From 09b05de4482931ce567fdb3a3ac5406df81c4b7d Mon Sep 17 00:00:00 2001
From: Massaki Archambault <marchambault@badjware.dev>
Date: Sat, 28 Dec 2024 16:17:37 -0500
Subject: [PATCH] enable gpu acceleration for openedai-speech

---
 kustomize/bases/ollama/ollama-deployment.yaml               | 2 +-
 .../bases/openedai-speech/openedai-speech-deployment.yaml   | 6 +++++-
 2 files changed, 6 insertions(+), 2 deletions(-)

diff --git a/kustomize/bases/ollama/ollama-deployment.yaml b/kustomize/bases/ollama/ollama-deployment.yaml
index 5de254a..4e95a8f 100644
--- a/kustomize/bases/ollama/ollama-deployment.yaml
+++ b/kustomize/bases/ollama/ollama-deployment.yaml
@@ -43,7 +43,7 @@ spec:
             limits:
               memory: 4Gi
               cpu: 3000m
-              nvidia.com/gpu: "2"
+              nvidia.com/gpu: "1"
               # amd.com/gpu: "1"
           ports:
             - name: http
diff --git a/kustomize/bases/openedai-speech/openedai-speech-deployment.yaml b/kustomize/bases/openedai-speech/openedai-speech-deployment.yaml
index b36ced9..ffea627 100644
--- a/kustomize/bases/openedai-speech/openedai-speech-deployment.yaml
+++ b/kustomize/bases/openedai-speech/openedai-speech-deployment.yaml
@@ -26,9 +26,13 @@ spec:
                     operator: In
                     values:
                       - amd64
+      runtimeClassName: nvidia
       containers:
         - name: openedai-speech
           image: ghcr.io/matatonic/openedai-speech
+          env:
+            - name: EXTRA_ARGS
+              value: --preload xtts_v2.0.2
           ports:
             - name: http
               containerPort: 8000
@@ -38,7 +42,7 @@ spec:
               memory: 1Gi
             limits:
               cpu: 1000m
-              memory: 1Gi
+              nvidia.com/gpu: "1"
           volumeMounts:
             - name: server-voices-pv
               mountPath: /app/voices