Couple of cleanups.

Remove queue limitation since we loop through HW queues now.
Add a DevLogError if we fail to create the hsa_queue. A ticket showed a regression there.

Change-Id: I4f58e405f88e75600a762f6d6352838c969cdb5e

rocclr/device/rocm/rocdevice.cpp

@@ -1963,6 +1963,7 @@ hsa_queue_t* Device::acquireQueue(uint32_t queue_size_hint, bool coop_queue) {
                           &queue) != HSA_STATUS_SUCCESS) {
     queue_size >>= 1;
     if (queue_size < 64) {
+      DevLogError("Device::acquireQueue: hsa_queue_create failed!");
       return nullptr;
     }
   }

rocclr/device/rocm/rocsettings.cpp

@@ -57,9 +57,6 @@ Settings::Settings() {
   nonCoherentMode = getenv("OPENCL_USE_NC_MEMORY_POLICY");
   enableNCMode_ = (nonCoherentMode) ? true : false;
 
-  commandQueues_ = 200;  //!< Field value set to maximum number
-                         //!< concurrent Virtual GPUs for ROCm backend
-
   // Disable image DMA by default (ROCM runtime doesn't support it)
   imageDMA_ = false;
 
@@ -188,10 +185,6 @@ void Settings::override() {
     maxWorkGroupSize3DZ_ = GPU_MAX_WORKGROUP_SIZE_3D_Z;
   }
 
-  if (!flagIsDefault(GPU_MAX_COMMAND_QUEUES)) {
-    commandQueues_ = GPU_MAX_COMMAND_QUEUES;
-  }
-
   if (!flagIsDefault(GPU_XFER_BUFFER_SIZE)) {
     xferBufSize_ = GPU_XFER_BUFFER_SIZE * Ki;
   }

rocclr/device/rocm/rocvirtual.cpp

@@ -694,11 +694,6 @@ VirtualGPU::~VirtualGPU() {
 }
 
 bool VirtualGPU::create() {
-  // Checking Virtual gpu unique index for ROCm backend
-  if (index() > device().settings().commandQueues_) {
-    return false;
-  }
-
   // Pick a reasonable queue size
   uint32_t queue_size = 1024;
   gpu_queue_ = roc_device_.acquireQueue(queue_size, cooperative_);