[jra101]

Have you tried reducing the register count in your FP32 FMA test by increasing the iteration count and reducing the number of values computed per loop?

Instead of computing 8 independent values, compute one with 8x more iterations:

    for (int i = 0; i < count * 8; i++) {
        v0 += acc * v0; 
    }

That plus inlining the iteration count so the compiler can unroll the loop might help get closer to SOL.

[clamchowder]

The problem is loop overhead matters on AMD, because AMD's compiler doesn't unroll the loop. Nvidia's does, so it doesn't matter for them.

[WithinReason]

unroll with #pragma unroll?