fix: revert dual queue to single GPU — light worker caused 2x regression

Root cause: render-worker and render-worker-light shared the same GPU,
causing contention. Complex TRB renders went from 17s → 36s (2x slower).

Changes:
- Thumbnails back to asset_pipeline queue (not asset_pipeline_light)
- Dispatch routing always uses asset_pipeline (no queue splitting)
- render-worker-light gated behind "multi-gpu" profile — only starts with:
  docker compose --profile multi-gpu up -d
- For single-GPU setups: all rendering is sequential on one worker

The dual queue approach is correct for multi-GPU machines where each
worker gets its own GPU. On single-GPU, serial execution is faster
than concurrent GPU contention.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

backend/app/domains/pipeline/tasks/render_order_line.py

@@ -40,35 +40,13 @@ def dispatch_order_line_render(order_line_id: str):
                 logger.info(f"OrderLine {order_line_id}: order {order.status.value} — not dispatching")
                 return
 
-        # Route light renders (small stills) to asset_pipeline_light,
+        # All renders go to asset_pipeline (single-GPU default).
-        # heavy renders (HQ stills, animations) stay on asset_pipeline.
+        # For multi-GPU setups: enable render-worker-light in docker-compose
-        is_animation = False
+        # and change target_queue logic below to route small stills to
-        max_dim = 0
+        # asset_pipeline_light for concurrent rendering.
-        if line:
+        pass
-            from app.models.output_type import OutputType
-            ot = session.execute(
-                select(OutputType).where(OutputType.id == line.output_type_id)
-            ).scalar_one_or_none() if line.output_type_id else None
-            if ot:
-                is_animation = bool(getattr(ot, 'is_animation', False))
-                rs = ot.render_settings or {}
-                w = int(rs.get("width", 0) or 0)
-                h = int(rs.get("height", 0) or 0)
-                max_dim = max(w, h)
-            # Apply render_overrides for routing decisions
-            ro = getattr(line, 'render_overrides', None) or {}
-            if ro.get("width"):
-                w = int(ro["width"])
-            if ro.get("height"):
-                h = int(ro["height"])
-            if w or h:
-                max_dim = max(w, h)
 
-    if max_dim > 0 and max_dim <= 1024 and not is_animation:
-        target_queue = "asset_pipeline_light"
-    else:
     target_queue = "asset_pipeline"
-
     logger.info(f"Dispatching render for order line: {order_line_id} -> queue={target_queue}")
     render_order_line_task.apply_async(args=[order_line_id], queue=target_queue)
 

backend/app/domains/pipeline/tasks/render_thumbnail.py

@@ -72,7 +72,7 @@ def _pipeline_session(tenant_id: str | None = None):
         engine.dispose()
 
 
-@celery_app.task(bind=True, name="app.tasks.step_tasks.render_step_thumbnail", queue="asset_pipeline_light")
+@celery_app.task(bind=True, name="app.tasks.step_tasks.render_step_thumbnail", queue="asset_pipeline")
 def render_step_thumbnail(self, cad_file_id: str):
     """Render the thumbnail for a freshly-processed STEP file.
 
@@ -188,7 +188,7 @@ def render_step_thumbnail(self, cad_file_id: str):
     pl.step_done("render_step_thumbnail")
 
 
-@celery_app.task(bind=True, name="app.tasks.step_tasks.regenerate_thumbnail", queue="asset_pipeline_light")
+@celery_app.task(bind=True, name="app.tasks.step_tasks.regenerate_thumbnail", queue="asset_pipeline")
 def regenerate_thumbnail(self, cad_file_id: str, part_colors: dict):
     """Regenerate thumbnail with per-part colours."""
     pl = PipelineLogger(task_id=self.request.id)

docker-compose.yml

@@ -157,6 +157,7 @@ services:
               capabilities: [gpu, compute, utility, graphics]
 
   render-worker-light:
+    profiles: ["multi-gpu"]   # Only start with: docker compose --profile multi-gpu up -d
     build:
       context: .
       dockerfile: render-worker/Dockerfile