• 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.

Chiphell - AMD HD8000 series GCN2.0 in Q3

I strongly doubt they will rebadge. AMD typically push new chip designs out for their own sake as kinda prototypes of new chip arrangements and so on, like the way the 5>6 series went.

The 6850, 6870, 6950 and 6970 weren't particularly faster than the cards they were replacing, they were however slightly cheaper and with completely different chips.

I was joking because the rumour said 256 bit bus which is what the GTX 680 comes with.:D
 
AMD need to do something, looking at the 7970 it really ought to quite a bit faster then it is given it uses more power then a Geforce Titan. While AMD have lost the engineering battle on 28nm imo it has won marketing battle with aggressive pricing across all levels except the ultra high end and first rate gaming bundles.
 
Was taking GCN2.0 into account but obviously there could be enhancements that I'm not aware of. I find it very unlikely those specs would give that kind of gains overall tho even in a best case scenario outside of limiting benchmarks to high-res multi monitor setups.

Maybe you're reading it wrong?

20% more power per stream processor, around 15% more stream processors, and a higher clock speed?

I was joking because the rumour said 256 bit bus which is what the GTX 680 comes with.:D

Whoosh! Went right over my head :D

AMD need to do something, looking at the 7970 it really ought to quite a bit faster then it is given it uses more power then a Geforce Titan. While AMD have lost the engineering battle on 28nm imo it has won marketing battle with aggressive pricing across all levels except the ultra high end and first rate gaming bundles.

It doesn't really work like that, power usage in and of itself doesn't indicate much when comparing two very different architectures.
 
It doesn't really work like that, power usage in and of itself doesn't indicate much when comparing two very different architectures.

I'm aware of that but it shows just how effective Nvidias architecture is at this point in time. For me the gtx670 is an engineering marvel, theres never been a card that small thats offered so much performance. If they wanted to Nvidia cod drop the price of the 670 down to 200 which make all of amd's offerings from 7970 down to the 7850 redundant. Nvidia really did have the opportunity to really stick the boot into AMD this time around but have chosen to protect its margins instead.
 
I'm aware of that but it shows just how effective Nvidias architecture is at this point in time. For me the gtx670 is an engineering marvel, theres never been a card that small thats offered so much performance. If they wanted to Nvidia cod drop the price of the 670 down to 200 which make all of amd's offerings from 7970 down to the 7850 redundant. Nvidia really did have the opportunity to really stick the boot into AMD this time around but have chosen to protect its margins instead.

Not particularly, the Titan is also significantly larger than the 7970's GPU, and Titan chips want to use more power but they are artificially limited in software in how much power they can draw.

The 670 really isn't an engineering marvel either, there's very basic reasons as for why it's the size that it is.

256 bit bus, that means the PCB is less complex and needs less traces, cheaper to produce and also means it can be smaller.

Cheaper power regulation, less chips, and other components, cheaper to produce too, and won't get as hot, so could get away easily without any sort of cooling, which makes for a much smaller PCB.

This is also the reason why they have been gimped in terms of voltage control for potentially big overclocks, and the cards that did come with the voltage control were built on much better PCBs, like the lightning GTX680s and so on.

This is what nVidia has done to protect its margins, as well as majorly reducing the size of their top tier desktop chips by seriously gimping the FP64 performance to such a ridiculous amount that the GTX460 has greater FP64 throughput than a GTX690.

The reason there's never been a card that small with that sort of power is because they've never felt the need to slash the cost of manufacture so hard.

They would never bottom the price out of their stuff because a lot of people who buy nVidia associate price with performance.
 
I'm aware of that but it shows just how effective Nvidias architecture is at this point in time. For me the gtx670 is an engineering marvel, theres never been a card that small thats offered so much performance. If they wanted to Nvidia cod drop the price of the 670 down to 200 which make all of amd's offerings from 7970 down to the 7850 redundant. Nvidia really did have the opportunity to really stick the boot into AMD this time around but have chosen to protect its margins instead.

As spoffle said, different architectures.

While the GTX Titan is good (the same, not better than the 7970) the GTX 670 is not as good as that, it also only has 2GB of 256Bit vRam and much much much less powerful GPU Compute than the 7970.
hell the 7870 beats it in GPU Compute, all that makes a difference in power consumption.

And that shows in GPU Compute intensive games, the GTX 600 chockes on them.

IMO the GTX 600 is and never was a good GPU, the architecture may have been good in 2008 and games for 2008 while the HD 7000 series has far better future proofing.
 
I wouldn't say that the 6 series is a bad architecture, I think the way nVidia has used it is bad, putting it on a 256bit bus PCB, with poor power circuitry to intentionally limit overclocking, and gimping the crap out of the FP64 performance.
 
I wouldn't say that the 6 series is a bad architecture, I think the way nVidia has used it is bad, putting it on a 256bit bus PCB, with poor power circuitry to intentionally limit overclocking, and gimping the crap out of the FP64 performance.

Its not 'bad' but its not good either, with advanced lighting, shadows and Physics all rendered by the GPU's compute engine the one in the GTX 600 is a long way from up to the job.

AMD have set new standards in that which has allowed developers to move that along nicely.
 
Maybe you're reading it wrong?

20% more power per stream processor, around 15% more stream processors, and a higher clock speed?

l'll be quite surprised if the 25% on paper increase in compute capabilities translates to an extra 25% overall performance in gaming and likewise the extra efficiency of GCN 2.0 translates to a direct 20% in gaming benchmarks on a GPU at that scale. More realistically for those specs it will be somewhere between 25 and 30% over a range of games.
 
Maybe it was gimped on purpose to make the 700 series 780 and titan look immensely better and to try and justify the price it sells at. To me and many others, its a big mistake.

I'm sure companies like this have the roadmap panned out well in advance, maybe even years in advance and changing it when they feel they can make more money from us.
 
Its not 'bad' but its not good either, with advanced lighting, shadows and Physics all rendered by the GPU's compute engine the one in the GTX 600 is a long way from up to the job.

AMD have set new standards in that which has allowed developers to move that along nicely.

The core architecture is sound, the process used for whats actually in the 680 was never originally intended for a high end GPU which severely gimps it. The process the 680 would have been on was forked to gk100 then later trashed, gk110 was never intended to be used for GeForce.
 
The core architecture is sound, the process used for whats actually in the 680 was never originally intended for a high end GPU which severely gimps it. The process the 680 would have been on was forked to gk100 then later trashed, gk110 was never intended to be used for GeForce.

I highly highly doubt this. It just doesn't make any sense, we all know nVidia wanted smaller GPUs, at around 300mm² it makes sense for the 680 to have always been intended to be where it is now.
 
If this is true then the days of the 7950 being 5% behind the 7970 will be long gone. If this is actually true then we know what caused it, the titan.

EDIT

To clarify the titan has set a new performance level for other cards to keep up with. I think its safe to assume, at least according to some of these leaks that this is going to affect AMD's card structure and could lead to a potential new 7970 pulling further away from the potential new 7950.
 
Last edited:
If it's true, no one is to blame but AMD, they choose to do what they do.

I don't think it is true anyway, but we shouldn't get in to the habit of blaming the actions of companies on others.
 
If this is actually true then we know what caused it, the titan.

lol that's rubbish. Honestly, I read it all the time. I've nothing against having GPU brand preference but to say one companies pricing is determined by a product launched by its competitor is ludicrous. It might be influenced by the Titan/780 prices but AMD are still ultimately responsible for their own pricing. If it is true (which I doubt) then why not pour the same levels of disdain upon AMD instead of incoherently blaming NVIDIA and Titan just because it doesn't fit into your own brand preference. Crazy.

As said above by spoffle and quite succinctly: AMD are in charge of their own pricing.
 
Last edited:
Back
Top Bottom