Add zero-copy GPU encoding pipeline

- New GPUHLSOutput class for direct GPU-to-NVENC encoding
- RGB→NV12 conversion via CUDA kernel (no CPU transfer)
- Uses PyNvVideoCodec for zero-copy GPU encoding
- ~220fps vs ~4fps with CPU pipe approach
- Automatically used when PyNvVideoCodec is available

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

streaming/stream_sexp_generic.py

@@ -863,8 +863,14 @@ class StreamInterpreter:
         # Import output classes - handle both package and direct execution
         try:
             from .output import PipeOutput, DisplayOutput, FileOutput, HLSOutput, IPFSHLSOutput
+            from .gpu_output import GPUHLSOutput, check_gpu_encode_available
         except ImportError:
             from output import PipeOutput, DisplayOutput, FileOutput, HLSOutput, IPFSHLSOutput
+            try:
+                from gpu_output import GPUHLSOutput, check_gpu_encode_available
+            except ImportError:
+                GPUHLSOutput = None
+                check_gpu_encode_available = lambda: False
 
         self._init()
 
@@ -909,8 +915,14 @@ class StreamInterpreter:
             hls_dir = output[:-9]  # Remove /ipfs-hls suffix
             import os
             ipfs_gateway = os.environ.get("IPFS_GATEWAY_URL", "https://ipfs.io/ipfs")
-            out = IPFSHLSOutput(hls_dir, size=(w, h), fps=fps, audio_source=audio, ipfs_gateway=ipfs_gateway,
-                               on_playlist_update=self.on_playlist_update)
+            # Use GPU encoding if available (zero-copy, much faster)
+            if GPUHLSOutput is not None and check_gpu_encode_available():
+                print(f"[StreamInterpreter] Using GPU zero-copy encoding", file=sys.stderr)
+                out = GPUHLSOutput(hls_dir, size=(w, h), fps=fps, audio_source=audio, ipfs_gateway=ipfs_gateway,
+                                   on_playlist_update=self.on_playlist_update)
+            else:
+                out = IPFSHLSOutput(hls_dir, size=(w, h), fps=fps, audio_source=audio, ipfs_gateway=ipfs_gateway,
+                                   on_playlist_update=self.on_playlist_update)
         else:
             out = FileOutput(output, size=(w, h), fps=fps, audio_source=audio)