Can CPU affect Graphics Score in 3DMark?

[Snowmirage]

I'm working to build a new version of my previous 2 gamers 1 CPU build.

I'm running a series of benchmarks and want to make sure I'm interpreting the results correctly.

List of tests
 

1. GPU directly in a x16 slot with all 24 threads enabled (EVGA SR2 motherboard if you are curious)
2. GPU on a PCI-E extension cable with all 24 threads enabled
3. GPU directly in a x16 slot with only two cores and hyper threading enabled
4. GPU on a PCI-E extension cable with only two cores and hyper threading enabled

My tests so far have confirmed no difference in performance with either an RX 480 or a GTX 980 when comparing directly in a 16x slot or with an extension when using all 24 threads

My next test addresses how much (if at all) I will be bottle necking each graphics card (RX 480 or GTX 980) when running with less CPU.

When testing the RX 480 with all 24 threads 
Fire Strike Score: 10716
Graphics Score: 12776
Test 1 FPS: 61.86
Test 2 FPS: 50.41
Phy Score: 13596
Comb Score: 4241

When testing the RX 480 with 2 cores and HT (4 total threads)
Fire Strike Score: 6490
Graphics Score: 12881
Test 1 FPS: 62.38
Test 2 FPS: 50.82
Phy Score: 3476
Comb Score: 1898

I of course expected a drastic drop in the Physics and overall score as a result.  

The question I have is can I correctly conclude that the 2nd test with limited cores active was not bottle-necking the GPU because the Graphics score was the same (with in margin of error, actually higher in this case oddly)?