• 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.

7970 After one day

So you are telling (forget benchmarks for a second) that in-game performance is as good if not better than SLI 570?! That's nuts. The problem I have with that though, is that it would be cheaper for me to buy another Gigabyte SOC 570 and SLI it than buy an AMD 7970. I could sell my current one and put it towards the 7970 but would the 7970 be as cool and quiet as my Gigabyte SOC (which by the way is almost totally silent idle and barely audible under load and keeps temps 70c and below while overclocked)?

The thing that puts me off SLI 570 as others have stated before with crossfire and SLI with other cards, is that you get the hassle of getting it set right. For some it works effortlessly without a hiccup then there are others who have been plagued with issues from day 1. Seems to me like it's 50-50 chance of going either way - something that I am finding more and more reluctant to gamble my money on with another 570.

I will be keeping my money for now, in the hope the 7970 will probably drop price soonish, and the anticipation of the Kepler rumours being true. Can't say that I am not tempted though, especially for Skyrim and Crysis 2 mods.
 
It also is that much faster, lucky that isn't it ;)

For the record, I said 70-80% faster over the 6970, not 50% faster than the 580gtx.

Take a reference 7970, overclock to max, take a reference 6970, overclock to max.......... what do you get, between 70 and 100% faster depending on the game, with the 100% being a rare case the average is likely somewhere around the high 70% mark.

Again, as with every other thread, if the 6970 was released but arbitrarily set to a 180w TDP with 30% slower clocks..... then the stock 7970 would have been 70-80% faster than the "underclocked" 6970.

The 7970, the CORE, the speed it can reach, the actual difference in games at the ONLY comparible settings, it IS around 80% faster than a 6970.

It's lucky for you they overclock so well as you were forecasting a 70-80% performance gain purely down to the architecture, which isn't the case.

It will be completely unsurprising when the 7970 releases with clock speeds between 950Mhz and 1Ghz, and even less surprising will be, if you overclock just about any gpu from 950 to 1Ghz, a sub 5% overclock, you would see MAYBE 3% in real FPS gain, which other than being not noticeable in use, also pales in comparison to a 70-80% performance gain from shader/architecture, memory clock, memory bus, and everything else.

IE clock speed is all but irrelevant in this situation, let's say you expect 1Ghz clock speeds, and get 900Mhz, because the card won't always be clock speed limited, this will likely make barely a 6-7% overall difference in the speed of the card........ or less than 10% of the total expected improvement in card speed.

Clock speed is the LEAST important difference in terms of the specs in a new gen card.
 
It's lucky for you they overclock so well as you were forecasting a 70-80% performance gain purely down to the architecture, which isn't the case.

And if the architecture dictates its designed for a different clock speed? I made the assumption that the clock speed range the chip was designed to operate in was 900-1000Mhz, it clearly isn't.

Chips don't overclock 40%, its really this simple, no matter how much you overclock a 580gtx on ANY air cooling it is not 100% faster than a 6970 .

The chip is underclocked, the architecture and design is 70-100% faster in so much as, when you take the chip you can play games and have a 70-100% faster framerate so I'm crazily calling it that much faster.

Clockspeed is STILL unimportant as I said. the difference is if you randomly remove 30% of it, the card is slower. in the example I said a 5% overclock wouldn't make much difference, which is true, or are you claiming that the 7970 overclocks by 5% and gains in performance far more than the clock speed? Because that is what I was saying.

I don't see me stating anywhere in there that a 6970 underclocked by 30% would barely perform differently.

We were working off the info we had, without any leaks to suggest it was vastly underclocked and capable of releasing easily at 1150Mhz, the assumption was that the architecture was designed to work at 900-950Mhz, it was clear as day(now we've seen it) designed to work quite a bit faster.

If you design a chip for 300Mhz clock speeds with crazy high IPC, or 1600Mhz and a lower IPC, those are architecture and design choices.

I didn't predict every single unreleasing detail of the chip and was working off the assumed info.

However the only claim I made was it will likely be 70-80% faster, it is, boo hoo for the pendantic.

Not that I can be bothered to look up which thread or the post you got that from because I'm not a pedant like some but I think you'll almost certainly find a post claiming that without knowing the clockspeed you couldn't possibly know how fast it was and that my post was an EXAMPLE of how clock speeds don't make a hell of a lot of difference.

IE had the chip come out and aimed for 1200Mhz, and only hit 1150Mhz for release...... that would be a VERY small portion of the overall speed gain.

Now lets think about those results on [H] 100% faster at 1.3Ghz in deus ex than the 6970, the 6970 can overclock circa 7-8% on the reference cooler most times. So lets call it 90% faster in Deus ex........ now lets knock 5% clock speed off the 7970......... so its 85.5% faster than the 6970....... oh know, my example was right, but the pedant in you wants to post me out of context, misrepresent my post and call foul.

Should also point out that, again my post indicated what a "real" stock speed is, something within 5-10% of the max overclock achievable on the reference cooler.
 
Last edited:
I have had a 7970 for a few days and my opinion is if you game at 1080p or less the 7970 is not worth the money as the 6970 is more than capable at 1080p or under. If you like your multiscreen eyefinity setups the 7970 is the card to have. I can almost get the same performance from a 6970 @ 1080p on the 7970 @ 5760. Which as I have said before is quite amazing.
 
I'm not trying to pick sides but Surveyor...did you fail to understand what you quoted or are you just trolling?

Thanks for not taking sides yet accussing me of not understanding or trolling.

DM said in the earlier post I quoted that he expected the HD 7970 to be released with clock speeds of 950MHz to 1GHz, the 70-80% increase would come mainly from the architectural improvements and that clock speed is the LEAST important difference in terms of the specs in a new gen card.

He was forecasting that a 950MHz to 1GHz HD 7970 would perform 70-80% better than a HD 6970.

That turns out not to be the case and a large part of the performace increase does come from increased clock speed.

A HD 7970 at around 1.2-1.3GHz performs something like 70-80% better than a HD 6970.

All I've pointed out is that DM was pretty much right just for the wrong reasons.
 
If the 6970 was shrunk onto the 28nm process, would it not be realistic to expect similar overclocks as for the 7970? At the same time, simply increasing the 6970's shader count by 30% and increasing it's memory from 256 to 384 would also make a nice difference.

I put it forward that the 7970 is merely a tweaked 6970 on a new 28nm process which allows space for beefed up shaders, memory bandwidth, and nice clocks. I do not see too many architectual changes that benefit real world gaming, other than from the process change itself. AMD did not invent and does not own the 28nm process. They merely take advantage of it. The same will probably apply for NVidia's Kepler when it eventually arrives in the form of a GTX580 on steroids.
 
I have had a 7970 for a few days and my opinion is if you game at 1080p or less the 7970 is not worth the money as the 6970 is more than capable at 1080p or under. If you like your multiscreen eyefinity setups the 7970 is the card to have. I can almost get the same performance from a 6970 @ 1080p on the 7970 @ 5760. Which as I have said before is quite amazing.

AMD have usually done very admirably at the very high resolutions with their top end gpu's. It's almost like it finds that extra gear. This reminds me of when not long after I bought my 4870X2, I spoke to another user that thought it really came into its own when he used it at a very high res. Though not relevant here, I do remember being surprised that it could perform so well what with the its lack of vram for 1440 / 1600p.

Anyway, I digress, glad to read that you're lovin' the performance. :)
 
Last edited:
Woah , wasnt expecting this amount of replies and im not trolling or picking sides.

The simple truth of it is i didnt pay 560 or anything close for it and im going to be getting the same money from selling the 570s as i paid for the 7970.

The heaven benches were just to show the difference is all, my card is clocked and runs fine in both bf3 swtor and also skyrim. With everything on max.
 
I can now see why you did go for the 7970, particularly when you factor water cooling in. I'm sure it will bring you hours of gaming bliss. :)
 
Last edited:
I will be posting my attempt at doing it on here also so im sure the people that disagree can have a good laugh at it if i get it wrong! :P
 
Woah , wasnt expecting this amount of replies and im not trolling or picking sides.

The simple truth of it is i didnt pay 560 or anything close for it and im going to be getting the same money from selling the 570s as i paid for the 7970.

The heaven benches were just to show the difference is all, my card is clocked and runs fine in both bf3 swtor and also skyrim. With everything on max.
 
Again its because this card is HEAVILY underclocked at stock.

I say it in reply to every fanboy statement, take a 580gtx, whack a weird(I give AMD zero credit for going with a low TDP limit and dropping the clocks) TDP on it, underclock it by 30% and then see how fast it is. Infact if you underclocked the 580gtx by the same amount as a 7970...... then compared a underclocked 580gtx with a stock(underclocked) 7970 and you'd see similar results again 50-60% performance difference.

AMD would have looked a WHOLE lot better if they simply released this card as a 300-325W 1150Mhz card WITH a low power setting in drivers that went to 225W, lower clocks and so on...... essentially had a low power mode for when you don't need that performance..... there are plenty of games out there that don't need the performance and plenty that do.

It's essentially like if Intel released the Sandybridge at 2.5Ghz for no apparent reason, but it easily did 3.3Ghz on stock and 3.9Ghz turbo..... same chip, just different clocks and odd benchmark results in reviews. The "real" speed of a Sandybridge wouldn't have changed. It really is terrible marketing/plan by AMD.

Look, I tend to like your posts because if nothing else they're articulate and erudite. You seemingly know your onions, both from a historical as well as future envisaged technical perspective.

What I take issue with is this:- greater minds than both yours and mine were the individuals that architected and released the 7970, and in doing so they chose a low TDP limit along with dropping the clocks. This strategy would appease those that are after the relative bang/watt scenario, but using your Intel analogy above, it quite literally DOES NOT MAKE SENSE. Why would Intel release a chip at 2.5Ghz if they were aware it had massive headroom to be stock clocked much higher, ***especially*** if they had direct competition at that performance point that might cause hesitation within the buying public from going with their solution? On top of that, why would they charge MORE than their direct competitor in that performance bracket for that same chip?

I'm afraid the answer is something that may be elusive to both of us, but I'll say this - AMD released the 7970 at the clocks they did for a very specific reason. Thus far I've not seen a plausible argument you've made for this - only conjecture that would appear to be your own ruminations on that very exasperation. Regardless, I'm in agreement that in a heavily overclocked scenario the 7970 is definitively a GTX 580 slayer - however there are vast tracts of the gaming populace that would baulk at doing so, due to it at this junture at least, being an unknown quantity. Therein lies the fence-sitting demographic who don't really think the 7970 is honestly all that - of which I unfortunately count myself amongst.
 
Last edited:
And if the architecture dictates its designed for a different clock speed? I made the assumption that the clock speed range the chip was designed to operate in was 900-1000Mhz, it clearly isn't.

etc. etc.

As you went on so much about the performance increase before the cards were launched you should expect what you said to be remembered.

As they say, if you can't take the heat stay out of the kitchen.

It turns out you were right, but for the wrong reasons, and there's nothing pedantic about me pointing that out.

See my reply to MichaelM.
 
Companies tend to release what they need to release either to beat the competition or win on some lesser count like power/heat/noise. Intel feel no need to release cpu's running at 5-6GHz as releasing something at 3.x gives them scope for something better later on that the 99% who don't or cannot overclock will lap up - and if we happen to get there in 4-5 steps of 200-300MHz each then thats even better for the bottom line.

AMD have released the 7970 at what they wanted to, admittedly perhaps slightly lower than where they should have - 1GHz would have had lots of kudos like the race they won to release the first 1GHz CPU and still leave room for them to release a faster one at 1150MHz or so and some exotic models at upto 1300MHz from Partners.
 
Back
Top Bottom