First AMD 7970 review up: Now with AT official review

[Flipped Gazelle]

I don't pay much attention to these sorts of things, and which websites are affiliated with each other, but this morning I noticed that [H] doesn't list Anandtech (among others) on their front-page Review Roundup. Same thing during the Bulldozer release. Is there animosity, or am I merely unobservant/uninformed?

 

[Annon2255]

What a ocing beast. For real it is. But way to expensive for me.

I agree. After Kepler comes out will the price on this card drop? Or will it stay about the same for a while?

 

[lol123]

Stock performance is disappointing but overclocked performance looks excellent. Stock clock frequencies may be too conservative but with a maturing 28nm process AMD will probably release higher-clocked SKUs. Also it's a new architecture so we can actually expect significant improvements from driver updates (I especially look forward to those in BF3). I can't help but think that AMD dropped the ball (Bulldozer style) with regard to product PR however. This is not the splash launch they needed.

 

[-Slacker-]

Disappointing. Very marginal performance increment. Still cannot break 30fps in metro2033. And only 5% better than 580 in BF3?!





In 2009 the 5850 outperforms GTX285 similarly and AMD charged $259 for it. Now the price is insanity!

I think anandtech's review comment sums it up nicey:

At the risk of incurring the community's wrath, there are about 20 other review links posted in this thread, a lot of them showing different results than the AT benches ... might be a good idea to consult a few more of them...

 

[3DVagabond]

At the risk of incurring the community's wrath, there are about 20 other review links posted in this thread, a lot of them showing different results than the AT benches ... might be a good idea to consult a few more of them...

Yes, it's a bit frustrating to see people pull one review (sometimes one bench from one review) and declare victory or defeat from it. Also the rhetoric and double standards we see being posted. We all have biases. It's only human, and difficult to over come. It takes effort though to be a BS artist. Some folks need to quit trying so hard.

AMD has a new CEO. He's running things differently. He was likely hired to do just that.



Now, about the card... Does anyone else think this chip pretty much fits apoppin's preview? Except for the crippled aspect I do. I think this chip is being run at no where near it's final potential. With voltage this chip's gonna break 1.2GHz, in all likelihood. Possibly well over since we've already seen reviews @ 1165MHz on stock voltage. On another forum there was a poster claiming to have a 6970. He/She was pretty tight lipped. The only bit of info he said was the clocks were compromised. The chip needs more power than AMD liked to run at it's max clocks. He too likened it to Fermi, in that respect.

So, is it unreasonable to think there's another 25%-30% more performance available from this as the process matures? Or, maybe with a respin?

 

[MrX8503]

I think anandtech's review comment sums it up nicey:

Do people seriously only read what they want to hear?

Anandtech also said this, which you conveniently left out

BF3 implements a wide array of technologies so it’s hard to say for sure, but there is one thing we know they implement in the engine that only NVIDIA can use: Driver Command Lists, the same “secret sauce” that boosted NVIDIA’s Civilization V performance by so much last year.

 

[Arzachel]

Now, about the card... Does anyone else think this chip pretty much fits apoppin's preview? Except for the crippled aspect I do. I think this chip is being run at no where near it's final potential. With voltage this chip's gonna break 1.2GHz, in all likelihood. Possibly well over since we've already seen reviews @ 1165MHz on stock voltage. On another forum there was a poster claiming to have a 6970. He/She was pretty tight lipped. The only bit of info he said was the clocks were compromised. The chip needs more power than AMD liked to run at it's max clocks. He too likened it to Fermi, in that respect.

So, is it unreasonable to think there's another 25%-30% more performance available from this as the process matures? Or, maybe with a respin?

He was right about the 25%-30% figure... and thats basically it. The card runs cool, has an amazing perf/w ratio, doesn't increase neither die size nor voltage over last gen's top card, overclocks like hell and is early on top of that. It's basically the anti-Fermi

 

[Flipped Gazelle]

I can't help but think that AMD dropped the ball (Bulldozer style) with regard to product PR however. This is not the splash launch they needed.

I disagree, and in fact I don't understand how you could feel this way. BD was a near-disaster, over-hyped, late, under-performing and almost universally panned. 7970 has received much acclaim from the review sites and is demonstrably better than the competition and is sanely priced, although of course I wish it were cheaper.

Also, this architecture, as it currently stands, clearly has something in reserve, unlike BD.

 

[Flipped Gazelle]

Now, about the card... Does anyone else think this chip pretty much fits apoppin's preview?

Yes, I do. I think people came down too hard on him. Sure, he editorialized too much in his preview, but he also threw a few bouquet's AMD's way - he described the 7970's power-conserving ability as "amazing". ABT does a fine job with benches, but the articles are generally poorly written, with too much opinion interjected, and too self-promoting. Hopefully as ABT matures they'll grow out of it.

 

[badb0y]

I am planning on breaking 1250 core at least.

 

[woolfe9999]

I honestly don't understand the negative reactions to this card. It has to be either a product of unrealistic expectations or just not liking AMD. Probably mostly the former. This card has similar price/performance to the 580. Ordinarily the fastest card has worse price/performance than the second fastest. It's also a strong OCer, excellent power consumption numbers, and good tessellation. It also adds a powerful compute engine which, though irrelevant for gamers, has to be factored into the price.

This is a good/strong launch for AMD. Not a great one, but quite good.

Comparisons to BD? Give me a break. That was a total fiasco. An embarrassment of epic proportions.

Oh, and can people please stop using cherry-picked benchmarks to prove either that the card is good or bad? It's an insult to peoples' intelligence.

- wolf

 

[Corporate Thug]

the potential in this card lies in its OC ability. Honestly, people buying a card on this level are enthusiasts so I think most will try to push the limits.

 

[Attic]

Do people seriously only read what they want to hear?

Yes 100%, you just have to accept it and know when you're dealing with this kind of person. It's in part confirmation bias. Folks look back from their conclusion and focus on anything (even incorrect or illogical thoughts) that proves their conclusion and dismiss anything (even clear logical thoughts) that points to an alternative conclusion.

If someone needs 3+3 to equal 6 because 6 serves their purpose it is sometimes impossible to get them to admit that they've misinterpreted 2+2 as 3+3, but you're better off doing this than arguing that the answer is 4. Though worse are people who know that they are dealing with 2+2 and insist on presenting it as 3+3 because they need a 6.

We see this in politics and business all the time.

 

[Lonyo]

Considering AMD's mantra has been the sweet spot strategy since the HD 3870, why would you offer that so boldly? What it is-is a flip flop -- a change of strategy. Personally don't have any problem with enthusiast priced single GPU's though and have been consistent about it and never believed in the sweet spot strategy because it left too much revenue potential on the table.

It would be pretty much suicidal and utterly pointless to price it at $400 or a "sweet spot", because it would kill their own other GPUs as well.

The question is, have they actually changed strategy.
It might still be a sweet-spot GPU, something they CAN price at $400, but that remains to be seen.
Even if it is a sweet spot GPU (die size suggests it could be, since it's slightly smaller than the HD6970), it won't be priced as such until there's a reason to price it there (competition).

The launch price is exactly like the GTX280, inflated above what you might think because there's no reason to make it cheaper.
When the 4800 series came out, it adjust the GTX200 series pricing, and when NV launches their new parts, it will adjust the HD7000 series pricing, until then, there's no incentive to AMD to play the value card.

Personally I was hoping for something cheaper so that I would have an idea of where prices might be in April when I want to buy, but I will just have to hope NV launches something by then (or releases benchmarks and pricing).

 

[TakeNoPrisoners]

It looks like AMD finally got the performance of the 5970 in one GPU.

It took 2 generations to double the performance of the 5870.

 

[SirPauly]

It would be pretty much suicidal and utterly pointless to price it at $400 or a "sweet spot", because it would kill their own other GPUs as well.

The question is, have they actually changed strategy.
It might still be a sweet-spot GPU, something they CAN price at $400, but that remains to be seen.
Even if it is a sweet spot GPU (die size suggests it could be, since it's slightly smaller than the HD6970), it won't be priced as such until there's a reason to price it there (competition).

The launch price is exactly like the GTX280, inflated above what you might think because there's no reason to make it cheaper.
When the 4800 series came out, it adjust the GTX200 series pricing, and when NV launches their new parts, it will adjust the HD7000 series pricing, until then, there's no incentive to AMD to play the value card.

Personally I was hoping for something cheaper so that I would have an idea of where prices might be in April when I want to buy, but I will just have to hope NV launches something by then (or releases benchmarks and pricing).

I would say the strategy has changed -- didn't hear the words sweet spot this round so far, hehe!

 

[3DVagabond]

He was right about the 25%-30% figure... and thats basically it. The card runs cool, has an amazing perf/w ratio, doesn't increase neither die size nor voltage over last gen's top card, overclocks like hell and is early on top of that. It's basically the anti-Fermi

I'm talking about the chip itself. Not his spin and editorializing.

 

[3DVagabond]

It looks like AMD finally got the performance of the 5970 in one GPU.

It took 2 generations to double the performance of the 5870.

That's because the last gen was borked being stuck on a node it wasn't designed for.

 

[Sable]

Wait, wat?

People bitching that ATI are charging less than the price of a 3GB GTX580 for a card that is faster than 3GB GTX580?!

 

[Edgy]

Seems pretty good if one considers this as AMD's attempt at their own "Fermi"...

Lest we all forget, original Fermi (GTX 480) at launch was $500 and only about 10-15% faster than 5870 which was around $400... and we all know about heat/noise issues let alone the fact that the card was 6 months late...

Of course, Nvidia followed up extremely well to improve upon the original which we can all hope AMD can do as well with theirs.

Looks like 28nm node really helped AMD avoid similar power/heat/noise issues with their new architecture and performance seems pretty good (stock) to excellent (if counting great OC potential).

 

[Deleted member 4644]

I honestly don't understand the negative reactions to this card. It has to be either a product of unrealistic expectations or just not liking AMD. Probably mostly the former. This card has similar price/performance to the 580. Ordinarily the fastest card has worse price/performance than the second fastest. It's also a strong OCer, excellent power consumption numbers, and good tessellation. It also adds a powerful compute engine which, though irrelevant for gamers, has to be factored into the price.

This is a good/strong launch for AMD. Not a great one, but quite good.

Comparisons to BD? Give me a break. That was a total fiasco. An embarrassment of epic proportions.

Oh, and can people please stop using cherry-picked benchmarks to prove either that the card is good or bad? It's an insult to peoples' intelligence.

- wolf

The problem is that the GTX580 has been out for a YEAR. This card is going to get trounced (in all probability) in about 6 months.

 

[AnandThenMan]

The problem is that the GTX580 has been out for a YEAR. This card is going to get trounced (in all probability) in about 6 months.

In 6 months, do you think AMD will have absolutely nothing faster in a single GPU card? Plus it gives AMD 6 months to benefit from the fastest card out there. If I was AMD, I would ecstatic about this.

But I don't think it will take Nvidia 6 months to respond, most likely we will see something much sooner, even if actual availability is several weeks or more out.

 

[sandorski]

The problem is that the GTX580 has been out for a YEAR. This card is going to get trounced (in all probability) in about 6 months.

Few cards don't get "trounced" in 6 months.

 

[Flipped Gazelle]

The problem is that the GTX580 has been out for a YEAR. This card is going to get trounced (in all probability) in about 6 months.

In 6 months AMD will have a different top of the line card, if Nvidia delivers with Kepler.

 

[MrX8503]

The problem is that the GTX580 has been out for a YEAR. This card is going to get trounced (in all probability) in about 6 months.

And that equates to the 7000 series being close to failure of Bulldozer how?