I'm noticing the current A2AR tasks have a low GPU utilization. My 980Ti is crunching one currently at 44% and has therefore downclocked to its base clock speed. My 750Ti is also reporting lower than usual GPU utilization. Has anyone else noticed this or should I start investigating for problems with my machine?

I'm noticing the current A2AR tasks have a low GPU utilization. My 980Ti is crunching one currently at 44% and has therefore downclocked to its base clock speed. My 750Ti is also reporting lower than usual GPU utilization. Has anyone else noticed this or should I start investigating for problems with my machine?

You're not alone with this issue.
See the following threads:
"low GPU utilization with recent Gerard CXCL12?"
"About the GERARD_A2AR batch"
"GPU not fully utilized +Solution."

I'm noticing the current A2AR tasks have a low GPU utilization. My 980Ti is crunching one currently at 44% and has therefore downclocked to its base clock speed. My 750Ti is also reporting lower than usual GPU utilization. Has anyone else noticed this or should I start investigating for problems with my machine?

Now that I got home I see that the workunits from the batch called "GERARD_A2AR_NUL" are those which are extremely slow.
I had to suspend all CPU tasks to get the GPU usage up to 42~48%.
I have no idea how to increase the GPU usage any further. (besides running two simultaneously on the same GPU)

No CPU tasks running on this machine. I've suspended GPUGrid on the 750Ti and am now running 2x on the 980Ti. I guess I'll keep it like this until we get this sorted out.

I've also noticed the CPU usage on these tasks to be quite a bit higher than normal.

I have a refinement to make regarding the batch name: the "GERARD_A2AR_NUL" workunits were fine. (this is a previous batch).
Now we have (probably a follow-up) batch called "GERARD_A2AR_NUL1D" which is extremely slow.