7990/Ares II Vs Ttian/690

[LtMatt]

Strolled into this thread a little late

100%+ scaling doesn't often happen, on 13.5 with bf3 we are approaching it though, 150fps avg at 1010mhz ultra pre set, sits around 70-80 fps on a single card - 1080 testing.

3dmark11 scales at around 80-90%, looking around 9000 stock single and 16-1700 on gpu scores.

Can easy try some of the other benches later, though some of those single to dual scores do seem a little off. I really wish in bench comparisons they would list the stock boost clocks of cards, makes the whole thing pointless otherwise.

Finally a friendly face!

Well google chrome translate is not working on that website at the moment for some reason but here is a list of the clock speeds.

Radeon HD 7970GE тестировалась на стоковых частотах и с разгоном Radeon HD 7970GE OC - по ядру 1160 МГц и по памяти 7160 МГц

Radeon HD 7990 тестировалась на стоковых частотах и с разгоном Radeon HD 7990 OC - по ядру 1100 МГц и по памяти 6500 МГц

ASUS ARES II тестировалась на стоковых частотах и с разгоном ASUS ARES II OC - по ядру 1200 МГц и по памяти 7200 МГц

GeForce GTX 680 тестировалась на стоковых частотах и с разгоном GeForce GTX 680 OC - по ядру 1150 МГц и по памяти 7000 МГц

GeForce GTX 690 тестировалась на стоковых частотах и с разгоном GeForce GTX 690 OC - по ядру 1020 МГц и по памяти 7000 МГц

GEFORCE GTX TITAN тестировалась на стоковых частотах и с разгоном GEFORCE GTX TITAN OC - по ядру 1080~1180 МГц и по памяти 6700 МГц


I don't think it also helps that they appear to be using a Nvidia beta driver which might have been pulled 320.20. Either that or its just a typo but i have found many references to it when searching google.

 

[andybird123]

I finally found something on it on the nvidia website - it's a notebook driver, under the skin the desktop part of it is the same as 320.00

 

[Kaapstad]

So they actually got 1 more fps on TR than your titan. Yikes they really are bad i better stop using them. Don't want the titans looking any better than they do already.

My Titans on water, it was just getting warmed up @1080p, that's why it's 1 fps better @1600p.:

 

[LtMatt]

My Titans on water, it was just getting warmed up @1080p, that's why it's 1 fps better @1600p.:

Furry muff

 

[Darujhistan]

Why are you now talking about something completely different to what I've posted about?

People are "crying" because the results don't look accurate based on what they are getting themselves. I think you're possibly in the wrong forum if you don't want to see debates about a few FPS here or there. Especially this generation when everything is so close .

But you didn't answer my question - what does it matter if x is 5 fps faster than y?

You've all got cards that can smash games at any settings at any res, isn't that what GPU's are for?

 

[Rusty0611]

But you didn't answer my question - what does it matter if x is 5 fps faster than y?

You've all got cards that can smash games at any settings at any res, isn't that what GPU's are for?

Read what I said above about the forum and you have your answer . It's still completely different to what we were originally discussing as well.

 

[Darujhistan]

Read what I said above about the forum and you have your answer . It's still completely different to what we were originally discussing as well.

That's not really an answer.

 

[Gregster]

Whether single GPU or dual GPU, whether Green or Red whether Titan or 7990, these cards are the very top end that play games at very high settings at very high res, right?

So why are people crying about a few FPS here or there on benchmarks?

What does it matter?

A few fps could be the difference between playable and not playable. It is all subjective and down to what each individual considers to be a playable rate but those few fps could have a deciding effect on the purchaser.

Some people are sensitive to micro stutter and dual cards invariably bring a certain degree of this and in some cases, SLI and CF do not have any support for some games and can infact have negative scaling and reduce a 7990 or 690 to perform less than the 7970/680 in some games. This is why I believe you should never compare SLI/CF with single core cards. It is fine to see how well dual cards scale compared to their single brothers but to compare a single core GPU to a dual core GPU and then say "The Titan is obsolete" is a very daft thing to say.

 

[Rusty0611]

That's not really an answer.

It is. Why does it matter? Because people are interested in it. If you aren't then don't let the door hit you on the way out .

 

[Gregster]

Strolled into this thread a little late

100%+ scaling doesn't often happen, on 13.5 with bf3 we are approaching it though, 150fps avg at 1010mhz ultra pre set, sits around 70-80 fps on a single card - 1080 testing.

3dmark11 scales at around 80-90%, looking around 9000 stock single and 16-1700 on gpu scores.

Can easy try some of the other benches later, though some of those single to dual scores do seem a little off. I really wish in bench comparisons they would list the stock boost clocks of cards, makes the whole thing pointless otherwise.

This is the thing. You take 3DMark scores and even with 100% scaling, no way would an Ares II score more than a 7970 Ghz card.

Kaapstad has a good understanding of overclocking and his single 7970 score was:

12921. 7970. 3970X @ 4.9 - Kaapstad
and with CF
19562. 7970. 3970X @ 4.9 - Kaapstad

As you can clearly see, 2x 7970's don't score anywhere near 39124 points. If we take just the GFX scores, it gets closer.

12477. 7970 - Kaapstad
And CF
21810. 7970 X 2 - Kaapstad

This equates to roughly 75% scaling and not even close to the 98% scaling in one test shown by GameGPU or the rather crazy 108% scaling in another test. This goes beyond the margin of error.

I appreciate it is hard to get unbiased bench results but I took what our guys did (Rusty/LtMatt/Meatloaf/Humbug/yourself/others) far more serious than some dodgy Russian site. Get the OcUK benchmark thread pumping and show how it is done fairly/honestly

 

[pgi947]

Ok I did a little test (vs 7990) on 3dmark11 extreme pre-set...I set my 7970's to 950/1500 as in the tests (stock 7990 clocks).

Their score: 5599

My score: 5259

I need to test a little more however, I have a hunch my primary PCI lane is for what ever reason, running at gen3 4x.

 

[Gregster]

This is my Titan run (single with no overclocks). http://www.3dmark.com/3dm11/6550358? Not matching up with GameGPU.

 

[LtMatt]

Ok I did a little test (vs 7990) on 3dmark11 extreme pre-set...I set my 7970's to 950/1500 as in the tests (stock 7990 clocks).

Their score: 5599

My score: 5259

I need to test a little more however, I have a hunch my primary PCI lane is for what ever reason, running at gen3 4x.

7990 stock clocks is 1000mhz on the core mate and 1500 on the memory.

 

[pgi947]

7990 stock clocks is 1000mhz on the core mate and 1500 on the memory.

Derp, I only read the header in the item description which says 950mhz, forgot the newer 7900's have boost clocks.

Retesting now...

Edit: Now 5491

The extra 100 points or so they got could be down to both chips being on a single card or, as I said earlier, my hunch that my primary lane isn't running at full speed.

 

[LtMatt]

Derp, I only read the header in the item description which says 950mhz, forgot the newer 7900's have boost clocks.

Retesting now...

Edit: Now 5491

The extra 100 points or so they got could be down to both chips being on a single card or, as I said earlier, my hunch that my primary lane isn't running at full speed.

It could be down to anything really. Thanks for taking the time pgi.