swap back to nvenc/HEVC and redefine hosts accordingly

encode_VOD_pyro.py

@@ -11,10 +11,34 @@ import shutil
 import time
 
 # Distributed mode requires tqdm and ffmpeg_distributed.py (SSH, Unix select.poll); on Windows use WSL or Linux.
-DISTRIBUTED_HOSTS_DEFAULT = ["Pyro", "RenderScrap", "root@GuiltsCurse", "PostIrony", "root@Godzilla"]
-DISTRIBUTED_REMOTE_ARGS_DEFAULT = "-c:v libaom-av1 -crf 0 -b:v 9000k -maxrate 9000k -bufsize 18000k -cpu-used 8 -row-mt 1 -an"
+# Workers = (ssh_host, gpu_index). Unraid (GuiltsCurse, Godzilla) excluded; RenderScrap has 2 GPUs.
+DISTRIBUTED_WORKERS_DEFAULT = [
+    ("Pyro", 0),
+    ("RenderScrap", 0),
+    ("RenderScrap", 1),
+    ("PostIrony", 0),
+]
+DISTRIBUTED_REMOTE_ARGS_DEFAULT = "-c:v hevc_nvenc -preset p7 -tune hq -rc vbr -rc-lookahead 32 -spatial-aq 1 -aq-strength 15 -cq 0 -b:v 9000k -maxrate 9000k -bufsize 18000k -an"
 DISTRIBUTED_SEGMENT_SECONDS = 60
 
+
+def _parse_workers_env(s):
+    """Parse DISTRIBUTED_WORKERS e.g. 'Pyro:0,RenderScrap:0,RenderScrap:1,PostIrony:0' -> [(host, gpu_id), ...]."""
+    out = []
+    for part in (s or "").strip().split(","):
+        part = part.strip()
+        if not part:
+            continue
+        if ":" in part:
+            host, gpu = part.rsplit(":", 1)
+            try:
+                out.append((host.strip(), int(gpu.strip())))
+            except ValueError:
+                pass
+        else:
+            out.append((part, 0))
+    return out
+
 # ANSI color codes
 class Colors:
     PURPLE = '\033[95m'
@@ -342,9 +366,9 @@ def encode_dvr(input_file, output_dir, gpu):
                       f"{Colors.RED}Unexpected error encoding {input_path}: {e}{Colors.ENDC}")
 
 
-def encode_dvr_distributed(input_file, output_dir, hosts, segment_seconds=60, remote_args=None, concat_args="-c:a copy", probe_host=None, probe_path=None, remote_ffmpeg_path=None):
-    """Encode one file using ffmpeg_distributed (split -> farm -> concat). Segment temp dirs go under script dir/tmp/.
-    If probe_host and probe_path are set, ffprobe runs there (faster when input is on NAS)."""
+def encode_dvr_distributed(input_file, output_dir, workers, segment_seconds=60, remote_args=None, concat_args="-c:a copy", probe_host=None, probe_path=None, remote_ffmpeg_path=None):
+    """Encode one file using ffmpeg_distributed (split -> farm -> concat). workers = [(host, gpu_id), ...].
+    Segment temp dirs go under script dir/tmp/. If probe_host and probe_path are set, ffprobe runs there (faster when input is on NAS)."""
     input_path = Path(input_file).resolve()
     output_path = (Path(output_dir) / f"{input_path.stem}{input_path.suffix}").resolve()
     if output_path.exists():
@@ -366,11 +390,11 @@ def encode_dvr_distributed(input_file, output_dir, hosts, segment_seconds=60, re
     try:
         os.chdir(output_dir)
         from ffmpeg_distributed import encode as distributed_encode
-        safe_log_info(f"Distributed encode: {input_path} -> {output_path} (hosts: {hosts})")
-        print(f"{Colors.BLUE}Distributed encode (AV1): {input_path.name}{Colors.ENDC}")
+        safe_log_info(f"Distributed encode: {input_path} -> {output_path} (workers: {workers})")
+        print(f"{Colors.BLUE}Distributed encode (HEVC): {input_path.name}{Colors.ENDC}")
         remote_ffmpeg = remote_ffmpeg_path or os.environ.get("DISTRIBUTED_REMOTE_FFMPEG_PATH")
         distributed_encode(
-            hosts,
+            workers,
             str(input_path),
             str(output_path),
             segment_seconds=segment_seconds,
@@ -399,13 +423,14 @@ if __name__ == "__main__":
     output_dir = "output"
     os.makedirs(output_dir, exist_ok=True)
 
-    hosts_str = os.environ.get("DISTRIBUTED_HOSTS")
-    if hosts_str:
-        hosts = [h.strip() for h in hosts_str.split(",") if h.strip()]
+    workers_str = os.environ.get("DISTRIBUTED_WORKERS")
+    if workers_str:
+        workers = _parse_workers_env(workers_str)
     else:
-        hosts = DISTRIBUTED_HOSTS_DEFAULT
-        print(f"{Colors.BLUE}Using hosts: {', '.join(hosts)}{Colors.ENDC}")
-    safe_log_info(f"Distributed mode; hosts: {hosts}")
+        workers = DISTRIBUTED_WORKERS_DEFAULT
+    workers_desc = ", ".join(f"{h}:gpu{g}" for h, g in workers)
+    print(f"{Colors.BLUE}Using workers: {workers_desc}{Colors.ENDC}")
+    safe_log_info(f"Distributed mode; workers: {workers}")
 
     files = [f for f in os.listdir(input_dir) if f.endswith(('.mp4', '.DVR.mp4'))]
     total_files = len(files)
@@ -419,4 +444,4 @@ if __name__ == "__main__":
             input_file = os.path.join(input_dir, file)
             safe_log_info(f"Processing file {i}/{total_files}: {file}")
             print(f"\n{Colors.BLUE}Processing file {i}/{total_files}: {file}{Colors.ENDC}")
-            encode_dvr_distributed(input_file, output_dir, hosts, segment_seconds=DISTRIBUTED_SEGMENT_SECONDS) 
+            encode_dvr_distributed(input_file, output_dir, workers, segment_seconds=DISTRIBUTED_SEGMENT_SECONDS)