GPUDirect - is TCC mode a requirement?

I’ve been trying to get gpu to gpu, system to system rdma working and found a message board post stating that GPUDirect requires the GPUs to be in TCC mode.
I’m not using the GPUs for video, and the nvidia-smi doesn’t show anything about TCC mode.
I also get a not supported message if I try to turn it on (see below).
Does all of this indicate that GPUDirect won’t work with my GPUs or system?
Is there some RHEL configuration change I can do to get them into TCC mode?
I can do rdma host/gpu pinned memory to host/gpu pinned memory, just not gpu-gpu.
The mellanox perftools all work with --use-cuda… so I don’t know what to believe. I have to grudgingly admit that must still point to my implementation. But I’m still asking the TCC question.

[root@D2703 rdmaX]# nvidia-smi
Thu Aug 1 19:18:36 2019
| NVIDIA-SMI 418.67 Driver Version: 418.67 CUDA Version: 10.1 |
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| 0 Quadro K600 Off | 00000000:05:00.0 Off | N/A |
| 25% 39C P8 N/A / N/A | 0MiB / 982MiB | 0% Default |

| Processes: GPU Memory |
| GPU PID Type Process name Usage |
| No running processes found |
[root@D2703 rdmaX]# nvidia-smi -g 0 -dm 1
Changing driver models is not supported for GPU 00000000:05:00.0 on this platform.
Treating as warning and moving on.
All done.
[root@D2703 rdmaX]#

TCC mode is applicable to GPUs operating in windows. It is not applicable (it has no relevance) to GPUs operating in linux.