From 9b353838d1b9bf6f256f26cee0458a4c987c023c Mon Sep 17 00:00:00 2001
From: osmarks <osmarks@mycorrhiza>
Date: Tue, 24 Mar 2026 15:46:06 +0000
Subject: [PATCH] =?UTF-8?q?Edit=20=E2=80=98autogollark=E2=80=99?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 autogollark.myco | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/autogollark.myco b/autogollark.myco
index 70ba389..808127b 100644
--- a/autogollark.myco
+++ b/autogollark.myco
@@ -39,7 +39,7 @@ Autogollark currently comprises the dataset, the search API server and the [[htt
 * https://arxiv.org/abs/2510.14901
 * https://github.com/d0rc/egg.c and https://eshyperscale.github.io/. Does this actually work (at scale)? Why? Would be really nice for using AMX units.
 * Maybe compute grants are available for training.
-* Substantial bandwidth bottleneck on CPU (230GB/s nominal; 200GB/s benchmarked; 100GB/s per NUMA node, which llama.cpp handles awfully). Specdec/MTP would be useful.
+* Substantial bandwidth bottleneck on CPU (230GB/s nominal; 200GB/s benchmarked; 100GB/s per NUMA node, which llama.cpp handles awfully). Specdec/MTP would be useful. Can anything use AMX well though?
 }
 * Search over conversations with non-gollark simulacra? Should find //something// to use spare parallelism on local inference. Best-of-n? https://arxiv.org/abs/2505.10475
 * {Longer context, mux several channels.