1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
|
#+TITLE: ratio - Desktop Workstation
#+DATE: 2026-01-27
#+HOSTNAME: ratio
* Automated Capabilities
# Maintained by the system-health-check workflow. Manual edits may be
# overwritten when the workflow detects drift between this drawer and the
# live system. Machine-readable capability signals used to dispatch which
# checks run on this host.
:PROPERTIES:
:FS: btrfs
:PM: pacman
:ORCH: topgrade
:SNAPSHOT: snapper
:MESH: tailscale
:BACKUP: source
:VIRT: libvirt,docker,podman
:INIT: systemd
:LAST_AUDIT: 2026-07-08
:END:
* Overview
| Field | Value |
|-------------+------------------------------------------|
| Hostname | ratio |
|-------------+------------------------------------------|
| Type | Desktop workstation |
|-------------+------------------------------------------|
| Maker | Framework |
|-------------+------------------------------------------|
| Model | Desktop (AMD Ryzen AI Max 300 Series) |
|-------------+------------------------------------------|
| Board | FRANMFCP06 v A6 |
|-------------+------------------------------------------|
| Firmware | INSYDE UEFI v03.03 (2025-09-16) |
|-------------+------------------------------------------|
| OS | Arch Linux |
|-------------+------------------------------------------|
| Kernel | 6.12.67-1-lts |
|-------------+------------------------------------------|
| Desktop | Hyprland (Wayland) |
|-------------+------------------------------------------|
| Previous ID | cogito (repurposed and renamed to ratio) |
|-------------+------------------------------------------|
* CPU
| Field | Value |
|---------------+---------------------------------------|
| Model | AMD Ryzen AI MAX+ 395 w/ Radeon 8060S |
|---------------+---------------------------------------|
| Architecture | Zen 5 (TSMC 4nm) |
|---------------+---------------------------------------|
| Cores/Threads | 16 cores / 32 threads |
|---------------+---------------------------------------|
| Base/Boost | 3.0 GHz / 5.15 GHz |
|---------------+---------------------------------------|
| Cache L1 | 1.2 MiB (d-16x48 KiB; i-16x32 KiB) |
|---------------+---------------------------------------|
| Cache L2 | 16 MiB (16x1024 KiB) |
|---------------+---------------------------------------|
| Cache L3 | 64 MiB (2x32 MiB) |
|---------------+---------------------------------------|
| NPU | XDNA 2, 50+ peak AI TOPS |
|---------------+---------------------------------------|
| Peak Perf | 59.4 FP16/BF16 TFLOPS @ 2.9GHz |
|---------------+---------------------------------------|
* GPU
| Field | Value |
|--------------+-----------------------------------------|
| Model | AMD Radeon 8060S (Strix Halo) |
|--------------+-----------------------------------------|
| Architecture | RDNA 3.5 |
|--------------+-----------------------------------------|
| CUs | 40 Compute Units |
|--------------+-----------------------------------------|
| Driver | amdgpu (kernel) |
|--------------+-----------------------------------------|
| Max VRAM | 96GB (via AMD Variable Graphics Memory) |
|--------------+-----------------------------------------|
| GPU Arch ID | gfx1151 |
|--------------+-----------------------------------------|
| PCIe | Gen 4, 16 GT/s, 16 lanes |
|--------------+-----------------------------------------|
* Memory
| Field | Value |
|--------------+--------------------------|
| Total | 128 GiB unified (LPDDR5) |
|--------------+--------------------------|
| Speed | 8000 MT/s |
|--------------+--------------------------|
| Channels | 8 (8x 16 GiB) |
|--------------+--------------------------|
| Manufacturer | Micron Technology |
|--------------+--------------------------|
| Part Number | MT62F4G32D8DV-023 WT |
|--------------+--------------------------|
* Storage
| Device | Model | Size | Interface |
|--------------+------------------------+-----------+-----------|
| /dev/nvme0n1 | WD BLACK SN850X 8000GB | 7.28 TiB | NVMe |
|--------------+------------------------+-----------+-----------|
| /dev/nvme1n1 | WD BLACK SN850X 8000GB | 7.28 TiB | NVMe |
|--------------+------------------------+-----------+-----------|
| *Total* | | 14.55 TiB | |
|--------------+------------------------+-----------+-----------|
Filesystem: btrfs with subvolumes (@, @home, @snapshots, @log, @pkg)
* Network
| Interface | Chipset | Speed | Type |
|-----------+-----------------+---------+----------|
| enp191s0 | Realtek RTL8126 | 5 GbE | Wired |
|-----------+-----------------+---------+----------|
| wlp192s0 | MediaTek MT7925 | Wi-Fi 7 | Wireless |
|-----------+-----------------+---------+----------|
Also has: docker0, tailscale0, virbr0 (virtual interfaces)
* Audio
- AMD Radeon High Definition Audio (HDMI, snd_hda_intel)
- AMD Ryzen HD Audio (3.5mm, snd_hda_intel)
- PipeWire v1.4.10 (pipewire-pulse, wireplumber)
* PCI Slots
| Slot | Info | Status |
|------+------------+-----------|
| 1 | M.2 JWLAN | Available |
|------+------------+-----------|
| 2 | M.2 JSSD1 | In use |
|------+------------+-----------|
| 3 | M.2 JSSD2 | In use |
|------+------------+-----------|
| 4 | PCIe Gen 4 | Available |
|------+------------+-----------|
* Connected Peripherals
- Monitor: Dell U3419W (HDMI-A-1, 3440x1440@60Hz)
- Keyboard: Das Keyboard 4 (Cherry MX Blue) — replacing with Keychron Q6 Pro
- Mouse: Logitech M650
- DAC/Amp: JDS Labs Element IV (ordered)
* AI/LLM Capability
Primary purpose includes local AI/LLM inference via Ollama.
| Model | Size | Purpose |
|------------------+-------+----------------------|
| deepseek-r1:70b | 42 GB | Main reasoning model |
|------------------+-------+----------------------|
| qwen3:30b | 18 GB | MoE chat (256K ctx) |
|------------------+-------+----------------------|
| deepseek-r1:14b | 9 GB | Light reasoning |
|------------------+-------+----------------------|
| nomic-embed-text | 274MB | RAG embeddings |
|------------------+-------+----------------------|
Inference benchmarks (AMD testing, LM Studio 0.3.11):
- Small models (1-3B): ~100+ tok/s
- Medium models (7-8B): ~60-80 tok/s
- Large models (20B): ~58 tok/s
- Very large models (120B): ~38 tok/s
|
