PCIe 2.0 Bottleneck?

[Derangedteddy]

Build:

CPU: i7 920 OC @ 3.80 GHz, HT Enabled, NZXT Kraken x61 AIO Liquid Cooler

RAM: 24 GB DDR3 1866 CS 10

Mobo: EVGA X58 SLI LE

GPU: Reference GTX 970 @ +230 Core, +580 Mem
Storage: 2x Samsung 850 EVO 256 GB in RAID 0

So I ran Steam's VR Readiness test and it seems that I have some strange results...

Nearly everyone else that has a GTX 970 in the forum has had "ready" status on their GPU, except me.  The only thing that I can come up with is that my GPU is being bottlenecked by the PCIe 2.0 connection belonging to my (seven-year-old) i7 920.  Can anyone confirm with their MSI AB graphs who has a PCIe 3.0 Bus and a GTX 970?  I am seeing that mine is only using about 50% of the GPU, but I'm not sure if this has to do with a frame limiter in the benchmark itself or if it's truly because PCIe 2.0 is bottlenecking my card.  One thing going against the bottleneck theory is that a higher OC results in slightly higher testing scores.

 

[79wjd]

The benchmark may just consider PCIE 2.0 to not be in spec, but PCIE 2.0x16 won't bottleneck a 980Ti let alone a 970. (and like you said, if PCIE 2.0 was the bottleneck, then overclocking wouldn't improve performance)

[Praesi]

The Benchmark has no frame limit.

[Derangedteddy]

3 hours ago, Praesi said:

The Benchmark has no frame limit.

That's what I thought.  Would there, then, be anything else that could be causing my GPU not to be used to its full potential other than a PCIe 2.0 bottleneck?

[Derangedteddy]

6 hours ago, djdwosk97 said:

The benchmark may just consider PCIE 2.0 to not be in spec, but PCIE 2.0x16 won't bottleneck a 980Ti let alone a 970. (and like you said, if PCIE 2.0 was the bottleneck, then overclocking wouldn't improve performance)

Perhaps, but the fact that my GPU was 50% utilized left me concerned.  However, now that I think about it, running other games allows my GPU to be maxed out 100% without V-Sync enabled.  Maybe it's just something weird about the benchmark itself, such as the fact that it's an extremely low-resolution benchmark?  Is there any way to force 1080p rendering in this benchmark?

[Praesi]

The Resolution is 2160x1200. Would not call that extreme low. Try to rerun the test. Use another driver or disable OC for a run. Can be several reasons. Hard to say from the distance.

[Derangedteddy]

Ok so I recently bought some upgrades to my machine (not for this reason, games like Cities: Skylines were slow on the i7 920), but I am using the same GPU (Reference GTX 970).  I re-ran the test and got "Ready" and the GPU is now green, with the overall system rating being near the top end of the range (7.9).  This is the same exact system image with the same drivers, etc.  In MSI AB, GPU usage is now pegged at 100% throughout the entire test.  Perhaps this could be a flaw in the X58 chipset?  After all, X58 was the first chipset designed by Intel after they stripped a lot of components out of the NB and migrated them to the CPU's IMC.  Anyways, here are the new specs:

 

(New) i7 6700k @ Stock 4.00 GHz
(New) 32 GB EVGA SuperSC DDR4 3200 RAM

(New) Gigabyte Z170X Gaming 5 Motherboard

(New) Samsung 950 Pro 512 GB PCIe M.2 SSD

(Old) Windows 10 x64

(Old) NVidia GTX 970 Reference

(Old) NVidia Drivers Version 361.91