**The OcUK Benchmarks thread!**

[LtMatt]

drum roll.....

And
The winner IS....

The 680...by a ball hair

I guess what we can take from this is that their is no 'clear' winner in this any more, though that AMD rep saying the 7970 out right beats a 680 is clearly wrong unless 12.10 is going to bring another 10% to the party.

Great results guys, very close indeed.

Re-run the test when the new drivers come out.

I think its a case of Nvidia worked closely with the devs on this game, i remember reading an interview about it with the game designer. I think AMD have been playing catchup ever since day 1 tbh but its great that they have closed the gap considerably and are still working on improvements with graphics core next.

 

[pgi947]

Yeah pretty much.

Here was my own Operation Swordbreaker run on a single card. From out the LAV to the end part where you clear the house with rockets...



The 680's average FPS seem (on average) a fair amount higher. I wonder why...

They do indeed, my last bench was from the time demo linked earlier (as was the scores taken from gregster).

My 7970 JC2S benches are up in the main thread, and for the lazy:

 

[Gregster]

There would be no chance of telling what GPU was used in either machine if you didn't use AB or similar. As much of an Nvidia fan I am, I can see AMD have done well with recent drivers and good to see

 

[LtMatt]

They do indeed, my last bench was from the time demo linked earlier (as was the scores taken from gregster).

My 7970 JC2S benches are up in the main thread, and for the lazy:

20fps onto the minimum fps, impressive. That's only a mild overclock.

There would be no chance of telling what GPU was used in either machine if you didn't use AB or similar. As much of an Nvidia fan I am, I can see AMD have done well with recent drivers and good to see

Good of you to say Greg, one of the least biased nvidia users on ocuk.

Its a strange title, but roll with it.

 

[pgi947]

There would be no chance of telling what GPU was used in either machine if you didn't use AB or similar. As much of an Nvidia fan I am, I can see AMD have done well with recent drivers and good to see

Exactly, with frame rates like that on either card you'd have to be looking at the FPS counter the whole time.

20fps onto the minimum fps, impressive. That's only a mild overclock.

Just Cause 2 seems to scale really well with the OC, to think I first played that game on a pair of 9800GTX+'s at around 40fps with med settings

Edit: Its worth pointing out that with either card on BF3, you can near double the frame rates by dropping HBAO and to 2xaa

 

[humbug]

Thanks pgi, JC2 looks ok to me. i was worried something was wrong but your ~30% faster than me.

I might run BF3 again later with just MSAA set to 2x.

 

[pgi947]

Just found this as well guys: http://downloads.guru3d.com/Benchmarks-&-Demo's_c6.html

 

[humbug]

gregster / Rusty / Mat / pgi ecte... When you say stock in your charts is that reference or branded clocks?

I'm board and putting together a preliminary stock GPU chart for myself, i know your doing one later and don't want to intrude on that.
I don't want to use factory overclocks as they are all different, for example calling mine stock (Gigabyte) is messy given that @ 1100Mhz its running 50Mhz faster than 70% of 7870's, and 100Mhz faster than reference.

Incidentally it might be a good idea if we all make benchmarks including (reference stock clocks) even if that means down clocking.

 

[Tonester0011]

gregster / Rusty / Mat / pgi ecte... When you say stock in your charts is that reference or branded clocks?

I'm board and putting together a preliminary stock GPU chart for myself, i know your doing one later and don't want to intrude on that.
I don't want to use factory overclocks as they are all different, for example calling mine stock (Gigabyte) is messy given that @ 1100Mhz its running 50Mhz faster than 70% of 7870's, and 100Mhz faster than reference.

Incidentally it might be a good idea id we all make benchmarks including (reference stock clocks) even if that means down clocking.

I know Pgi is doing reference stock clocks as he stated in his original post (and charts ) Not too sure on everyone else.
I'm also doing ref stock and my stable CF clocks (just easier, single card clocks higher...just not when CF'd)

 

[humbug]

Ok here is BF3, No stock GTX 680's in this yet. i put the EVGA GTX 680 in as FOC

If this is enthusiast to top end level we are missing GTX 660 / GTX 660TI / GTX 670 / 7950 / 7950 BE / 7970 GE.

Come on guys this is a role call, if you have one of the above get your backsides in here

 

[LtMatt]

I'll run some stock and overclock tests again when the new AMD drviers come out, should be this afternoon with a bit of luck.

 

[pgi947]

That 12.9 whql or 12.10 beta? I'll re-run vital games on new drivers also.

Running kepler at stock will be tricky due to boost, the only way around it is to disable boost and set static 3d clocks at the stated boost minimum. Not sure if you can set a negative boost?

Edit, it's 12.11 see final8ys thread will update tonight. Seems to gain a good 10-15% over 12.7

 

[Rusty0611]

That 12.9 whql or 12.10 beta? I'll re-run vital games on new drivers also.

Running kepler at stock will be tricky due to boost, the only way around it is to disable boost and set static 3d clocks at the stated boost minimum. Not sure if you can set a negative boost?

Edit, it's 12.11 see final8ys thread will update tonight. Seems to gain a good 10-15% over 12.7

Yeah you can set a negative offset.

 

[pgi947]

Then it should be possible to wind boost back to stock min boost, though the unpredictable boost is one of keplers charms.

 

[whyscotty]

IIRC you can turn boost off via Nvidia Inspector.

 

[Rusty0611]

Then it should be possible to wind boost back to stock min boost, though the unpredictable boost is one of keplers charms.

Yes you can. AFAIK the stock boost of 1066 is the bare minimum. I've not seen one boost to this out the box.

12.11 drivers look to have eroded the gap between the 680 and 7970 to a couple of per cent in BF3.

Shame it took so long but great news now for potential buyers and the existing owners. Hopefully this means the 8000 series when released will have far better release day performance.

 

[psychodil]

Isn't the stock boost of the 680 typically 1100mhz on average on reference cards? Perhaps a good idea to clock to that for 'stock benches'.

 

[Gregster]

Isn't the stock boost of the 680 typically 1100mhz on average on reference cards? Perhaps a good idea to clock to that for 'stock benches'.

I agree with this. When looking at all 680 cards on here, they all boost to 1100 or above. This for me is 'stock' and going any lower would be pointless.

 

[Twirlywoos]

Great thread. This needs to be stickied. Much prefer real life gaming benches than the synthetic and somewhat biased stuff you get from online reviewers (with the exception of Tom's Hardware love them guys). Needs to also include if people are using that Vitru MVP crap though.

 

[Gregster]

Great thread. This needs to be stickied. Much prefer real life gaming benches than the synthetic and somewhat biased stuff you get from online reviewers (with the exception of Tom's Hardware love them guys). Needs to also include if people are using that Vitru MVP crap though.

I would like to think that honesty is something being displayed by all users here and if they was to use it, they would say they are using it. It would be cool to see some results with that being used though