Add support for SR-IOV GPU stats (#15796)

* Add option to treat GPU as SRIOV in order for stats to work correctly

* Add to intel docs

* fix tests

frigate/config/telemetry.py

@@ -11,6 +11,9 @@ class StatsConfig(FrigateBaseModel):
     network_bandwidth: bool = Field(
         default=False, title="Enable network bandwidth for ffmpeg processes."
     )
+    sriov: bool = Field(
+        default=False, title="Treat device as SR-IOV to support GPU stats."
+    )
 
 
 class TelemetryConfig(FrigateBaseModel):

frigate/stats/util.py

@@ -195,7 +195,7 @@ async def set_gpu_stats(
                 continue
 
             # intel QSV GPU
-            intel_usage = get_intel_gpu_stats()
+            intel_usage = get_intel_gpu_stats(config.telemetry.stats.sriov)
 
             if intel_usage is not None:
                 stats["intel-qsv"] = intel_usage or {"gpu": "", "mem": ""}
@@ -220,7 +220,7 @@ async def set_gpu_stats(
                     continue
 
                 # intel VAAPI GPU
-                intel_usage = get_intel_gpu_stats()
+                intel_usage = get_intel_gpu_stats(config.telemetry.stats.sriov)
 
                 if intel_usage is not None:
                     stats["intel-vaapi"] = intel_usage or {"gpu": "", "mem": ""}

frigate/test/test_gpu_stats.py

@@ -38,7 +38,7 @@ class TestGpuStats(unittest.TestCase):
         process.returncode = 124
         process.stdout = self.intel_results
         sp.return_value = process
-        intel_stats = get_intel_gpu_stats()
+        intel_stats = get_intel_gpu_stats(False)
         print(f"the intel stats are {intel_stats}")
         assert intel_stats == {
             "gpu": "1.13%",

frigate/util/services.py

@@ -255,7 +255,7 @@ def get_amd_gpu_stats() -> dict[str, str]:
         return results
 
 
-def get_intel_gpu_stats() -> dict[str, str]:
+def get_intel_gpu_stats(sriov: bool) -> dict[str, str]:
     """Get stats using intel_gpu_top."""
 
     def get_stats_manually(output: str) -> dict[str, str]:
@@ -302,6 +302,9 @@ def get_intel_gpu_stats() -> dict[str, str]:
         "1",
     ]
 
+    if sriov:
+        intel_gpu_top_command += ["-d", "drm:/dev/dri/card0"]
+
     p = sp.run(
         intel_gpu_top_command,
         encoding="ascii",