PCI-E 3.0 vs PCI-E 2.0 vs PCI-E 1.1

[AdamK47]

I was curious to see what kind of difference I could see switching between the three modes. My motherboard, an Asus P9X79 Deluxe, has the ability to switch between Gen1, Gen2, and Gen3 modes. The results posted are from Metro 2033 with every game setting maxed at a standard 1920x1080 with PhysX effects disabled.

Specs:
Intel Core i7 3960X @ 4300MHz
Asus P9X79 Deluxe @ 43 x 100MHz
16GB Corsair XMS3 @ 2133 DDR
Three HD7970 in Tri-Fire @ 1100/1600
120GB Vertex 3 SSD
Two 3TB Deskstar 7K3000 in RAID-0 - 6TB
1TB SpinPoint F1
Pioneer BDR-206 BD-RW
Sound Blaster X-Fi HD USB
Cooler Master HAF-X case
Prolimatech Megahalems cooling
Corsair AX1200 power supply


PCI-E 1.1
Run 0 (Frontline)
•Total Frames: 5196, Total Time: 59.34303 sec
•Average Framerate: 87.66
•Max. Framerate: 248.57 (Frame: 4790)
•Min. Framerate: 13.55 (Frame: 2409)

Run 1 (Frontline)
•Total Frames: 5282, Total Time: 59.77334 sec
•Average Framerate: 88.47
•Max. Framerate: 255.82 (Frame: 5068)
•Min. Framerate: 13.79 (Frame: 2456)

Run 2 (Frontline)
•Total Frames: 5252, Total Time: 59.77408 sec
•Average Framerate: 87.96
•Max. Framerate: 252.72 (Frame: 4796)
•Min. Framerate: 14.80 (Frame: 2378)

Average Results
•Average Framerate: 88.00
•Max. Framerate: 255.82
•Min. Framerate: 13.55
-------------------------------------------------------------------------


PCI-E 2.0
Run 0 (Frontline)
•Total Frames: 5489, Total Time: 59.43546 sec
•Average Framerate: 92.45
•Max. Framerate: 245.82 (Frame: 5251)
•Min. Framerate: 16.52 (Frame: 3)

Run 1 (Frontline)
•Total Frames: 5515, Total Time: 59.83518 sec
•Average Framerate: 92.27
•Max. Framerate: 264.13 (Frame: 5375)
•Min. Framerate: 19.43 (Frame: 2385)

Run 2 (Frontline)
•Total Frames: 5523, Total Time: 59.83703 sec
•Average Framerate: 92.40
•Max. Framerate: 250.56 (Frame: 5357)
•Min. Framerate: 21.65 (Frame: 2547)

Average Results
•Average Framerate: 92.33
•Max. Framerate: 264.13
•Min. Framerate: 16.52
-------------------------------------------------------------------------

PCI-E 3.0
Run 0 (Frontline)
•Total Frames: 5602, Total Time: 59.43283 sec
•Average Framerate: 94.36
•Max. Framerate: 255.62 (Frame: 5102)
•Min. Framerate: 19.72 (Frame: 2623)

Run 1 (Frontline)
•Total Frames: 5608, Total Time: 59.85704 sec
•Average Framerate: 93.79
•Max. Framerate: 259.34 (Frame: 5095)
•Min. Framerate: 15.50 (Frame: 2573)

Run 2 (Frontline)
•Total Frames: 5630, Total Time: 59.85656 sec
•Average Framerate: 94.16
•Max. Framerate: 261.71 (Frame: 5240)
•Min. Framerate: 15.61 (Frame: 2588)

Average Results
•Average Framerate: 94.00
•Max. Framerate: 261.71
•Min. Framerate: 15.50

 

[AdamK47]

Another run at PCI-E 3.0:


METRO 2033 BENCHMARK RESULTS
1/15/2012 8:24:43 PM
frontline

--------------------------------------------------------------------------------
Options: Resolution: 1920 x 1080; DirectX: DirectX 11; Quality: Very High; Antialiasing: MSAA 4X; Texture filtering: AF 16X; Advanced PhysX: Disabled; Tesselation: Enabled; DOF: Enabled

Run 0 (Frontline)
•Total Frames: 5529, Total Time: 59.47736 sec
•Average Framerate: 93.06
•Max. Framerate: 253.94 (Frame: 5088)
•Min. Framerate: 15.04 (Frame: 2549)

Run 1 (Frontline)
•Total Frames: 5604, Total Time: 59.86073 sec
•Average Framerate: 93.72
•Max. Framerate: 255.49 (Frame: 5278)
•Min. Framerate: 18.78 (Frame: 2649)

Run 2 (Frontline)
•Total Frames: 5611, Total Time: 59.85937 sec
•Average Framerate: 93.84
•Max. Framerate: 255.30 (Frame: 5488)
•Min. Framerate: 16.79 (Frame: 2598)

Average Results
•Average Framerate: 93.33
•Max. Framerate: 255.49
•Min. Framerate: 15.04


The minimum seems to be taken during the loading between benchmark loops. I don't see those minimums anywhere during the actual playback of the benchmark demo.

 

[96Firebird]

I thought PCI-E 3.0 is only available with upcoming Ivy Bridge CPUs?

Edit - Seems X79 and SB-E support it as well... http://www.anandtech.com/show/5264/sandy-bridge-e-x79-pcie-30-it-works

 

[TakeNoPrisoners]

I think everyone can agree that no game would need 3.0. It is suprising however that 1.1 didn't show much of a drop in performance.

 

[AdamK47]

Yeah, it seems a lot of people forget SB-E / X79 has PCI-E 3.0

Here's 3DMark 11.

PCI-E 1.1

PCI-E 2.0

PCI-E 3.0

I'll see if I can do more tests with this later.

 

[AdamK47]

Last one for tonight. Batman Arkham City.

Settings (PhysX Off)

PCI-E 1.1

PCI-E 2.0

PCI-E 3.0

Kinda funny that only one frame seperates each on average.

 

[notty22]

You don't have F1 2010 do you ? That game seemed to like Sandy Bridge and the nf200 chip , allowing more pcie- bandwidth, in a Hocp article.
http://www.hardocp.com/article/2011/05/03/nvidia_3way_sli_amd_trifire_redux/2
F1 2010

​

What you are seeing above is no joke, we triple checked our performance to make sure this was correct, and it was. Our new 4.8GHz test system was a whopping 31.3% faster than the old system with the same GTX 580 3-Way SLI configuration using the same drivers. 3-Way SLI was being held back a lot with our previous system. The added CPU frequency has really brought out the true beast of performance GTX 580 3-Way SLI can be apparently, and it didn't stop there.

 

[Grooveriding]

Can you try a compute oriented bench ? I think this one fits the bill http://www.luxrender.net/wiki/LuxMark

Anand said he saw the greatest benefits from PCIE 3.0 in compute runs: http://www.anandtech.com/show/5264/sandy-bridge-e-x79-pcie-30-it-works

 

[ThatsABigOne]

Very informative benches. I presume that PCIe 2.0 at 8x is same as PCIe 1.1 at 16x?

 

[BrightCandle]

Very informative benches. I presume that PCIe 2.0 at 8x is same as PCIe 1.1 at 16x?

In bandwidth terms it should be near identical. There are tweaks and such between them that its not identical but yes that is the way it basically works.