Catch exceptions from rocm-smi (#85)

Signed-off-by: coleramos425 <colramos@amd.com>
2023-02-21 09:47:45 -06:00
parent 1d82bc67b0
commit f77173a2f1
1 changed files with 4 additions and 0 deletions
@@ -216,6 +216,10 @@ def get_machine_specs(devicenum):
    ) = gpuinfo()
    rocm_smi = run(["rocm-smi"])

+    if search(r"ERROR(\S*)", rocm_smi) != None:
+        print("ERROR: No GPU detected. Unable to load rocm-smi")
+        sys.exit(1)
+
    device = rf"^\s*{devicenum}(.*)"

    hostname = socket.gethostname()