engineering/lumbda
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

bend port flip: 9091 → 8320 (BEND mnemonic)

Browse source 
Port mnemonic embedded verbatim across our source files:

  8 ~= B (implied infinity B flattened; bake a cake; baby & me)
  3 ~= E (backward)
  2 ~= N (pivoted 90 degrees)
  0 ~= D (flattened)

Files touched:
- examples/cuda-fanout/gpu-worker.lsp (*worker-port*)
- examples/cuda-fanout/bend.lsp (*bend-worker-port*)
- examples/cuda-fanout/mock-worker.py (PORT)
- examples/cuda-fanout/bench_tiers.py (asm tier fixed port)
- examples/cuda-fanout/smoke-bend.lsp + smoke-bend-asm.lsp
- examples/cuda-fanout/README.md
- www/bend.html (catalog + multi-host text)
- Makefile (PORT default + comment)

bend.html updates 3090-ai + ai (4090) fleet table to active 2-node
mesh on 8320 — qwen moves off ai, bend takes over.
This commit is contained in:
russell@unturf.com 2026-06-06 15:06:18 -04:00
parent 9a547ee7df
commit 8d66bc01f1
No known key found for this signature in database
17 changed files with 76 additions and 20 deletions
Download patch file Download diff file Expand all files Collapse all files

14
www/bend.html
View file

@ -30,7 +30,13 @@
make gpu-worker
# → builds examples/cuda-fanout/shake256-fanout
# → builds the C tier (~10× faster wire orchestration than Python)
# → launches gpu-worker.lsp on port 9091
# → launches gpu-worker.lsp on port 8320 (BEND)
# Port 8320 = BEND mnemonic:
# 8 ~= B (implied infinity B flattened; bake a cake; baby & me)
# 3 ~= E (backward)
# 2 ~= N (pivoted 90 degrees)
# 0 ~= D (flattened)
# Override tier or port:
make gpu-worker LUMBDA=python PORT=9001 # easier debugging
@ -74,11 +80,11 @@ make gpu-worker LUMBDA=asm # smallest footprint</code></pre>
<table>
<thead><tr><th>host</th><th>GPU</th><th>arch</th><th>port</th><th>status</th></tr></thead>
<tbody>
<tr><td><code>3090-ai.foxhop.net</code></td><td>RTX 3090 (24 GB)</td><td>sm_86</td><td>9091</td><td><strong>active</strong> &mdash; production worker</td></tr>
<tr><td><code>ai.foxhop.net</code></td><td>RTX 4090 (24 GB)</td><td>sm_89</td><td>9092</td><td>reserved for qwen LLM (llama.cpp); bend worker enabled per workload</td></tr>
<tr><td><code>3090-ai.foxhop.net</code></td><td>RTX 3090 (24 GB)</td><td>sm_86</td><td>8320</td><td><strong>active</strong> &mdash; production worker</td></tr>
<tr><td><code>ai.foxhop.net</code></td><td>RTX 4090 (24 GB)</td><td>sm_89</td><td>8320</td><td><strong>active</strong> &mdash; 2-node mesh</td></tr>
</tbody>
</table>
<p>Multi-host fan-out was validated at 1.6&times; aggregate throughput on small workloads, but routine round-robin against the 4090 would steal VRAM from qwen. Caller opts in explicitly when a workload justifies fan-out: <code>(bend-set-workers! '((&quot;3090-ai.foxhop.net&quot; . 9091) (&quot;ai.foxhop.net&quot; . 9092)))</code> brings the 4090 online for that call.</p>
<p>Multi-host fan-out enabled. Both 3090 + 4090 serve bend workloads on port 8320 (BEND mnemonic). Round-robin or explicit selection via <code>(bend-set-workers! '((&quot;3090-ai.foxhop.net&quot; . 8320) (&quot;ai.foxhop.net&quot; . 8320)))</code>.</p>
</section>
<section id="health">