Skip to content

Fix ISO CI awscli on 24.04 and 3.5.0 hardware/FAQ/benchmarks copy #2

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
ShadowfetchLinux wants to merge 2 commits into
base: main
Choose a base branch
from
Draft

Fix ISO CI awscli on 24.04 and 3.5.0 hardware/FAQ/benchmarks copy #2

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
026b3c4
Install AWS CLI v2 in ISO CI and correct 3.5.0 site copy
cursoragent Aug 29, 2026
0c22d2f
Use curl -fsSL for the AWS CLI v2 installer download
cursoragent Aug 29, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 7 additions & 1 deletion .github/workflows/build-iso.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,7 +38,13 @@ jobs:
reprepro gnupg \
xorriso isolinux syslinux-common syslinux-utils \
grub-pc-bin grub-efi-amd64-bin mtools \
xz-utils python3 awscli
xz-utils python3 curl unzip

- name: Install AWS CLI v2
run: |
curl -fsSL https://awscli.amazonaws.com/awscli-exe-linux-x86_64.zip -o awscliv2.zip
unzip awscliv2.zip
sudo ./aws/install

- name: Verify release version and source gates
run: |
Expand Down
1 change: 1 addition & 0 deletions SHA256SUMS
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,3 +2,4 @@
81a906788ec48150d4a4527b4d9e7b09a974d3d577f1bd32ba3f333df8a1a86b shadowfetch-2.1.4-amd64.iso
848f043e4d6f85c3607e7034ba911a1ce8b4a317674feebef8b07fcd8f531c24 shadowfetch-2.1.5-amd64.iso
110b0d075e699a05a8a2f8f8dcd05f19454bc8ae09acd0745ca0d947db8c5e3c shadowfetch-3.0.0-amd64.iso
2af853b1f5dedfca17a7a63783f4c881e72e912f26082b10c07d45aafe57b995 shadowfetch-3.5.0-amd64.iso
127 changes: 124 additions & 3 deletions web/shadowfetch-linux-worker/src/index.js
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -39,6 +39,7 @@ export default {
if (route === "/linux/security") return html(securityPage());
if (route === "/linux/roadmap") return html(roadmapPage(await latestRelease(env)));
if (route === "/linux/faq") return html(faqPage());
if (route === "/linux/benchmarks") return html(benchmarksPage());
if (route === "/linux/agents") {
return new Response("This feature has been removed.\n", {
status: 410,
Expand Down Expand Up @@ -1351,12 +1352,12 @@ function hardwarePage() {
<li><strong>CPU:</strong> 64-bit Intel or AMD processor. Newer multi-core CPUs make local AI, video, and creative workloads much nicer.</li>
<li><strong>Memory:</strong> 4 GB minimum, 8 GB recommended for the desktop, 16 GB+ recommended for local AI and heavier creative work.</li>
<li><strong>Storage:</strong> 40 GB minimum, 100 GB+ recommended. Local AI models can consume several GB each.</li>
<li><strong>Graphics:</strong> Intel and AMD graphics use the normal Mesa stack. NVIDIA systems ship with the proprietary driver stack and hybrid laptops may still need manual tuning.</li>
<li><strong>Graphics:</strong> Intel and AMD graphics use the normal Mesa stack. NVIDIA setup is an explicit, simulate-first workflow that refuses removals and can create Phoenix Points on Btrfs. Hybrid laptops and physical accelerator performance still need hardware-specific validation.</li>
<li><strong>Firmware:</strong> UEFI or legacy BIOS. Secure Boot is not signed yet, so disable Secure Boot before booting the ISO.</li>
</ul>

<h2>NVIDIA and hybrid laptops</h2>
<p>The ISO includes the proprietary NVIDIA stack so NVIDIA desktops and laptops have a working path on first boot. Non-NVIDIA systems remove the stack after first boot to reclaim disk. Hybrid laptops can still vary by vendor firmware. PRIME offload is the intended path for discrete-GPU apps.</p>
<p>NVIDIA is an explicit, simulate-first path. The proprietary stack is not auto-installed and is not included then removed after first boot. The workflow refuses removals and can create Phoenix Points on Btrfs. Hybrid laptops and physical accelerator performance still need hardware-specific validation. After a user-installed NVIDIA driver, PRIME offload is the intended path for discrete-GPU apps.</p>
<pre><code>__NV_PRIME_RENDER_OFFLOAD=1 __GLX_VENDOR_LIBRARY_NAME=nvidia blender</code></pre>

<h2>Wi-Fi, Bluetooth, and peripherals</h2>
Expand Down Expand Up @@ -1486,7 +1487,7 @@ function faqPage() {
<p>Anyone who needs Secure Boot signing, enterprise support, a Debian-stable base, or a zero-surprise production workstation should wait and follow the changelog.</p>

<h2>Does it support NVIDIA?</h2>
<p>The proprietary NVIDIA stack is included and hybrid laptops use PRIME offload, but laptop firmware varies. Read the hardware notes and known issues before installing.</p>
<p>NVIDIA setup is an explicit, simulate-first workflow that refuses removals and can create Phoenix Points on Btrfs. It is not auto-installed and is not included then removed after first boot. Hybrid laptops and physical accelerator performance still need hardware-specific validation. Read the hardware notes and known issues before installing.</p>

<h2>Does local AI phone home?</h2>
<p>The local chat stack is designed to run on your machine. If you download models, those downloads come from the model host you choose. Shadowfetch does not add an account requirement or telemetry daemon to use the OS.</p>
Expand All @@ -1501,6 +1502,126 @@ function faqPage() {
});
}

function benchmarksPage() {
return shell({
title: "Historical 16GB local LLM benchmark — Shadowfetch Linux",
canonical: "/linux/benchmarks",
description: "A dated August 2026 Ollama benchmark on one NVIDIA RTX 5060 Ti: measured tokens per second, VRAM footprint, method, and explicit limits. It is not a Shadowfetch Linux 3.5.0 runtime claim.",
body: `
<section class="narrow">
<h1>Historical 16GB local LLM benchmark</h1>
<p class="lede">Dated August 2026. This is the old RTX 5060 Ti / <code>ollama pull</code> run. It is not a Shadowfetch Linux 3.5.0 runtime claim, and no 3.5.0 benchmark results are published here.</p>

<h2>What fits your card?</h2>
<p>Speed is measured on the 16 GB card above. Whether a model fits is mostly its memory footprint, which travels between cards far better than its speed does.</p>

<h2>The measured ladder</h2>
<div style="overflow-x:auto">
<table>
<thead>
<tr>
<th>Model</th>
<th>Params</th>
<th>Quant</th>
<th>VRAM used</th>
<th>On GPU</th>
<th>Context</th>
<th>Generation</th>
<th>Prompt</th>
<th>Cold load</th>
</tr>
</thead>
<tbody>
<tr>
<td>Gemma 3 · 1B<br><code>ollama pull gemma3:1b</code></td>
<td>999.89M</td>
<td>Q4_K_M</td>
<td>1.3 GB</td>
<td>100% GPU</td>
<td>32768</td>
<td>263 tok/s</td>
<td>7931 tok/s</td>
<td>2.61s</td>
</tr>
<tr>
<td>Qwen2.5 · 3B Instruct<br><code>ollama pull qwen2.5:3b-instruct</code></td>
<td>3.1B</td>
<td>Q4_K_M</td>
<td>3.1 GB</td>
<td>100% GPU</td>
<td>32768</td>
<td>175 tok/s</td>
<td>9768 tok/s</td>
<td>2.66s</td>
</tr>
<tr>
<td>Gemma 3 · 4B<br><code>ollama pull gemma3:4b</code></td>
<td>4.3B</td>
<td>Q4_K_M</td>
<td>4.6 GB</td>
<td>100% GPU</td>
<td>131072</td>
<td>120 tok/s</td>
<td>3697 tok/s</td>
<td>3.14s</td>
</tr>
<tr>
<td>Llama 3.2 · 3B<br><code>ollama pull llama3.2:3b</code></td>
<td>3.2B</td>
<td>Q4_K_M</td>
<td>4.8 GB</td>
<td>100% GPU</td>
<td>131072</td>
<td>174 tok/s</td>
<td>8985 tok/s</td>
<td>2.92s</td>
</tr>
<tr>
<td>Qwen2.5 · 7B Instruct<br><code>ollama pull qwen2.5:7b-instruct</code></td>
<td>7.6B</td>
<td>Q4_K_M</td>
<td>6.3 GB</td>
<td>100% GPU</td>
<td>32768</td>
<td>88 tok/s</td>
<td>4960 tok/s</td>
<td>8.55s</td>
</tr>
<tr>
<td>Qwen2.5 · 14B<br><code>ollama pull qwen2.5:14b</code></td>
<td>14.8B</td>
<td>Q4_K_M</td>
<td>13 GB</td>
<td>100% GPU</td>
<td>32768</td>
<td>45 tok/s</td>
<td>2548 tok/s</td>
<td>9.97s</td>
</tr>
</tbody>
</table>
</div>

<h3>Where to start</h3>
<ul>
<li>On a 16 GB card, start with Qwen2.5 · 14B — the largest measured model that fits, at 13 GB and 45 tok/s.</li>
<li>Want it snappier? Drop to Qwen2.5 · 7B Instruct at 6.3 GB and 88 tok/s — about 2× the generation speed.</li>
<li>On an 8 GB card, run Llama 3.2 · 3B — 4.8 GB and 174 tok/s.</li>
</ul>
<p>For the current release boundary, see local AI and coding agents. Buzz owns current model selection and download after confirmation.</p>

<h2>How these were measured</h2>
<ul>
<li>One NVIDIA GeForce RTX 5060 Ti (16 GB). Generation rate is eval tokens ÷ eval time from the Ollama API, taken as the median of 3 timed runs of 200 tokens after a warm-up.</li>
<li>VRAM used and “on GPU” are what <code>ollama ps</code> reports for the resident model — the model's real footprint, and whether all of it sits on the GPU or spills to the CPU (which is the moment speed falls off a cliff).</li>
<li>Params, quantisation and architecture are read from <code>ollama show</code> — the model file's own metadata, not typed by us. Everything shown is Q4_K_M-class weights, the quant most people actually run.</li>
<li>Speed is specific to this card; your tokens/sec will differ on other hardware. Footprint and fit travel much better — that's why fit is judged by memory, not speed.</li>
</ul>
</section>
`,
});
}

function notFoundPage() {
return shell({
title: "Not found — Shadowfetch Linux",
Expand Down