3D Mark 11 Results

[EventHorizon27]

So, now that I've had my new PC for a while it's time to start benching for the OC3D team

I've decided to start with 3D Mark 11.

My specs are: (for 24/7 use)

Intel i5 2500k @ 4.8GHz

Asus p8p67 pro

G.Skill 2x4GB 2133MHz

2x Asus GTX580 Matrix @ Stock

2x OCZ Vertex 2 (RAID 0)

At these settings I scored a depressing: P11166

It also warns that my system scored lower then it should compared to others of similar spec, pointing in the direction of the graphics. That's ok I thought as I hadn't overclocked them yet.

I went from:

GPU 816

Memory 1002

Shader 1632

To:

GPU 919

Memory 1151

Shader 1838

This took my score up to..... P11192

At this point I can only assume this is due to the cards maxing out the bandwidth of the x8/x8 slots they are in.

Do you think If I went up to a board that gave me x16/x16 it would give significantly higher results?

I'm also hitting below 20FPS occasionally in Crysis 2 and would like to stay above 30FPS at all times.

Thanks for any help or advice you can give.

 

[AdrianD2]

Is it possible that the SSD's are messing things up?

 

[EventHorizon27]

Is it possible that the SSD's are messing things up?

Possible? I guess so, I'll go test them quick.

 

[EventHorizon27]

Read test completed with the only tool I could find that I haven't already used the trial for.

Max read: 412.68MB/s

Access Time: 0.23ms

So It's starting to show some slow down due to lack of TRIM support over RAID but not enough to effect things to a noticeable level yet.

Any other ideas?

 

[ruthless_villain]

sli drivers?

 

[yuri04]

It also warns that my system scored lower then it should compared to others of similar spec, pointing in the direction of the graphics. That's ok I thought as I hadn't overclocked them yet.

Is that the site itself? if so, it does the same on World records

don't worry about that.

You should however get better FPS in Crysis I think (not sure, don't have the game nor de graphics cards

)

about the score...

looking at HWbot, your scores are in the ballpark though

Source

 

[EventHorizon27]

sli drivers?

I'm using 280.26. I'm going to have to wait for the drivers to get approved before I can submit to HWbot.

Is that the site itself? if so, it does the same on World records

don't worry about that.

You should however get better FPS in Crysis I think (not sure, don't have the game nor de graphics cards

)

about the score...

looking at HWbot, your scores are in the ballpark though

Source

Its the 3D Mark website that's telling me that there may be a problem.

My FPS is what id expect with Crysis as I'm playing in 5760x1080. I'm just trying to weigh up if getting a new motherboard is worth it. If it helps with benching then that's great but it needs to improve my real world gaming performance to be worth the money. The HWbot scores have made me feel a bit happier but I was still expecting my scores to go up alot more once overclocked. I'm going to try 1x 580, stock vs overclocked to see if that makes a bigger impact.

Thanks for the help so far

 

[EventHorizon27]

So it seems that how I imagined overclocking your GPU would effect your score was incorrect. It doesn't really make a difference to your P score.

2x GTX 580s (x8/x8)

Stock = 11166

12% overclock = 11192

1x GTX 580 (x16)

Stock = 6569

12% overclock = 6572

There was me thinking a 12% GPU overclock would give you at least a 5% increase in score... silly me

So, is submitting a bad bench better then not submitting one at all?

 

[S1EJ0]

Those single 580 figures don't make sense 3 points difference!

I'm sure I get a much bigger difference from stock and OC'd I'll run some benches later and post results.

Oh and no I doubt you will see significant increases with a x16/x16 board.

 

[Stromatolites]

I'm using 280.26. I'm going to have to wait for the drivers to get approved before I can submit to HWbot.

You don't need to wait for the drivers to be approved. That is only for 3dmark website.

HWbot will accept anything and everything as it stands.

Edit: As for your over clock results, its a number of factors that influence the score, some gpu and memory clock variations will cause a lower score than say a lower gpu clock and a higher memory clock.

Benching is about finding the best set balance of individual component over clocks for each test. It takes time.