• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

[TechSpot] The Best Graphics Cards: Nvidia vs. AMD Current-Gen Comparison

Yes he does your quite right.
So lets look at some extra high resolution results from here

http://www.techpowerup.com/reviews/AMD/Catalyst_12.11_Performance/1.html

2 things:

a) no overclock at all so if the 7950s are at stock settings it does represent somewhat of a false comparison
b) in some of the tests they put the 660Ti quite close at this resolution which I find hard to believe

Gregster knows he can get better performance from a pair of 7950s and has accepted it. It doesn't really matter to him now does it? He's had superior performance for months before 12.11.

As I said while testing the difference between the two can be summarised between playing at maximum settings and not. Quite impressive for a £240 card no?

The gap for gregster is smaller as he has better 680s but it will still be a big gap as my own 680s were pretty golden.

Stop posting things to get a negative reaction Rusty, it's not cool!

(it looks like I've got 2 good 7950s, they seem fine doing 1200/1600.)

Nice one. Mine can do 1200 but I don't really need the extra for a single screen so run at 1100 and 1500. My card gets a little cranky above 1550.
 
Last edited:
Could really do with some OC'ing. Based on my 7850, they should use a 1200 core and 1450 memory clock at stock voltage. I makes a huge difference, it would give a stock 7950 a run for its money (until that was overclocked that is!).
 
now realise that my 680 or both of them are more than capable cards and I did have 10 months of stomping on the 7970s :p (last bit for you Tommy)

When did you get the Lightnings?:p

Since release, they were always ~even unless both were@stock clocks, both oc'ed the 7970+680 didn't have much to split them at all, with the performance drivers, the 7970 just edges it now when both@stock.

He's had superior performance for months before 12.11.

As above, it doesn't paint the correct picture though.

12.8 betas brought the initial shift in performance back in July-when I decided it was time for the change, the 12.11's just took them that little bit further.

:)
 
It's only really relevant to people buying now, HOWEVER you could sell your GTX680s and get 7950s, make a profit and have more performance.

So in that regard, I think it's relevant for people to say, though I know it's very unlikely that you'd consider selling your 680s anyway.

I seriously considered this for a spell and I would have gone for 2*7970s instead (purely for the benching). After common sense kicked in, I realised I have 2 of the best 680's available and games over 1 or 3 screens run fantastic. I generally play games with no fps counters running or even any overclocking (I forget to start AB up). They run smooth enough for my liking with most of the games on ultra detail.

I also love pushing my hardware as far as it can go and I do hold the fastest SLI 680's in the 3DMark11 thread :)

And I don't tend to game so much in 3D as I used to but I still realy enjoy when I do, so this keeps me Nvidia bound for now.

Gregster knows he can get better performance from a pair of 7950s and has accepted it. It doesn't really matter to him now does it? He's had superior performance for months before 12.11.

As I said while testing the difference between the two can be summarised between playing at maximum settings and not. Quite impressive for a £240 card no?

The gap for gregster is smaller as he has better 680s but it will still be a big gap as my own 680s were pretty golden.



Nice one. Mine can do 1200 but I don't really need the extra for a single screen so run at 1100 and 1500. My card gets a little cranky above 1550.

I have known since day one and possibly chose to ignore the fact that the lower memory bus on the 680's would be a deciding factor when playing across 3 screens. I expected 2 * 7970's to beat me but wasn't expecting 2 * 7950's to beat me if I am honest...That kinda hurt a little :D

I have no intention of swapping out 'The Precious' though, as they can clock much higher than other 680's I have seen and I am not representative of normal 680's IMO.

When did you get the Lightnings?:p

lol bro, I knew you would have something to come back with :D. My Original 680 from EVGA was the first released batch and was pretty special. It clocked up very nicely and held the fastest 680 in the Heaven thread for a long time. It was only after seeing OptimaLNRG's score, I decided to get a lightning for SLI and 3 screens and then seeing how well one clocked, I gambled on getting a cherry second (which worked). My second isn't as fast as my first but it isn't far off.

Banter aside though, I still go back to my statement early in the thread and say without a fps monitor running, Joe Bloggs would have no idea what GPU's were in which machine when gaming.

Edit:

Rroff hits on a point which is valid, The latest Beta drivers are not 'performance drivers', they are broken betas and give less performance in BF3 (I checked these against 304.79 which seems to give the most performance in BF3 in SLI). Hopefully they can pull something out the bag and not sure what is left in the tank to do this.
 
Last edited:
Don't worry Gregster just wait for Nv to release their Never Doubt drivers with 20% performance boost lol

It is only stupid graphs that made me consider swapping in the first place. If I did swap out for 2*7970's, I would probably end up smashing something, as AMD/X79/CF does not work on certain systems and is a complete fail. AMD are aware of the problem and have been for a number of months but have done nothing about it. I will stick to my system that works flawlessly :)
 
The thing is before the latest drivers AMD weren't THAT much faster at triple screen resolution.

Faster undoubtedly but not as much as people liked to make out especially when both were overclocked.
 
One thing that would I feel has become even more apparent as the time has gone on, is that AMD really did release the 7970/50 well under their potential seeing as now both cards have had a faster version released and both still have a good amount of overclocking headroom.
Makes me wonder if they did it deliberately.
 
To the people who have regreted buying a 670 over a 7950, spare a thought for me who paid £480 at relaese day for a 7970 when today for the same money i could buy 7950 crossfire and have change left over. Sometimes its just not worth thinking about
 
One thing that would I feel has become even more apparent as the time has gone on, is that AMD really did release the 7970/50 well under their potential seeing as now both cards have had a faster version released and both still have a good amount of overclocking headroom.
Makes me wonder if they did it deliberately.

It came out early because of Amd's secret power delivery that enables a lot more cores to be used instead of being thrown out.

That was also the reason for the low stock clocks too.

The AMD tech on overclock forums stated that most 79's will clock between 1000-1100MHz, but wouldn't go into any more detail as it was a 'secret', Asic quality is a hint on how it's achieved.

If they didn't incorporate the circuitry, it would have taken a lot longer to amass the chips needed to produce the numbers needed for release-much like what happened with Nvidia with the low yeilds.

To the people who have regreted buying a 670 over a 7950, spare a thought for me who paid £480 at relaese day for a 7970 when today for the same money i could buy 7950 crossfire and have change left over. Sometimes its just not worth thinking about

You were bumped.:(

I could have got one for £399 delivered on release, others here got theirs for £379.
 
I think i paid £460 for release 7970. I'll double check for exact price in a second.
I was happy when i bought the card, if i wasn't happy to shell out i wouldn't have bought.
I also don't look back and think "i could have got this, that and the other if i waited 10 months"

Edit: £461, not a bad guess.
 
Last edited:
Nice article. Makes me feel quite good about getting a 7870.

It's a card that seems to fall between 2 stools (i.e. the 7850 and 660ti). The main competitor is the 660, which in the text they say the 7870 is better value, but in the 'top pick' table they choose the 660.

Still, I'm pretty happy with my 7870's performance.
 
Last edited:
The AMD tech on overclock forums stated that most 79's will clock between 1000-1100MHz, but wouldn't go into any more detail as it was a 'secret', Asic quality is a hint on how it's achieved.

He said almost all 7970's will do 1100-1200 core tommy, i remember it well. :)
 
He said almost all 7970's will do 1100-1200 core tommy, i remember it well. :)

It could be that it's the chips themselves that will do that, but the board they are mounted on determines whether it will hit it or not.

The same way the same CPU might hit different overclocks on different motherboards.
 
It could be that it's the chips themselves that will do that, but the board they are mounted on determines whether it will hit it or not.

The same way the same CPU might hit different overclocks on different motherboards.

After a bit of digging around i dug up the exact post and question.

I213p.jpg
 
It could be that it's the chips themselves that will do that, but the board they are mounted on determines whether it will hit it or not.

The same way the same CPU might hit different overclocks on different motherboards.

Apart from all reference board are the same, other than the core itself.
The aftermarket boards are another matter, but they still dont guarantee a good overclocker as it's still down to the silicon lottery.
 
Back
Top Bottom