feat(dcgmi-v4): bump to dcgmi v4 to be compatible with 580 nvidia drivers

Author: shvbsle

test/images/nvidia/Dockerfile

@@ -37,7 +37,8 @@ RUN apt update -y \
       libhwloc-dev \
       freeglut3-dev \
       libglu1-mesa-dev \
-      datacenter-gpu-manager
+      datacenter-gpu-manager-4-cuda12 \
+      datacenter-gpu-manager-4-cuda13
 
 RUN ldconfig
 

test/images/nvidia/gpu_unit_tests/tests/test_basic.sh

@@ -37,10 +37,5 @@ test_04_dcgm_diagnostics()
     fi
 
     # https://docs.nvidia.com/datacenter/dcgm/latest/user-guide/dcgm-diagnostics.html#run-levels-and-tests
-    if [[ $EC2_INSTANCE_TYPE == g* ]]; then
-        # The G series instance don't have nvlink and GPU p2p communication
-        assert_status_code 0 'dcgmi diag -r "software,memory,memory_bandwidth"'
-    else
-        assert_status_code 0 "dcgmi diag -r 2"
-    fi
+    assert_status_code 0 "dcgmi diag -r 2"
 }