AMD's Tonga - R9 285 (Specs) and R9 285X (Partial Specs)

[Dribble]

285/x is successor of 280/x obviously... What's the Comedy Central doing here?

You wanted amd to name it r9 370 so much that you believed yourself. ^^

The 265 is faster then the 260X, hence the 285 is not a good name as it's not faster then 280X.

Most depressing thing is it's not a competitor for maxwell, there is no competitor for maxwell - AMD are still competing with kepler with in this case what is basically a 7950. They don't really have a new architecture. This will mean Nvidia can screw us more on prices for maxwell chips due to lack of strong competition.

 

[Abwx]

Most depressing thing is it's not a competitor for maxwell, there is no competitor for maxwell - AMD are still competing with kepler with in this case what is basically a 7950.

Where is this maxwell based card that is above a 260X that AMD should compete against.?.

 

[champion-7891]

Unless Nvidia releases a Maxwell based card that performs in the ballpark of 7870/270X or higher with similar $ and significantly lower power, there is absolutely no reason for AMD to be worried.

Its not as if everyone here is desperate to get a 750Ti performance level card in their rig. Absolutely sure we would be hearing the same if it was AMD who were the first to come up with 750Ti like card, we would be hearing that nobody buys in this performance range and AMD cannot scale it up properly and so on.

 

[alexruiz]

Unless Nvidia releases a Maxwell based card that performs in the ballpark of 7870/270X or higher with similar $ and significantly lower power, there is absolutely no reason for AMD to be worried.

Its not as if everyone here is desperate to get a 750Ti performance level card in their rig. Absolutely sure we would be hearing the same if it was AMD who were the first to come up with 750Ti like card, we would be hearing that nobody buys in this performance range and AMD cannot scale it up properly and so on.

This is a great example on why post count matters little in regards to the quality of a post. You are a fairly new member, and you make much more sense than many of the "seasoned" members here. Thumps up!

People, the quoted text is absolutely right!
Unless nvidia comes with a maxwell derivative that is close in price, close in performance and much lower consumption, AMD will be fine.

 

[FatherMurphy]

Where is this maxwell based card that is above a 260X that AMD should compete against.?.

You mean the GTX 750ti?

http://www.anandtech.com/bench/product/1044?vs=1130

 

[AtenRa]

The fact that they are using a 30months old GPU to compete with AMD's latest GPU architecture should tell you enough.

They care so much about Tongo that they will sell the successor of the GTX770 for $499 while AMD is selling the successor of Tahiti for the same or less. :|

So let me ask you this,

If NVIDIA was expecting to have GM204 in late September, why lower MSRP of GTX-770 a few weeks before launch ??

ps: GTX-770 is not 30months old GPU, you mean Kepler architecture is 30month old. Same with GCN, it was released in January 2012, GCN 1.1 is only an update.

 

[Cloudfire777]

So let me ask you this,

If NVIDIA was expecting to have GM204 in late September, why lower MSRP of GTX-770 a few weeks before launch ??

ps: GTX-770 is not 30months old GPU, you mean Kepler architecture is 30month old. Same with GCN, it was released in January 2012, GCN 1.1 is only an update.

Because GTX 960 is coming.

GTX 780 have also recieved a major price cut. GTX 970 coming. It will be the direct competitor to R9 290X with muuuch lower TDP, and combined with a very good price, it will undoubtly sell well.

 

[FatherMurphy]

If NVIDIA was expecting to have GM204 in late September, why lower MSRP of GTX-770 a few weeks before launch ??

I think it means that GM204 is competing in an entirely different performance bracket than GK104, which should seem obvious, but given Tonga's performance-matching (to predecessor Tahiti) debut, perhaps we shouldn't take such advances for granted.

 

[AtenRa]

I think it means that GM204 is competing in an entirely different performance bracket than GK104, which should seem obvious, but given Tonga's performance-matching (to predecessor Tahiti) debut, perhaps we shouldn't take such advances for granted.

You mean NV will compete with GTX-770 against Tonga in that price segment ??

 

[caswow]

Because GTX 960 is coming.

GTX 780 have also recieved a major price cut. GTX 970 coming. It will be the direct competitor to R9 290X with muuuch lower TDP, and combined with a very good price, it will undoubtly sell well.

I read here in this thread tonga has been hyped much. absolutely every thread maxwell gets hyped so much without knowing anything about the cards. thats what you call hyping.

 

[Cloudfire777]

I read here in this thread tonga has been hyped much. absolutely every thread maxwell gets hyped so much without knowing anything about the cards. thats what you call hyping.

So are you expecting anything else considering we have already seen Maxwell and what it can do against current architectures?

We know what Maxwell is capable of.
http://tpucdn.com/reviews/NVIDIA/GeForce_GTX_750_Ti/images/perfwatt.gif

 

[96Firebird]

So let me ask you this,

If NVIDIA was expecting to have GM204 in late September, why lower MSRP of GTX-770 a few weeks before launch ??

To sell existing stock before the new cards hit. If they are planning a release for late September, they lowered the price of the cards too late IMO. We will see the same scenario there, where the old stock makes the new card look bad in perf/$.

ps: GTX-770 is not 30months old GPU, you mean Kepler architecture is 30month old. Same with GCN, it was released in January 2012, GCN 1.1 is only an update.

He said GPU, which is GK104. Which is 30 months old. GPU =/ graphics card.

Mr. Smith, where is your review?!

 

[ElFenix]

nm stop going down tangents

 

[MathMan]

Forget for a second about Maxwell, AMD managed to released a mid-range GPU to compete against another mid-range GPU, both with cores disabled, but one 20% larger than the other. And one silicon 2 1/2 years later than the other.

And even though the AMD GPU beats it by a bit, they still can't even manage to ask money for it.

That's a really scary situation to be in.

You guys care about low prices or perf/$, but AMD has higher cost and can't even monetize a slightly better performance. Over the long term, that never ends well. And the fact that they release this chip now doesn't bode well for the mid-term future: you don't make new silicon to replace it 6 months later with something else.

We don't know what Maxwell will bring in terms of perf/W but does anyone have any doubt that it will spank the Kepler architecture, and thus the GCN too?

That's even scarier.

AMD has always competed reasonably well from a technical point of view. But they were never able to get top dollar and market share for it. With Kepler, Nvidia was able to match or slightly surpass AMD from a technical point of view.
Now it looks like it's going to lose even that last glimmer of hope.

 

[SlowSpyder]

The fact that they are using a 30months old GPU to compete with AMD's latest GPU architecture should tell you enough.

They care so much about Tongo that they will sell the successor of the GTX770 for $499 while AMD is selling the successor of Tahiti for the same or less. :|

My guess is the 285 has more to do with being more inexpensive for AMD to produce than it being a true next gen architecture. Also, it brings FreeSync capability to a price and performance level where it previously was not. I bet most of AMD's resources are being put into future 20nm GPU parts.


Isn't the MSRP of the 280X is $299? They sell well below that. I bet 285 prices will settle fairly quickly after launch.

 

[parvadomus]

Forget for a second about Maxwell, AMD managed to released a mid-range GPU to compete against another mid-range GPU, both with cores disabled, but one 20% larger than the other. And one silicon 2 1/2 years later than the other.

And even though the AMD GPU beats it by a bit, they still can't even manage to ask money for it.

That's a really scary situation to be in.

You guys care about low prices or perf/$, but AMD has higher cost and can't even monetize a slightly better performance. Over the long term, that never ends well. And the fact that they release this chip now doesn't bode well for the mid-term future: you don't make new silicon to replace it 6 months later with something else.

We don't know what Maxwell will bring in terms of perf/W but does anyone have any doubt that it will spank the Kepler architecture, and thus the GCN too?

That's even scarier.

AMD has always competed reasonably well from a technical point of view. But they were never able to get top dollar and market share for it. With Kepler, Nvidia was able to match or slightly surpass AMD from a technical point of view.
Now it looks like it's going to lose even that last glimmer of hope.

1. You dont know how much AMD or Nvidia pays for each die, neither the ratio of functional dies.
2. You dont know whats the performance of the full Tonga die. I bet it destroy GK104 in every way and even Maxwell in performance/mm2.
3. Doubt any company would release any product that cost them the same as the previous one, and gets the same performance.

 

[FatherMurphy]

You mean NV will compete with GTX-770 against Tonga in that price segment ??

Yeah, that's what I anticipate. I think they won't come out with a GM206 on 28nm because they can compete with Tonga with the GK104. GK104 is smaller (about 294mm2 compared to Tonga's 350mm2 or so), as power efficient, and not so deprived of features (i.e. GK104 doesn't lack features that other Nvidia cards have, like Tahiti did compared to Hawaii/Tonga). As is rumored at TPU, if the GTX770 (or a re-branded version thereof) is lowered in price to $270, and GTX 760 (or a re-branded version thereof) is lowered to $220-230, then Nvidia already has its answer to Tonga.

Even if the full-blown Tonga die in the 285x outperforms the 770 somewhat (I expect it will), there are no indications why Nvidia couldn't sell the 770 for a little cheaper than the 285x. Also, as we have seen time and time again, Nvidia can compete even if it doesn't have the performance/$$ advantage.

 

[realibrad]

You mean the GTX 750ti?

http://www.anandtech.com/bench/product/1044?vs=1130

To be fair. A 260x costs $99 where as the 750ti costs $130ish. You are talking about a 5 maybe 10% difference in fps, and at such low numbers, that is a few frames here or there. There are quite a few games whee a 260x also beats the 750ti.

Now, if you compare an AMD card that costs as much as the 750ti such as the 265, you get this result.


http://www.anandtech.com/bench/product/1127?vs=1130

The 265 beats the 750ti pretty much in every game.

 

[ocre]

Yeah, that's what I anticipate. I think they won't come out with a GM206 on 28nm because they can compete with Tonga with the GK104. GK104 is smaller (about 294mm2 compared to Tonga's 350mm2 or so), as power efficient, and not so deprived of features (i.e. GK104 doesn't lack features that other Nvidia cards have, like Tahiti did compared to Hawaii/Tonga). As is rumored at TPU, if the GTX770 (or a re-branded version thereof) is lowered in price to $270, and GTX 760 (or a re-branded version thereof) is lowered to $220-230, then Nvidia already has its answer to Tonga.

Even if the full-blown Tonga die in the 285x outperforms the 770 somewhat (I expect it will), there are no indications why Nvidia couldn't sell the 770 for a little cheaper than the 285x. Also, as we have seen time and time again, Nvidia can compete even if it doesn't have the performance/$$ advantage.

there are 760s going for 219.99 on newegg right now. Some mir and i think one is just priced there.

 

[ocre]

Forget for a second about Maxwell, AMD managed to released a mid-range GPU to compete against another mid-range GPU, both with cores disabled, but one 20% larger than the other. And one silicon 2 1/2 years later than the other.

And even though the AMD GPU beats it by a bit, they still can't even manage to ask money for it.

That's a really scary situation to be in.

You guys care about low prices or perf/$, but AMD has higher cost and can't even monetize a slightly better performance. Over the long term, that never ends well. And the fact that they release this chip now doesn't bode well for the mid-term future: you don't make new silicon to replace it 6 months later with something else.

We don't know what Maxwell will bring in terms of perf/W but does anyone have any doubt that it will spank the Kepler architecture, and thus the GCN too?

That's even scarier.

AMD has always competed reasonably well from a technical point of view. But they were never able to get top dollar and market share for it. With Kepler, Nvidia was able to match or slightly surpass AMD from a technical point of view.
Now it looks like it's going to lose even that last glimmer of hope.

I think this was great food for thought. I think its great observation.

My biggest let down was the efficiency. Not the name, not the price. It was the efficiency. And its not a let down cause i wanted to buy it. It is kinda worrisome. Perhaps its nothing to think on but i find myself concerned with it when looking at the bigger picture. No one wants to see AMD fall far behind.

Perhaps we are looking too deep and the situation doesnt get bleak. It can go anyway from here.
Nvidia may simply position the gk104 in the 900 series. The gm204 may be efficient but not very powerful. It would be a boring time in GPU history and everything work out fine in the end.

 

[realibrad]

I think this was great food for thought. I think its great observation.

My biggest let down was the efficiency. Not the name, not the price. It was the efficiency. And its not a let down cause i wanted to buy it. It is kinda worrisome. Perhaps its nothing to think on but i find myself concerned with it when looking at the bigger picture. No one wants to see AMD fall far behind.

Perhaps we are looking too deep and the situation doesnt get bleak. It can go anyway from here.
Nvidia may simply position the gk104 in the 900 series. The gm204 may be efficient but not very powerful. It would be a boring time in GPU history and everything work out fine in the end.

I dont think things are nearly as bad as he makes them sound. When you compare the power usage of a AMD 265 vs 750 ti it seems pretty big. When you compare total system power things get a lot better. Then, you realize that the 265 gives better perf by about 10-15%, and you then look at the fact that the 265 uses 10-15% more power, things are not as bad.

For sure the 750 ti uses very little power. But once you start to OC that chip, the power gets closer to a 265. To get a 750 ti to perform equal to a 265, total system power is very very close.

 

[tviceman]

It's not too far from the truth. The 285 came at a bad time for us - Anand's retirement, labor day weekend, and a business trip I'm in the middle of. We've never missed a review like this before (and we won't again), but it was exceptional circumstances.

You'll see it tomorrow.🙂

So tomorrow came and went, but the review never showed up. Are we still getting a review and your perspective on the card???

 

[ShintaiDK]

So tomorrow came and went, but the review never showed up. Are we still getting a review and your perspective on the card???

After the iPhone 6 review? 😉

 

[MathMan]

1. You dont know how much AMD or Nvidia pays for each die, neither the ratio of functional dies.

True that. But I am confident that Nvidia is a much bigger customer of TSMC than AMD, and that Nvidia must be getting a pretty good deal there.
In addition, since both have cores enabled, the yield of both should be very high on such a mature process. And, finally, the yield on the fully functional dies has to tilt in favor of Nvidia because their die is 20% smaller.

I think that's sufficient arguments to conclude that Nvidia's die cost will be significantly lower.

2. You dont know whats the performance of the full Tonga die. I bet it destroy GK104 in every way and even Maxwell in performance/mm2.

I bet it will destroy GK104 in absolute performance. At the cost of extra DRAMs. I bet it will be destroyed by GM204 despite the latter having 2 DRAMs less. And if gm107 is an indication, I bet gm204 will destroy Tonga as well in perf/mm2.
And, finally, gm204 will have better perf/W as well, so the perf/$ for cooling will be better for gm204 as well.

Bloodbath coming.

3. Doubt any company would release any product that cost them the same as the previous one, and gets the same performance.

Tonga has better perf/$ than Tahiti. So much is obvious. It's just that they only now catch up with GK104 in that respect. They're 2 years late.

You really wonder what AMD has been doing all this time.

 

[monstercameron]

True that. But I am confident that Nvidia is a much bigger customer of TSMC than AMD, and that Nvidia must be getting a pretty good deal there.
In addition, since both have cores enabled, the yield of both should be very high on such a mature process. And, finally, the yield on the fully functional dies has to tilt in favor of Nvidia because their die is 20% smaller.

I think that's sufficient arguments to conclude that Nvidia's die cost will be significantly lower.


I bet it will destroy GK104 in absolute performance. At the cost of extra DRAMs. I bet it will be destroyed by GM204 despite the latter having 2 DRAMs less. And if gm107 is an indication, I bet gm204 will destroy Tonga as well in perf/mm2.
And, finally, gm204 will have better perf/W as well, so the perf/$ for cooling will be better for gm204 as well.

Bloodbath coming.


Tonga has better perf/$ than Tahiti. So much is obvious. It's just that they only now catch up with GK104 in that respect. They're 2 years late.

You really wonder what AMD has been doing all this time.

I disagree with everything here especially the bolded bit, how can you be confident? You do know that AMD also builds small core apus via tsmc?