• 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.

***AMD Radeon HD7970 GHZ Edition - Official Reviews Thread***

Well the techpowerup review is using the 12.7 beta drivers for the review, which are known to give performance increases. It would be fairer to compare to 670/680 scores with the newest drivers available also, but its just comparing results to the results they got in the original 670/680 reviews.

I am a fan of techpowerup reviews myself, but this one isnt particularly fair in my eyes.

One thing that is clear is that if the 7970 GHz matches the 680 at 1080p, it will generally be faster at 1600p and multi monitor resolutions, and especially if using AA. The performance hit of using AA is smaller on the 7970 architecture because of the bandwidth advantage it has.
 
So upto 1080p the 680 retains the fastest single gpu crown (in some reviews, in others ir states the 7970 ghz is 2% faster), but above that the 7970 ghz edition smashes it's back doors in.
 
Last edited:
Lol, according to Anandtech, the 680 beats the GHz edition at GPUcompute:rolleyes:

A nice checque from Nvidia must be coming their way. Their review certainly shows the GHz edition in the least favourable way, think I will discount its validity as well as the TPU review.

I know when scoring the olympics they discount the highest and lowest score to try and remove bias, so i will do the same thing :)
 
Looks unlikely to me. As things stand, how much quicker is Ghz Edition in 3 monitor configuration over 670 according to Techpowerup (in alphabetical order):

Alan Wake 40%
AvP 37%
Arkham City 18%
BF3 21%
Battleforge 43%
CoD4 minus 8.5% (that old quake 3 engine game)
Civilisation V Nvidia failed to run
Crisis 21%
Crisis2 15%
Dirt 3 Nvidia failed to run
Dragon Age2 10%
Hard Reset Nvidia failed to run
Metro 2033 21%
Stalker Pripyat 19%
Shogun2 2.5%
Skyrim 20%


As for TPU being inconsistent. Please show me one of their other reviews where you think they got it wrong?

Why would I the unlikely the 4GB would be quicker, the main reason the 2GB one is slower is because it has 1GB less.

Will do when i get home, at work so browsing through sites is kind of frowned upon, but look through previous TPU reviews and you will see the variations.

The thing that stands out the most is the fact they couldn’t get Dirt3 working at 5760x1080 on the 670/680 and several other websites could.
This clearly shows that they don’t know what they were doing, if they made a mistake with that, how many other mistakes did they make?
 
Why would I the unlikely the 4GB would be quicker, the main reason the 2GB one is slower is because it has 1GB less.

Will do when i get home, at work so browsing through sites is kind of frowned upon, but look through previous TPU reviews and you will see the variations.

The thing that stands out the most is the fact they couldn’t get Dirt3 working at 5760x1080 on the 670/680 and several other websites could.
This clearly shows that they don’t know what they were doing, if they made a mistake with that, how many other mistakes did they make?

No, the main reason the 7970 is more powerful at these resolutions is because it has a 384-bit memory bus and its bandwidth is far superior to the 680. An extra 2 GB may help it slightly but I doubt it looking at 4GB 680 reviews, but I am happy to be proven wrong. I just havent seen any meaningful gains from the extra 2GB in 4GB 680 reviews.
 
Yeah, reviews are all over the place, there is some sites even showing the 680 beating the 7970 at idle power etc. theres some massive discrepancies between them.

PS review sites have more bias than fanboys these days lol. I would trust the user reviews over the review sites anyday.
 
Did you you even bothered reading the 1st post of this topic?

I did, but has been since edited a couple of times :)

From the links provided on the first page or any other one can muster up, its seems anybody could prove their case either way they want, sad really. this is what it has come to and its going to get worse as more sites pop up favoring their favorite hardware company. the review sites have only themselves to blame.

We will have intel vs amd vs nvidia sites... cant wait (sarcasm)
 
No, the main reason the 7970 is more powerful at these resolutions is because it has a 384-bit memory bus and its bandwidth is far superior to the 680. An extra 2 GB may help it slightly but I doubt it looking at 4GB 680 reviews, but I am happy to be proven wrong. I just havent seen any meaningful gains from the extra 2GB in 4GB 680 reviews.

Maybe, but it’s not easy to find reviews comparing the 4GB GTX680 against the 7970 at 5760x1080, never mind the 4GB 670.
 
Interesting tidbit from a chap who works for AMD:

http://forum.beyond3d.com/showpost.php?p=1652097&postcount=3701

AMD CHAP said:
The GHz edition ASIC is explicitly designed to fit within the same thermal / electrical / physical footprint as the standard 7970 and as such, given that every one of the partners have transistioned to their own designs they will be using their own fansink designs. The "reference" design is not much more than our qualification mule now and is sampled for the purposes of performance testing.
 

They're hardly going to say it's nothing special when they expect people to buy them over less expensive 7970 are they?

Even if the 7970GE are cherry picked and slightly less leaky it just means that a 7970GE will run slightly cooler than a SE at any given speed/voltage.

If the 7970GE ASIC were something special you'd expect their voltage to be equal or lower than a standard 7970 which it isn't in its boost state, they just look like a standard 7970 with a software based overclocking mechanism bolted on.

Has anybody come across a standard 7970 that won't do 1050mhz on 1218mV?
 
Why would I the unlikely the 4GB would be quicker, the main reason the 2GB one is slower is because it has 1GB less.

Will do when i get home, at work so browsing through sites is kind of frowned upon, but look through previous TPU reviews and you will see the variations.

The thing that stands out the most is the fact they couldn’t get Dirt3 working at 5760x1080 on the 670/680 and several other websites could.
This clearly shows that they don’t know what they were doing, if they made a mistake with that, how many other mistakes did they make?

It's worth noting, but I can't remember the game, that [H] made a bold claim about surround being better on Nvidia and AMD drivers sucking(after saying the opposite on forums) and it turns out the game only works when hacked, as default the drivers wouldn't let the game play in surround.

Some sites use common sense and get a game working, other times they just go with what the driver supports full stop with no hacking.

For me especially with [H], they managed to force this game to work in surround, but have repeatedly failed to do the same with other games when AMD is concerned. THey did a little Star Trek online DX11 performance thing where Nvidia worked great and AMD sucked, when all other websites had AMD ahead in performance both in dx 9 and 11, and [h] had AMD training quite a lot in dx11 with again bold claims of AMD drivers sucking and lack of support for the game.

In other words, reviews are dodgy, any reviewer is capable of making mistakes, making bold claims about poor drivers for a reviewer mistake, is simply poor reviewing, they can go look at everyone else's reviews and point out their own mistakes or continue claiming theres some fundamental problem that no one else has.
 

AMD CHAP said:
The GHz edition ASIC is explicitly designed to fit within the same thermal / electrical / physical footprint as the standard 7970 and as such, given that every one of the partners have transistioned to their own designs they will be using their own fansink designs. The "reference" design is not much more than our qualification mule now and is sampled for the purposes of performance testing.

They're hardly going to say it's nothing special when they expect people to buy them over less expensive 7970 are they?

Even if the 7970GE are cherry picked and slightly less leaky it just means that a 7970GE will run slightly cooler than a SE at any given speed/voltage.

If the 7970GE ASIC were something special you'd expect their voltage to be equal or lower than a standard 7970 which it isn't in its boost state, they just look like a standard 7970 with a software based overclocking mechanism bolted on.

Has anybody come across a standard 7970 that won't do 1050mhz on 1218mV?

??

The post is about the fact the HD7970 V2 will have non-reference coolers at launch unlike the HD7970 V1. Hence cooling and noise levels of retail examples will be better than those in the reviews. Considering that the reference HD7970 V2 cooler is very noisy,this is good news as the card should run cooler and produce less noise.
 
Last edited:
It's worth noting, but I can't remember the game, that [H] made a bold claim about surround being better on Nvidia and AMD drivers sucking(after saying the opposite on forums) and it turns out the game only works when hacked, as default the drivers wouldn't let the game play in surround.

Some sites use common sense and get a game working, other times they just go with what the driver supports full stop with no hacking.

For me especially with [H], they managed to force this game to work in surround, but have repeatedly failed to do the same with other games when AMD is concerned. THey did a little Star Trek online DX11 performance thing where Nvidia worked great and AMD sucked, when all other websites had AMD ahead in performance both in dx 9 and 11, and [h] had AMD training quite a lot in dx11 with again bold claims of AMD drivers sucking and lack of support for the game.

Just been looking over a few reviews then read your post.

In [H]'s GHz oc'ing page, it is nothing short of blatant brown nosing, seriously, they have the audacity to go into detail and even post pictures of what happens when you exceed a cores limits!:eek:

'Often times when overclocking we find a video cards limitations not by driver crashing, but by visual anomalies that appear in game and ruin the image quality or fail to load textures. When we tried to overclock the AMD Radeon HD 7970 GHz Edition past 1180MHz these anomalies showed up everywhere in both Battlefield 3 and The Witcher 2: Enhanced Edition. In Max Payne 3 we would just get driver and system crashes.'

Well **** me, I can't thank you enough [H], I never new that happened when you oc too far.

I tried to get a hold of Brent on speed dial to query why they added this to the review, but the missus said he couldn't answer the phone because he was in the red JCB for the last 5 hours scraping the **** off of Grady's head.

She said I would need to ring back in another 5 hours as he still had half his head to go!:eek:

Nothing short of total embarrassment tbph.
 
Last edited:
??

The post is about the fact the HD7970 V2 will have non-reference coolers at launch unlike the HD7970 V1. Hence cooling and noise levels of retail examples will be better than those in the reviews. Considering that the reference HD7970 V2 cooler is very noisy,this is good news as the card should run cooler and produce less noise.

He states that the ASIC (GPU) is "explicitly designed" which would suggest it was higher binned/superior to a standard 7970 GPU.

My point is the GPU itself looks to be no better than a standard 7970.

Many standard 7970 have a core voltage of around 1120mV and will do 1050mhz easily, the very worst 7970SE have a core voltage of 1175mV and I'd be surprised if they didn't also do 1050mhz using the same voltages 7970GE has.

If the 7970GE ASIC was so superior you would expect it to run 1050mhz at 1100mV or even 1000mV, when quite the opposite is in fact the case with cards reported at 1225mV and even 1256mV under boost.

To me and a few equally suspicious people on Beyond3d 7970GE looks to be nothing more than a bog standard 7970 with a new bios and drivers offering a fancy boost mechanism.

http://forum.beyond3d.com/showpost.php?p=1652280&postcount=3730

I think it is well known how the boost works from Kepler
The problem I see is the lower power-efficiency:

All measurements below only for the card, not the whole system
HT4U: 21% more power (HawX)
PCGH: 33% more power (BFBC2)
techpowerup: 28% more power (Crysis 2)
Hardware.fr: 20% more power (Anno 2070)

All for a maximum of 13% better performance. It is especially worrying that ht4u measures 254W for the card itself during gaming while the TDP is only 250W.
http://forum.beyond3d.com/showpost.php?p=1652285&postcount=3731

Well, the numbers aren't identical.

For HD 7970, AMD stated: "typical board power ~210 W" and "maximum board power 250 W" for 7970.

The HD 7970 GE guide says: "typical power draw ~250 W" / "typical board power ~250 W"

So, PowerTune limit ("maximum power") wasn't probably changed, but typical power draw was raised by 40 watts.

So AMD have just overclocked a standard 7970 and made non-reference heatsinks mandatory then? and AMD are expecting people to pay more money for them? if anything the boost just gets in the way of overclocking so I'd rather have a standard 7970.
 
Last edited:
the very worst 7970SE have a core voltage of 1175mV and I'd be surprised if they didn't also do 1050mhz using the same voltages 7970GE has.

I have a bog standard ref 7970SE that is 1175mv stock, 68.4% asic, it runs upto 1250mhz at 1.25v and 24/7 oc is 1150/1650 at stock volts.
A few reviewers have tried the GE BIOS and it doesn't work on the SE.
 
Back
Top Bottom