 4090</td>
<td>7.00it/s</td>
<td>0.14s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Windows 11 24H2</td>
</tr>
<tr>
<td>AMD RX 7900 XTX</td>
<td>4.92it/s</td>
<td>0.20s/it</td>
<td>ROCm 6.4.1</td>
<td>SDNext (aa0652ca)</td>
<td>Arch Linux</td>
<td>INT8 Matmul, Flash Attention, OC</td>
</tr>
<tr>
<td>AMD RX 7900 XTX</td>
<td>4.80it/s</td>
<td>0.21s/it</td>
<td>ROCm 6.4.1</td>
<td>SDNext (aa0652ca)</td>
<td>Arch Linux</td>
<td>Flash Attention, OC</td>
</tr>
<tr>
<td>NVIDIA RTX 3090</td>
<td>4.00it/s</td>
<td>0.25s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX A4000</td>
<td>2.81it/s</td>
<td>0.36s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 5060 Ti</td>
<td>2.60it/s</td>
<td>0.39s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>Intel A770</td>
<td>2.40it/s</td>
<td>0.42s/it</td>
<td>OpenVINO 2025.2</td>
<td>SDNext (a9c65c0e)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>NVIDIA RTX 3070</td>
<td>2.34it/s</td>
<td>0.43s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Unknown</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 3060 Ti</td>
<td>1.96it/s</td>
<td>0.51s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>Intel A770</td>
<td>1.75it/s</td>
<td>0.56s/it</td>
<td>PyTorch XPU 2.7.1</td>
<td>SDNext (a9c65c0e)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>NVIDIA RTX 3060</td>
<td>1.49it/s</td>
<td>0.67s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Unknown</td>
<td></td>
</tr>
<tr>
<td>NVIDIA GTX 1080 Ti</td>
<td>0.31it/s</td>
<td>3.22s/it</td>
<td>CUDA 12.6</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>NVIDIA GTX 980</td>
<td>0.18it/s</td>
<td>5.35s/it</td>
<td>CUDA 12.4</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>AMD EPYC 9654</td>
<td>0.14it/s</td>
<td>7.13s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>AMD Pro W5500</td>
<td>0.13it/s</td>
<td>7.35s/it</td>
<td>Vulkan 1.3</td>
<td>KoboldCPP (1.95.1)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>NVIDIA GTX 1650 Mobile</td>
<td>0.13it/s</td>
<td>7.57s/it</td>
<td>CUDA 12.8</td>
<td>ComfyUI (5b12b55)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>NVIDIA GTX 1050 Ti</td>
<td>0.0833it/s</td>
<td>12.00s/it</td>
<td>CUDA 12.4</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>AMD Pro W5500</td>
<td>0.0699it/s</td>
<td>14.31s/it</td>
<td>DirectML</td>
<td>ComfyUI (Unknown)</td>
<td>Windows 11 22H2</td>
<td></td>
</tr>
<tr>
<td>AMD Ryzen 5800X</td>
<td>0.0365it/s</td>
<td>27.42s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>NVIDIA MX 330</td>
<td>0.0289it/s</td>
<td>34.60s/it</td>
<td>CUDA 12.4</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>Intel i7-10700K</td>
<td>0.0274it/s</td>
<td>36.44s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>AMD Pro WX 4100</td>
<td>0.0247it/s</td>
<td>40.50s/it</td>
<td>DirectML</td>
<td>ComfyUI (Unknown)</td>
<td>Windows 11 22H2</td>
<td></td>
</tr>
<tr>
<td>AMD Pro W5500</td>
<td>0.0147it/s</td>
<td>68.04s/it</td>
<td>Vulkan 1.3</td>
<td>KoboldCPP (1.95.1)</td>
<td>Windows 11 22H2</td>
</tr>
<tr>
<td>Intel i7-1165G7</td>
<td>0.0145it/s</td>
<td>69.08s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
</table>
<h3>512px</h3>
<table>
<tr>
<th>Chip</th>
<th>it/s</th>
<th>s/it</th>
<th>Backend</th>
<th>App (Commit)</th>
<th>OS</th>
<th>Notes</th>
</tr>
<tr>
<td>NVIDIA RTX PRO 6000 B</td>
<td>24.49it/s</td>
<td>0.0408s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 5090</td>
<td>23.42it/s</td>
<td>0.0427s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (37d620a)</td>
<td>Fedora 42</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 5090</td>
<td>21.52it/s</td>
<td>0.0465s/it</td>
<td>CUDA 12.8</td>
<td>ComfyUI (Unknown)</td>
<td>Windows 11 24H2</td>
</tr>
<tr>
<td>NVIDIA RTX 4090</td>
<td>18.5it/s</td>
<td>0.05s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Windows 11 24H2</td>
</tr>
<tr>
<td>NVIDIA RTX A4000</td>
<td>15.62it/s</td>
<td>0.06s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>AMD RX 7900 XTX</td>
<td>14.39it/s</td>
<td>0.07s/it</td>
<td>ROCm 6.4.1</td>
<td>SDNext (aa0652ca)</td>
<td>Arch Linux</td>
<td>Flash Attention, OC</td>
</tr>
<tr>
<td>NVIDIA RTX 3090</td>
<td>12.39it/s</td>
<td>0.0807s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX A4000</td>
<td>10.80it/s</td>
<td>0.09s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 3060 Ti</td>
<td>9.54it/s</td>
<td>0.10s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 3070</td>
<td>8.54it/s</td>
<td>0.12s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Unknown</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 5060 Ti</td>
<td>8.53it/s</td>
<td>0.12s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>Intel A770</td>
<td>7.20it/s</td>
<td>0.14s/it</td>
<td>OpenVINO 2025.2</td>
<td>SDNext (a9c65c0e)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>AMD RX 7900 XTX</td>
<td>6.60it/s</td>
<td>0.15s/it</td>
<td>ROCm 6.4.1</td>
<td>SDNext (aa0652ca)</td>
<td>Arch Linux</td>
<td>INT8 Matmul, Flash Attention, OC</td>
</tr>
<tr>
<td>NVIDIA RTX 3060</td>
<td>4.17it/s</td>
<td>0.24s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Unknown</td>
<td></td>
</tr>
<tr>
<td>Intel A770</td>
<td>1.70it/s</td>
<td>0.59s/it</td>
<td>Pytorch XPU 2.7.1</td>
<td>SDNext (a9c65c0e)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>NVIDIA GTX 1080 Ti</td>
<td>1.20it/s</td>
<td>0.83s/it</td>
<td>CUDA 12.6</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>AMD EPYC 9654</td>
<td>0.93it/s</td>
<td>1.08s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>NVIDIA GTX 980</td>
<td>0.69it/s</td>
<td>1.45s/it</td>
<td>CUDA 12.4</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>AMD Pro W5500</td>
<td>0.54it/s</td>
<td>1.85s/it</td>
<td>Vulkan 1.3</td>
<td>KoboldCPP (1.95.1)</td>
<td>Windows 11 22H2</td>
</tr>
<tr>
<td>NVIDIA GTX 1650 Mobile</td>
<td>0.54it/s</td>
<td>1.85s/it</td>
<td>CUDA 12.8</td>
<td>ComfyUI (5b12b55)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>AMD EPYC 9654</td>
<td>0.47it/s</td>
<td>2.12s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>AMD Pro W5500</td>
<td>0.42it/s</td>
<td>2.38s/it</td>
<td>DirectML</td>
<td>ComfyUI (Unknown)</td>
<td>Windows 11 22H2</td>
<td></td>
</tr>
<tr>
<td>NVIDIA GTX 1050 Ti</td>
<td>0.29it/s</td>
<td>3.40s/it</td>
<td>CUDA 12.4</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>AMD Pro W5500</td>
<td>0.20it/s</td>
<td>5.06s/it</td>
<td>Vulkan 1.3</td>
<td>KoboldCPP (1.95.1)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>AMD Ryzen 5800X</td>
<td>0.19it/s</td>
<td>5.32s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>Intel i7-10700K</td>
<td>0.17it/s</td>
<td>5.85s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>NVIDIA MX 330</td>
<td>0.11it/s</td>
<td>9.25s/it</td>
<td>CUDA 12.4</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>AMD HD 7790</td>
<td>0.11it/s</td>
<td>9.39s/it</td>
<td>DirectML</td>
<td>ComfyUI (Unknown)</td>
<td>Windows 11 22H2</td>
<td></td>
</tr>
<tr>
<td>AMD Pro WX 4100</td>
<td>0.1043it/s</td>
<td>9.59s/it</td>
<td>DirectML</td>
<td>ComfyUI (Unknown)</td>
<td>Windows 11 22H2</td>
<td></td>
</tr>
<tr>
<td>Intel i7-1165G7</td>
<td>0.0717it/s</td>
<td>13.94s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
</table>
</div>
<div class="subsection-div">
<h2>SD1.5</h2>
<h3>512px</h3>
<table>
<tr>
<th>Chip</th>
<th>it/s</th>
<th>s/it</th>
<th>Backend</th>
<th>App (Commit)</th>
<th>OS</th>
<th>Notes</th>
</tr>
<tr>
<td>NVIDIA RTX PRO 6000 B</td>
<td>51.83it/s</td>
<td>0.019s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 5090</td>
<td>46.93it/s</td>
<td>0.0213s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (37d620a)</td>
<td>Fedora 42</td>
<td></td>
</tr>
<tr>
<td>AMD RX 7900 XTX</td>
<td>26.69it/s</td>
<td>0.04s/it</td>
<td>ROCm 6.4.1</td>
<td>SDNext (aa0652ca)</td>
<td>Arch Linux</td>
<td>Flash Attention, OC</td>
</tr>
<tr>
<td>AMD RX 7900 XTX</td>
<td>22.75it/s</td>
<td>0.04s/it</td>
<td>ROCm 6.4.1</td>
<td>SDNext (aa0652ca)</td>
<td>Arch Linux</td>
<td>INT8 Matmul, Flash Attention, OC</td>
</tr>
<tr>
<td>NVIDIA RTX 3090</td>
<td>20.58it/s</td>
<td>0.0486s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 5060 Ti</td>
<td>14.21it/s</td>
<td>0.0704s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX A4000</td>
<td>15.62it/s</td>
<td>0.06s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>NVIDIA RTX 3070</td>
<td>13.78it/s</td>
<td>0.07s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Unknown</td>
<td></td>
</tr>
<tr>
<td>Intel A770</td>
<td>11.85it/s</td>
<td>0.08s/it</td>
<td>OpenVINO 2025.2</td>
<td>SDNext (a9c65c0e)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>NVIDIA RTX 3060 Ti</td>
<td>11.58it/s</td>
<td>0.09s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
<td></td>
</tr>
<tr>
<td>Intel A770</td>
<td>6.40it/s</td>
<td>0.15s/it</td>
<td>PyTorch XPU 2.7.1</td>
<td>SDNext (a9c65c0e)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>NVIDIA RTX 3060</td>
<td>4.17it/s</td>
<td>0.24s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Unknown</td>
<td></td>
</tr>
<tr>
<td>NVIDIA GTX 1080 Ti</td>
<td>3.20it/s</td>
<td>0.31s/it</td>
<td>CUDA 12.6</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>Qualcomm SD 8 Gen 1</td>
<td>2.35it/s</td>
<td>0.42s/it</td>
<td>QNN NPU 2.39</td>
<td>Local Dream (2.1.1)</td>
<td>OneUI 7.0 (A15 - 5.10.226)</td>
<td></td>
</tr>
<tr>
<td>NVIDIA GTX 980</td>
<td>1.59it/s</td>
<td>0.63it/s</td>
<td>CUDA 12.4</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>NVIDIA GTX 1650 Mobile</td>
<td>1.22it/s</td>
<td>0.82s/it</td>
<td>CUDA 12.8</td>
<td>ComfyUI (5b12b55)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>AMD Pro W5500</td>
<td>1.01it/s</td>
<td>0.99s/it</td>
<td>Vulkan 1.3</td>
<td>KoboldCPP (1.95.1)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>AMD Pro W5500</td>
<td>0.78it/s</td>
<td>1.27s/it</td>
<td>Vulkan 1.3</td>
<td>KoboldCPP (1.95.1)</td>
<td>Windows 11 22H2</td>
</tr>
<tr>
<td>AMD Pro W5500</td>
<td>0.75it/s</td>
<td>1.32s/it</td>
<td>DirectML</td>
<td>ComfyUI (Unknown)</td>
<td>Windows 11 22H2</td>
<td></td>
</tr>
<tr>
<td>NVIDIA GTX 1050 Ti</td>
<td>0.71it/s</td>
<td>1.40s/it</td>
<td>CUDA 12.4</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>AMD Ryzen 5800X3D</td>
<td>0.31it/s</td>
<td>3.20s/it</td>
<td>CPU</td>
<td>SDNext (a9c65c0e)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>NVIDIA MX 330</td>
<td>0.30it/s</td>
<td>3.37s/it</td>
<td>CUDA 12.4</td>
<td>ComfyUI (Unknown)</td>
<td>Atch Linux</td>
</tr>
<tr>
<td>AMD RG 512SP Mobile</td>
<td>0.26it/s</td>
<td>3.78s/it</td>
<td>Vulkan 1.3</td>
<td>KoboldCPP (1.100.1)</td>
<td>Arh Linux</td>
<td></td>
</tr>
<tr>
<td>AMD Pro WX 4100</td>
<td>0.24it/s</td>
<td>4.07s/it</td>
<td>DirectML</td>
<td>ComfyUI (Unknown)</td>
<td>Windows 11 22H2</td>
</tr>
<tr>
<td>AMD Pro WX 4100</td>
<td>0.22it/s</td>
<td>4.38s/it</td>
<td>Vulkan 1.3</td>
<td>KoboldCPP (1.95.1)</td>
<td>Windows 11 22H2</td>
</tr>
<tr>
<td>AMD Ryzen 5800X</td>
<td>0.22it/s</td>
<td>4.73s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>Intel i7-10700K</td>
<td>0.21it/s</td>
<td>4.87s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>Intel i7-1165G7</td>
<td>0.11it/s</td>
<td>8.97s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>Qualcomm SD 8 Gen 1</td>
<td>0.0714it/s</td>
<td>14.00s/it</td>
<td>MNN CPU</td>
<td>Local Dream (2.1.1)</td>
<td>OneUI 7.0 (A15 - 5.10.226)</td>
<td></td>
</tr>
<tr>
<td>AMD RX 550</td>
<td>0.0651it/s</td>
<td>15.35s/it</td>
<td>Vulkan 1.3</td>
<td>KoboldCPP (1.95.1)</td>
<td>Arch Linux</td>
<td>64 bit bus</td>
</tr>
<tr>
<td>Intel i7-3770</td>
<td>0.0569it/s</td>
<td>17.57s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>Intel i5-5257U</td>
<td>0.05456it/s</td>
<td>18.33s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>Intel i5-6300U</td>
<td>0.0295it/s</td>
<td>33.85s/it</td>
<td>CPU</td>
<td>KoboldCPP (1.95.1)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>MediaTek Dimensity 1080</td>
<td>0.0240it/s</td>
<td>41.71s/it</td>
<td>CPU</td>
<td>SDNext (a9c65c0e)</td>
<td>HyperOS 2.0.5.0 (A14 - 4.19.191)</td>
</tr>
<tr>
<td>Qualcomm SD 8 Gen 1</td>
<td>0.0227it/s</td>
<td>44.10s/it</td>
<td>CPU</td>
<td>KoboldCPP (1.100.1)</td>
<td>OneUI 7.0 (A15 - 5.10.226)</td>
<td>8 Threads. Prone to crashing</td>
</tr>
<tr>
<td>Intel i5-4300U</td>
<td>0.0117it/s</td>
<td>85.34s/it</td>
<td>CPU</td>
<td>KoboldCPP (1.95.1)</td>
<td>Arch Linux</td>
</tr>
</table>
<h3>256px</h3>
<p>Runable even on 1GB of VRAM!</p>
<table>
<tr>
<th>Chip</th>
<th>it/s</th>
<th>s/it</th>
<th>Backend</th>
<th>App (Commit)</th>
<th>OS</th>
<th>Notes</th>
</tr>
<tr>
<td>NVIDIA RTX PRO 6000 B</td>
<td>53.24it/s</td>
<td>0.0188s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
</tr>
<tr>
<td>NVIDIA RTX 5090</td>
<td>47.02it/s</td>
<td>0.0213s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (37d620a)</td>
<td>Fedora 42</td>
<td></td>
</tr>
<tr>
<td>AMD RX 7900 XTX</td>
<td>34.70it/s</td>
<td>0.03s/it</td>
<td>ROCm 6.4.1</td>
<td>SDNext (aa0652ca)</td>
<td>Arch Linux</td>
<td>Flash Attention, OC</td>
</tr>
<tr>
<td>NVIDIA RTX 3090</td>
<td>33.85it/s</td>
<td>0.0295s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>NVIDIA RTX A4000</td>
<td>31.52it/s</td>
<td>0.03s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
</tr>
<tr>
<td>NVIDIA RTX 3070</td>
<td>30.35it/s</td>
<td>0.03s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Unknown</td>
</tr>
<tr>
<td>NVIDIA RTX 3060 Ti</td>
<td>27.28it/s</td>
<td>0.04s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
</tr>
<tr>
<td>Intel A770</td>
<td>26.30it/s</td>
<td>0.04s/it</td>
<td>OpenVINO 2025.2</td>
<td>SDNext (a9c65c0e)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>NVIDIA RTX 5060 Ti</td>
<td>25.45it/s</td>
<td>0.0409s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>AMD RX 7900 XTX</td>
<td>20.51it/s</td>
<td>0.05s/it</td>
<td>ROCm 6.4.1</td>
<td>SDNext (aa0652ca)</td>
<td>Arch Linux</td>
<td>INT8 Matmul, Flash Attention, OC</td>
</tr>
<tr>
<td>NVIDIA RTX 3060</td>
<td>15.56it/s</td>
<td>0.0643s/it</td>
<td>CUDA 12.9</td>
<td>ComfyUI (Unknown)</td>
<td>Unknown</td>
</tr>
<tr>
<td>NVIDIA GTX 1080 Ti</td>
<td>8.23it/s</td>
<td>0.12it/s</td>
<td>CUDA 12.6</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>Intel A770</td>
<td>6.20it/s</td>
<td>0.16s/it</td>
<td>PyTorch XPU 2.7.1</td>
<td>SDNext (a9c65c0e)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>NVIDIA GTX 980</td>
<td>4.43it/s</td>
<td>0.23s/it</td>
<td>CUDA 12.4</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>AMD Pro W5500</td>
<td>3.84it/s</td>
<td>0.26s/it</td>
<td>Vulkan 1.3</td>
<td>KoboldCPP (1.95.1)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>NVIDIA GTX 1650 Mobile</td>
<td>3.63it/s</td>
<td>0.28s/it</td>
<td>CUDA 12.8</td>
<td>ComfyUI (5b12b55)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>AMD Pro W5500</td>
<td>2.84it/s</td>
<td>0.35s/it</td>
<td>Vulkan 1.3</td>
<td>KoboldCPP (1.95.1)</td>
<td>Windows 11 22H2</td>
</tr>
<tr>
<td>AMD Pro W5500</td>
<td>2.05it/s</td>
<td>0.48s/it</td>
<td>DirectML</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>NVIDIA GTX 1050 Ti</td>
<td>1.82it/s</td>
<td>0.55s/it</td>
<td>CUDA 12.4</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>AMD EPYC 9654</td>
<td>1.78it/s</td>
<td>0.56s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Ubuntu Server 24.04.2 LTS</td>
</tr>
<tr>
<td>AMD RG 512SP Mobile</td>
<td>1.19it/s</td>
<td>0.84s/it</td>
<td>Vulkan 1.3</td>
<td>KoboldCPP (1.100.1)</td>
<td>Arh Linux</td>
<td></td>
</tr>
<tr>
<td>AMD Ryzen 5800X3D</td>
<td>1.16it/s</td>
<td>0.86s/it</td>
<td>CPU</td>
<td>SDNext (a9c65c0e)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>AMD Ryzen 5800X</td>
<td>1.02it/s</td>
<td>0.98s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>Intel i7-10700K</td>
<td>0.83it/s</td>
<td>1.21s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>NVIDIA MX 330</td>
<td>0.75it/s</td>
<td>1.34s/it</td>
<td>CUDA 12.4</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>AMD Pro WX 4100</td>
<td>0.71it/s</td>
<td>1.41s/it</td>
<td>Vulkan 1.3</td>
<td>KoboldCPP (1.95.1)</td>
<td>Windows 11 22H2</td>
</tr>
<tr>
<td>Qualcomm SD 8 Gen 1</td>
<td>0.70it/s</td>
<td>1.42s/it</td>
<td>MNN GPU</td>
<td>Local Dream (2.1.1)</td>
<td>OneUI 7.0 (A15 - 5.10.226)</td>
<td></td>
</tr>
<tr>
<td>AMD Pro WX 4100</td>
<td>0.66it/s</td>
<td>1.50s/it</td>
<td>DirectML</td>
<td>ComfyUI (Unknown)</td>
<td>Windows 11 22H2</td>
</tr>
<tr>
<td>AMD HD 7790</td>
<td>0.60it/s</td>
<td>1.65s/it</td>
<td>DirectML</td>
<td>ComfyUI (Unknown)</td>
<td>Windows 11 22H2</td>
</tr>
<tr>
<td>Qualcomm SD 8 Gen 1</td>
<td>0.49it/s</td>
<td>2.05s/it</td>
<td>MNN CPU</td>
<td>Local Dream (2.1.1)</td>
<td>OneUI 7.0 (A15 - 5.10.226)</td>
<td></td>
</tr>
<tr>
<td>Intel i7-1165G7</td>
<td>0.48it/s</td>
<td>2.07s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>AMD HD 7750</td>
<td>0.48it/s</td>
<td>2.08s/it</td>
<td>DirectML</td>
<td>ComfyUI (Unknown)</td>
<td>Windows 11 22H2</td>
</tr>
<tr>
<td>Intel i7-4790K</td>
<td>0.40it/s</td>
<td>2.46s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>Intel i7-3770</td>
<td>0.26it/s</td>
<td>3.84s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>Intel i5-6300U</td>
<td>0.14it/s</td>
<td>6.98s/it</td>
<td>CPU</td>
<td>KoboldCPP (1.95.1)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>Intel i5-5257U</td>
<td>0.23it/s</td>
<td>4.31s/it</td>
<td>CPU</td>
<td>ComfyUI (Unknown)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>MediaTek Dimensity 1080</td>
<td>0.10it/s</td>
<td>9.57s/it</td>
<td>CPU</td>
<td>SDNext (a9c65c0e)</td>
<td>HyperOS 2.0.5.0 (A14 - 4.19.191)</td>
</tr>
<tr>
<td>Intel i7-3770</td>
<td>0.0316it/s</td>
<td>31.68s/it</td>
<td>Old CPU</td>
<td>KoboldCPP (1.95.1)</td>
<td>Arch Linux</td>
</tr>
<tr>
<td>Intel Core 2 Quad Q9300</td>
<td>0.0081it/s</td>
<td>123.34s/it</td>
<td>CPU Failsafe</td>
<td>KoboldCPP (1.95.1)</td>
<td>Arch Linux</td>
<td></td>
</tr>
<tr>
<td>Intel Core 2 Duo T9300</td>
<td>0.0049it/s</td>
<td>204.41s/it</td>
<td>CPU Failsafe</td>
<td>KoboldCPP (1.95.1)</td>
<td>Arch Linux</td>
<td></td>
</tr>
</table>
</div>
</div>
<div class="section-div">
<h1>How do I make my gens faster?</h1>
<ul>
<li>Use simple samplers such as Euler instead of double step ones such as DPM 2M</li>
<li>Lower your image sizes. SDXL can work coherently down to 384px and SD1.5 can go down to 128px.</li>
<li>Use addons such as TeaCache</li>
<li>Use low step LoRAs such as DMD2</li>
<li>As a last resort, disable CFG by setting your CFG to 1. This will disable your negative prompt but also increase your speeds drastically. This will also severely affect your output quality. The lack of negative prompt can be circumvented with the use of NAG</li>
<li>Upgrade your potato with a new GPU if all else fails.</li>
</ul>
</div>
</div>
- Task
- Text-to-Image
- License
- gfdl
- Languages
- en
- Created
- Jun 26, 2025
- Updated
- Nov 30, 2025