AMD 7000 series, any news?

pcgamer420

Limp Gawd
Joined
Jan 5, 2009
Messages
197
Any idea when we will see the 7000 series from AMD? All I know is there will be a die shrink, will this mean a huge leap forward in performance, like 50%+ gains over the 6000 series? Will we see these cards out by Christmas of 2011? I'm just wondering if I should start saving my pennies for a brand new Ivy Bridge + 7000 series rig.
 
Wow, really, that fast? Yes! Upgrade time for me! :) Hopefully, Nvidia will follow with same time frame.
 
that link is BS. you are living in a fantasyland if you think the 7000 series will be out in Q2. volume 28nm production will not even be ready until late Q3 or likely even Q4.
 
50%?

hehehe, no.

But die shrinks are always great. Gives ATi time to eek more performance out of the archetecture, as well as increase the brute power and decrease power usage.

Historically I think these revisions average at about a 20% increase, unless some glaring error was fixed.
 
50%?

hehehe, no.

But die shrinks are always great. Gives ATi time to eek more performance out of the archetecture, as well as increase the brute power and decrease power usage.

Historically I think these revisions average at about a 20% increase, unless some glaring error was fixed.
its a new line of gpus so even 50% increase in performance would be a joke.
 
50%?

hehehe, no.

But die shrinks are always great. Gives ATi time to eek more performance out of the archetecture, as well as increase the brute power and decrease power usage.

Historically I think these revisions average at about a 20% increase, unless some glaring error was fixed.

I think you are contradicting yourself here. Its not a revision. Its is a die shrink.

Nobody was even talking about revision in this thread. Every single die shrink has yielded alot more than 20% in the past.

its a new line of gpus so even 50% increase in performance would be a joke.

I think the GT 280 to GTX 480 was around 40-50% , 55nm to 40nm.


Most likely Q4 this year and that is at the earliest. 28nm is giving all sorts of problems right now to TSMC and GF.
 
Last edited:
that link is BS. you are living in a fantasyland if you think the 7000 series will be out in Q2. volume 28nm production will not even be ready until late Q3 or likely even Q4.

Never mind that AMD's drivers won't hardly even be acceptable for public consumption 'til Q1/Q2 next year... (Q4 if they hire Nvidia's guys...)
 
It's two die shrinks since 40nm->32nm was scrapped and we're jumping to 28nm. You can fit double the transistors since 40^2 is 1600, 28^2 is 784~800. The performance jump can be huge if it scales correctly, 100% if perfect but I'm betting >60%.

Same applies to Nvidia.
 
This is an issue I have with upgrading now. Basically with two die shrinks the boost in theory should be huge, bigger then previous ones. However at the same time I need (well more like want haha) to upgrade this 5770. A GTX 560/6950/GTX 570 is probably what would feel like an upgrade, but the 28nm might make them feel so dated and kill the resell value.

It really seems like the perfect upgrade time as well, since the part you buy then will likely time your next upgrade into DX 12 parts.
 
If everything goes as expected, AMD next generation of video cards (Southern Islands aka Radeon HD 7000) will be released in the late Q2 of 2011 or early Q3.
Lolno. Completely false. I'd say between November 2011 at the absolute earliest, to, more likely, around January-March 2012, for 28nm HD7 series cards.
Performance per watt should be up around 40% on the HD6 series, so the 150W midrange card (where the HD6870 is now) will perform like the GTX580 does now. How much further up the scale will extend I couldn't say.
 
its a new line of gpus so even 50% increase in performance would be a joke.

How so?

I think you are contradicting yourself here. Its not a revision. Its is a die shrink.

Nobody was even talking about revision in this thread. Every single die shrink has yielded alot more than 20% in the past.

Whoops! Sorry, I meant the word "revision" as in the dictionary definition, not what it means when it comes to video cards. I only meant to imply that it's a new iteration of the card, that's all.

As far as die shrinks yielding more than 20%, I mean in terms of FPS. And if you could link me to an example, I'd be more than happy to admit I'm wrong.
 
How so?



Whoops! Sorry, I meant the word "revision" as in the dictionary definition, not what it means when it comes to video cards. I only meant to imply that it's a new iteration of the card, that's all.

As far as die shrinks yielding more than 20%, I mean in terms of FPS. And if you could link me to an example, I'd be more than happy to admit I'm wrong.

FPS is the only metric used (usually average across games fps) and the only one we are talking about here. No link needed just do a search yourself for every single die shrink transition in the past 5 years. You are the only one that made that claim against common knowledge. Wouldn't you say the burden of proof is on you?
 
How so?



Whoops! Sorry, I meant the word "revision" as in the dictionary definition, not what it means when it comes to video cards. I only meant to imply that it's a new iteration of the card, that's all.

As far as die shrinks yielding more than 20%, I mean in terms of FPS. And if you could link me to an example, I'd be more than happy to admit I'm wrong.
you cant be serious. a new generation of cards with a die shrink would be a failure if it only produced 20% better performance.
 
Considering there are only a few games that even challenge the 580 or 6970, even at 2560x1600, I doubt either company will rush into 28nm. The demand simply would not justify the risk. AMD and Nvidia have both, painfully, learned their lesson from that in the past. Early 2012 at best for either company.
 
Hahah wow! The hostility.

Okay okay, forget it.
I wasn't being hostile at all. I just cant believe you think going from 40nm to 28nm and having a new or revised architecture would only yield 20% improvement. just the die shrink alone would allow for way more than 20% improvement in clock rates at the same power usage.
 
Guys, all this discussion about performance and wattage about new generations made me interested in performance and watt scaling from generation to generation. I decided to do some math. I used the performance for all resolutions and average power consumption from here to see how different cards stack up against one another. For the perf/watt listing, I divided 'relative performance' with 'average consumption' to get the final value. For one thing, it shows that while you can make generalizations about die shrink performance, you can't do the same for perf/watt and consumption. It also shows outliers in efficiency for the 5870 and 480 compared to each of their predecessors. The 6870 outperforms the 5850, but the high clock rate makes it use more power relative to the 5850, so it isn't as efficient. Also, look the efficiency difference between the 480 and 580 compared to the 285. It's HUGE!!!

Relative Performance

5850 -> 6870 = 1.08

4670 -> 5670 = 1.24

4890 -> 5870 = 1.48

GTX 285 -> GTX 480 = 1.49

GTS 250 -> GTX 285 = 1.68

GTS 250 -> GTS 460 1GB = 1.69

GTX 285 -> GTX 580 = 1.69

4870 -> 5870 = 1.72



Average Consumption


5850 -> 6870 = 1.17

4670 -> 5670 = 1.1

4890 -> 5870 = 0.93

GTX 285 -> GTX 480 = 1.73

GTS 250 -> GTX 285 = 1.3

GTS 250 -> GTS 460 1GB = 1.02

GTX 285 -> GTX 580 = 1.53

4870 -> 5870 = 0.88



Perf/Watt


5850 -> 6870 = 0.92

4670 -> 5670 = 1.13

4890 -> 5870 = 1.59

GTX 285 -> GTX 480 = 0.86

GTS 250 -> GTX 285 = 1.29

GTS 250 -> GTS 460 1GB = 1.66

GTX 285 -> GTX 580 = 1.1

4870 -> 5870 = 1.95
 
Last edited:
#1. Don't expect new cards from AMD until the end of this year at the earliest. They haven't even released the 6990 yet...
They still need time to make money from their 6000 cards. I would expect Nvidia to be about the same amount of time, though if anyone releases cards sooner, it will be them.

#2. Multi-monitor gaming is going to be the new performance milestone. While the fastest cards easily tackle single panel gaming, they still struggle with multiple screens. Hell, look at Vega's thread. It takes $1800 of graphics cards to power 3x 30" screens. There is still a LOT of room for graphics cards to improve, even if we are being limited by consoles and the restrictions they impose on game development.
 
Troll bait :D
Don't take it.

Awww! But it's so much fun to feed the Troll! It's so easy to pretend to sound all indignant and insulted by the Troll's obvious baiting, and then watch the Troll spring into action like an Olympic athlete (with a big, hairy paunch of course) and try to pick apart your response. See, the Troll thinks HE'S controlling YOU, when in fact.... ;)
 
That looks to be for mobile. But desktop refreshes normally happen before mobile, right? Could be a good sign!

they could be using the mobile platforms to test the 28nm gpu's, just like ATI/AMD did with the 4770 being 40nm and released almost a year in advance of the 5k series to make sure the process actually works on a large scale instead of just a limited amount of Engie samples.
 
they could be using the mobile platforms to test the 28nm gpu's, just like ATI/AMD did with the 4770 being 40nm and released almost a year in advance of the 5k series to make sure the process actually works on a large scale instead of just a limited amount of Engie samples.

It makes perfect sense to use a die shrink on mobile parts first, due to the lower power usage...

I'm wondering how much of a market there will be for discrete mobile graphics - however - now that Llano and - by the time these are supposed to launch - possibly Trinity have been released.
 
Back
Top