From 44492c64681389db9f916dbbc41465ebec994f05 Mon Sep 17 00:00:00 2001
From: Thierry Pouplier <thierrypouplier@gmail.com>
Date: Mon, 27 Apr 2026 14:49:14 -0400
Subject: [PATCH] Ollama tweeks

---
 ai/compose.yml | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/ai/compose.yml b/ai/compose.yml
index 948c868..49f66fc 100644
--- a/ai/compose.yml
+++ b/ai/compose.yml
@@ -68,11 +68,13 @@ services:
       - HCC_AMDGPU_TARGET=gfx906
       - HIP_VISIBLE_DEVICES=0,1
       - ROCR_VISIBLE_DEVICES=0,1
-      - HSA_ENABLE_SDMA=0 
+      - HSA_ENABLE_SDMA=1
+      - OLLAMA_MAX_LOADED_MODELS=1
+      - OLLAMA_MAX_QUEUE=512
       - OLLAMA_HOST=0.0.0.0
       - OLLAMA_DEBUG=1
-      - OLLAMA_FLASH_ATTENTION=0
-      - OLLAMA_NUM_PARALLEL=2
+      - OLLAMA_FLASH_ATTENTION=1
+      - OLLAMA_NUM_PARALLEL=1
     devices:
       # Map the render nodes and KFD for ROCm to work inside the container
       - /dev/kfd:/dev/kfd