will/swarm-master
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: add OpenVINO NPU prototype services

Browse Source
This commit is contained in:
William Valentin
2026-06-04 11:41:55 -07:00
parent d67c259187
commit 5b01b1bd11
26 changed files with 2165 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files
scripts/npu-service-health.sh
+4 -4
View File
@@ -68,7 +68,7 @@ section "HTTP health"
http_json "RAG endpoint" "http://127.0.0.1:18810/healthz" || true
http_json "RAG/embedding health wrapper" "http://127.0.0.1:18814/healthz" || true
http_json "Whisper NPU" "http://127.0.0.1:18816/health" || true
http_json "OpenVINO embeddings" "http://127.0.0.1:18817/health" || true
http_json "OpenVINO embeddings" "http://127.0.0.1:18817/healthz" || true
# Prototypes are expected to be unavailable until explicitly started/approved.
http_json "NPU reranker prototype" "http://127.0.0.1:18818/readyz" || true
http_json "NPU router classifier prototype" "http://127.0.0.1:18819/healthz" || true
@@ -91,10 +91,10 @@ if [[ -z "$response" ]]; then
fi
delta=$((after - before))
printf 'sysfs_before_us=%s\nsysfs_after_us=%s\nsysfs_delta_us=%s\n' "$before" "$after" "$delta"
printf '%s' "$response" | python - <<'PY' || true
import json, sys
RESPONSE_JSON="$response" python - <<'PY' || true
import json, os
try:
data = json.load(sys.stdin)
data = json.loads(os.environ.get('RESPONSE_JSON', ''))
except Exception as exc:
print(f'response_parse_error={type(exc).__name__}: {exc}')
raise SystemExit(0)