AMD to launch new flagship Radeon HD 7970 desktop graphics card on December 22.

[Olikan]

Unless AMD design team knows some trick nVidia doesn't, I have to believe these new cores are crap compared to a fermi core. My guess is Fermi can complete 2 times the work per core than one of the GC. I hope I wrong, though. This would probably make sense since Fermi 512 @ 1600Mhz = 1024 @ 800Mhz.. Just throwing numbers out there. Meaning, I think a Fermi core might be capable of performing nearly twice the work. I hope I am wrong... I probably am wrong. All these leaks! I can't wait to see the release....

yes....but in gpu word, this is actually a good thing.
fermi is 30% bigger than a cayman, with only beeing 10% better

if a cayman were 30% bigger than a fermi, it would be at least 30% better

 

[Zargon]

Even CF 6950 spank gtx590 for 2/3 the cost. Add some AA and the gtx590 becomes a slideshow at 3x1080p. So, no, CF 7950s onwards will steam roll.

but still not be enough to do high res eyefinity at decent quality

I really dont want to have to do trifire, but if performance isnt high enough its cheaper to go tri with one new card than go dual new more expensive cards

got a 1kw kingwin in this week, should be enough for 3x6950's even if they are unlocked

I am not arguing that the 590 wouldnt be a better deal, just looking at the big res picture on what is and isnt doable with new cards

and for most people 50% perf increase probably isnt going to let them get what they want over xfire

it MIGHT though. depends on what the 7950 will do, and how its priced, as its usually the better deal for going crossfire

 

[tviceman]

any explanation as to how they fit 4 x 8 (32) ROPs with a 6 channel memory bus?

What if the memory controllers are 96-bit instead of the usual 64-bit? Is that even possible? I admit I have a large ignorance when it comes to the more intimate technical details of various computer hardware.

 

[notty22]

yes....but in gpu word, this is actually a good thing.
fermi is 30% bigger than a cayman, with only beeing 10% better

if a cayman were 30% bigger than a fermi, it would be at least 30% better

There is no way to say whether that would be true, unless they made this mythical gpu. When AMD shrunk the die size with Barts, this logic was mistakenly used as some new baseline to foretell performance of Cayman.
With its new larger die size. It turned out differently.

 

[Lonbjerg]

Never really liked the VLIW idea,.. think its a good thing their going with a approch more akin to nvidias.

If it helps them get better GPGPU/OpenCL and tessellation at the cost of slightly bigger chips, slightly more power use, im fine with that.

From:
http://www.anandtech.com/show/4455/amds-graphics-core-next-preview-amd-architects-for-compute/2

VLIW wastes 1.6 out of 5 on avg. (~32% waste)

Those games Radeon cards do bad in? probably games where VLIW is very wastefull.
Graphic core next, will fix that. Performance will go up drastically in those cases.

GCN = shaders that dont waste anything (more consistant gameing / better min fps?) = great.

I dont care if you can squeeze in more when you do VLIW, if it just ends up wasteing ~33% of it.

They still use their fixed tessellation engine it has nothing to do with the cores.
They are still not fully unified in their design.

 

[Will Robinson]

And the artwork keeps on coming...RAWR

 

[tweakboy]

Looks like I might have to put my 5870's to bed finally.

James you got a sick rig,, but your bottlenecking with that 6GB of ram only

put in 12GB and do justice to that 6 core 12 thread monster!!!!!!!

The 980x OCed to 5Ghz is faster then a Sandy Mandy IMO ,, just my opinion.... I rather have the 4 extra cores including HT I mean,,,,,

 

[Riek]

Why would there be a need for the ROP to be coupled with the memory channels? I don't see that in their current architecture overview... i do see a l2 cache coupling though.

 

[AtenRa]

^ this goes hand in hand with this below:

because:

Tomshardware and linking donanimhaber, as a source and saying:

• 4.50 billion transistors, die-area of ~380 mm², built on TSMC 28 nm process
• Advanced GCN 1D architecture
• 2048 1D processing cores
• 128 TMUs, 48 ROPs
• 384-bit wide GDDR5 memory interface, clock slightly below 1 GHz, target bandwidth of 240~264 GB/s
• 6pin + 8pin power connector required
• PCI Express Gen 3.0
• DirectX 11.1 support

However theyre probably wrong and its still 32 ROPs:

reason? this pic posted by Silver:

In this pic, atleast you see 32 ROPs and 128 TMUs.

GCN has 8 Render Back-ends
Each Render Back-End has x16 Z/Stencils and x4 Colors.

8 RBE x 4 Colors = 32 ROPs
8 RBE x16 Z/Stencils = 128 ROPs

 

[PrayForDeath]

Can anyone explain this?

 

[AtenRa]

Can anyone explain this?

Redefining TDP With PowerTune

 

[james1701]

James you got a sick rig,, but your bottlenecking with that 6GB of ram only

put in 12GB and do justice to that 6 core 12 thread monster!!!!!!!

The 980x OCed to 5Ghz is faster then a Sandy Mandy IMO ,, just my opinion.... I rather have the 4 extra cores including HT I mean,,,,,

Thanks. That has been on my planned update list, when I went to do it, they stopped making the memory I have in it now, so I figured I would wait a little as the prices finish bottoming. I will probably do that after the start of the year, then get the new cards.

 

[BrightCandle]

This is a further generation of PowerTune though where I think we are about to see them clock up as well as down to stay on or near the limit more of the time.

 

[bunnyfubbles]

What if the memory controllers are 96-bit instead of the usual 64-bit? Is that even possible? I admit I have a large ignorance when it comes to the more intimate technical details of various computer hardware.

well, the diagram shows 6 x 64bit channels...so regardless

can't say I'm too knowledgeable on the subject, which is why I asked how this all fits in the first place.

 

[vdgamer]

Just think logically guys, do you think amd would price 7970 higher than nvidia's 580 if it didn't outperform it? 580s are about 480-530 depending on the model, if 7970 is $550 for reference and $600 for msi lightning or some other high end model the performance should be 20-30% better, this obviously is just logical guess

 

[Rifter]

Who thinks the top end new ATI part will be twice as fast than SLI 460's?

Im trying to plan my next GPU upgrade, and want to go back to single card, but wont upgrade till i get twice the performance.

My CPU will be good enough to get me to Ivy-E or haswell I hope.

Will spend up to $500-600.

Or do you guys think i'll be waiting for the 680?

 

[toyota]

Who thinks the top end new ATI part will be twice as fast than SLI 460's?

Im trying to plan my next GPU upgrade, and want to go back to single card, but wont upgrade till i get twice the performance.

My CPU will be good enough to get me to Ivy-E or haswell I hope.

Will spend up to $500-600.

Or do you guys think i'll be waiting for the 680?

you have to be kidding if you are referring to the 7970.

 

[Grooveriding]

Who thinks the top end new ATI part will be twice as fast than SLI 460's?

Im trying to plan my next GPU upgrade, and want to go back to single card, but wont upgrade till i get twice the performance.

My CPU will be good enough to get me to Ivy-E or haswell I hope.

Will spend up to $500-600.

Or do you guys think i'll be waiting for the 680?

You're going to be waiting a lot longer than either of those cards if you want twice the speed of 460 SLI in a single-gpu card.

 

[Arkadrel]

Who thinks the top end new ATI part will be twice as fast than SLI 460's?

I dont >_> (not 100% (twice) as fast) more like ~40% or so.

460 1gb SLI is within ~5% or so of a 580.

The 7970 is probably gonna be 35% or so faster than the 580.

So you ll probably go from 95% of a 580 -> 135% (a gain of ~42%), if you upg from 460 SLI -> 7970.

(if the 7970 ends up around the 35% mark faster)

 

[PrayForDeath]

Who thinks the top end new ATI part will be twice as fast than SLI 460's?

Im trying to plan my next GPU upgrade, and want to go back to single card, but wont upgrade till i get twice the performance.

My CPU will be good enough to get me to Ivy-E or haswell I hope.

Will spend up to $500-600.

Or do you guys think i'll be waiting for the 680?

The question is, do you NEED to upgrade your SLI setup? It should be fine for any game out there.

 

[Will Robinson]

Its all about WANT...not need.

 

[maddie]

Who thinks the top end new ATI part will be twice as fast than SLI 460's?

Im trying to plan my next GPU upgrade, and want to go back to single card, but wont upgrade till i get twice the performance.

My CPU will be good enough to get me to Ivy-E or haswell I hope.

Will spend up to $500-600.

Or do you guys think i'll be waiting for the 680?

7990? Probably achievable.

 

[aphelion02]

I assume when he said single-card he meant single GPU. There is no reason to get a single dual-gpu card imo.

 

[imaheadcase]

The problem with performance claims anymore is they don't really mean anything, its really all about what a card performs with whatever games you like to play.

Sad but true, AMD cards own in some, are terrible in others.

For example, say BF3 the new card is could be %20 faster than 580, but its not even if it shows it on paper because nvidia favors that engine/driver.

Sucks, but its true.

 

[Ares1214]

The problem with performance claims anymore is they don't really mean anything, its really all about what a card performs with whatever games you like to play.

Sad but true, AMD cards own in some, are terrible in others.

For example, say BF3 the new card is could be %20 faster than 580, but its not even if it shows it on paper because nvidia favors that engine/driver.

Sucks, but its true.

Really isn't true...