Of SKUs, OCs and what constitutes a midrange part

Mintmaster

Veteran
Both Nvidia and AMD struggle to make money in this division. AMD has barely been profitable compared to how much it spent into aquiring ATI. In the long run, the deal has costed them more than it has made them.
Go look up what a sunk cost is. It doesn't matter what ATI was acquired for in the past. Going forward, their GPU unit is far more competitive and closer to profit.

On top of that, it's critical for their Fusion line. If they don't top Intel in GPU performance, their entire mobile division is useless. Poorer power consumption (due primarily to process deficiency), poorer IPC. Haswell is going to have a big graphics boost, as will the Atom lineup. Without a GPU edge, AMD will have to hope that Intel prices its product high enough for them to find room to break even.
As a result, Nvidia's midrange catching up to AMD top end should not be a surprise this generation. Considering how much more Nvidia spends, it was likely to happen.
What are you talking about? 680 vs 7970? First of all, the 7970 is still faster. Secondly, the 680 isn't NVidia's midrange. Finally, all of AMD's other parts are a bit better than NVidia's. Pitcairn is a bit faster than GK106 despite being smaller, Cape Verde beat GK107, and now Bonaire blows away any other 128-bit GPU.

NVidia has a leg up in the GPGPU market, but it's not a done deal. AMD is doing better in most OpenCL tests, which surprisingly seems like it will become the API of choice due to portable computing seeking power-efficient parallel computation. Adobe has finally implemented OpenCL support and they have more reason to drop the CUDA path. Folding@Home is arguably the biggest GPGPU app in the world, and has gone OpenCL.
 
Last edited by a moderator:
First, the 680 and 7970 (non GHz) are equally fast:
http://www.computerbase.de/artikel/grafikkarten/2013/test-nvidia-geforce-gtx-titan/7/

Second, the 680 is a beefed up midrange card. It is marketed as highend, but marketing can totally distort things. For Nvidia standards (die size, memory bus width, power consumption, performance), the 680 is definitely not highend.

Third, the fastest GK106 solution, the GTX660 is pretty close to the fastest Pitcairn solution, the 7870. And Pitcairn is not really smaller (212 vs 214 mm2)

Fourth, while the 7770 is a good deal faster than the 650, it uses proportionately more power, too. You cannot really say one beats the other, they just occupy slightly different positions.
 
What are you talking about? 680 vs 7970? First of all, the 7970 is still faster. Secondly, the 680 isn't NVidia's midrange. Finally, all of AMD's other parts are a bit better than NVidia's. Pitcairn is a bit faster than GK106 despite being smaller, Cape Verde beat GK107, and now Bonaire blows away any other 128-bit GPU.

Ghz edition is faster the regular one isn't and that's with 384bit. It can be argued that GK110/Titan is the high end and that 294mm2 is midrange for nVidia, but let's not because that is getting super old and it boils down to marketing semantics. Pitcairn has a 256-bit memory bus vs the 192-bit that the GK106 has. GK106 should be pretty competitive with pitcairn if it had 256-bit bus imo. nVidia has a 660ti, 670, 680 and Titan on top of the 660,so they also have pretty good presence at many price points. 660 is cheaper than 7870 and 650ti boost is faster than 7790 and costs the same when they have same amount of memory.

edit: didn't see the post above mine, took some time to check prices...
 
Last edited by a moderator:
First, the 680 and 7970 (non GHz) are equally fast:
http://www.computerbase.de/artikel/grafikkarten/2013/test-nvidia-geforce-gtx-titan/7/

Second, the 680 is a beefed up midrange card. It is marketed as highend, but marketing can totally distort things. For Nvidia standards (die size, memory bus width, power consumption, performance), the 680 is definitely not highend.

What you're saying is Nvidia's "beefed up midrange card" is about 15% slower than AMD's fastest card. That's not quite the same as Nvidia's midrange catching up to AMD's top end.

Third, the fastest GK106 solution, the GTX660 is pretty close to the fastest Pitcairn solution, the 7870. And Pitcairn is not really smaller (212 vs 214 mm2)
If you consider that both of these are actually "midrange" and the AMD card is faster (while drawing more power), it's a bit more accurate to say that both companies midrange cards are actually quite close together.

Had AMD to do a "beefed up Pitcairn" with say 28 CU's, it would likely be pretty close to the 680 in every way.

Fourth, while the 7770 is a good deal faster than the 650, it uses proportionately more power, too. You cannot really say one beats the other, they just occupy slightly different positions.
You can say the same thing about the 7870 and 680 - you basically did by calling the 680 "beefed up midrange". Assume the 7770 is "beefed up entry level" and you have the exact same argument going back.

Fact is there is really nothing much between them. It feels like a victory to Nvidia because they were getting beaten up on metrics such as perf/W, perf/mm2 etc for years and now they are on a similar level (and that's without taking AMD's faster TTM into consideration, which is a technical disadvantage).
 
What you're saying is Nvidia's "beefed up midrange card" is about 15% slower than AMD's fastest card. That's not quite the same as Nvidia's midrange catching up to AMD's top end.

No, I'm not saying that, stop twisting the words in my mouth. Why do you compare a 7970 OC (which the 7970 GHz is) with a 680 non-OC? The 680 plays in the same league as the 7970. Simple fact.

If you consider that both of these are actually "midrange" and the AMD card is faster (while drawing more power), it's a bit more accurate to say that both companies midrange cards are actually quite close together.

Do you want to play the paraphrasing game? What is this? You're not really saying anything different than I was saying.

Had AMD to do a "beefed up Pitcairn" with say 28 CU's, it would likely be pretty close to the 680 in every way.

Had Nvidia to do a "beefed up GK106" with say 7 SMX, it would likely be pretty close to your beefed up Pitcairn. So what? Btw by "beefed up" I meant clocks only. As far as I know, the original clock target for GK104s fastest SKU was about 900-950 MHz. That is not very far from the 1006/1058 MHz of the GTX680.

You can say the same thing about the 7870 and 680 - you basically did by calling the 680 "beefed up midrange". Assume the 7770 is "beefed up entry level" and you have the exact same argument going back.

Fact is there is really nothing much between them. It feels like a victory to Nvidia because they were getting beaten up on metrics such as perf/W, perf/mm2 etc for years and now they are on a similar level (and that's without taking AMD's faster TTM into consideration, which is a technical disadvantage).

See my earlier explanation of "beefed up". I was contesting the claim that AMD is beating Nvidia on every level. Which by your own admission, they are not - they are pretty equal.

So I don't get why you have to barge into this discussion when you're just confirming what I'm saying.
 
No, I'm not saying that, stop twisting the words in my mouth. Why do you compare a 7970 OC (which the 7970 GHz is) with a 680 non-OC? The 680 plays in the same league as the 7970. Simple fact.

So we're just going to discount the 7970 "OC" as you put it because...?

What happens when you OC a 7950? Does it play in the same league as the 680? I'm sure it does.

Do you want to play the paraphrasing game? What is this? You're not really saying anything different than I was saying.

Had Nvidia to do a "beefed up GK106" with say 7 SMX, it would likely be pretty close to your beefed up Pitcairn. So what? Btw by "beefed up" I meant clocks only. As far as I know, the original clock target for GK104s fastest SKU was about 900-950 MHz. That is not very far from the 1006/1058 MHz of the GTX680.

See my earlier explanation of "beefed up". I was contesting the claim that AMD is beating Nvidia on every level. Which by your own admission, they are not - they are pretty equal.

So I don't get why you have to barge into this discussion when you're just confirming what I'm saying.
Yep I agree with most of it, but the problem is you continue to believe that Nvidia is somehow a mile ahead because their - in your own words - "beefed up midrange card" is - in my words - "15% slower than AMD's fastest card".

It's ridiculous to claim that the 7970 GHz doesn't count because it's "overclocked" - why not assume the 7970 was underclocked when most analysts realise that AMD was playing it extremly safe? You can still continue to overclock GHz editions, by far...and they will pull away from OC'd 680's. If Nvidia had enough good 680's to make an official overclocked SKU out of them they'd have done so months ago in order to compete with the GHz edition btw.
 
I'm not going to get into the pissing match of 680 versus 7970, however I am going to state this:

NVIDIA marketed the 680 as their high end card, they numbered it as their high end card, and they put a price tag on it commensurate with their high end card. For more than a year, it was their defacto high end card.

There's no excuses here for "it was only midrange"; NVIDIA made nothing higher than this, it is therefore defacto their high end card.
 
What if the 580 is much faster under different loads?

I'm not sure why this matters. NVIDIA treated the 680 in every way, shape and form as their high end product. That unquestionably makes it their high end product. In the several cases where the 580 was faster, it could be argued that the 680 was a shoddy "high end" replacement.

If I wanted to snide, I'd instead point out that the "old high end" beating the "new high end" further proves that NVIDIA doesn't give two shits about serving their enthusiast members. THey'd rather squeeze another $500 for a "high end product" that cannot consistently beat their prior model.

But I realize that might be too logical for certain populous to swallow, which is (I'm assuming) why this whole game of "Well it's just a midrange part with a high end numbering scheme, high end pricing scheme, high end positioning, and high end comparisons being touted" is nice for the loyalists.

I'm not among the loyalists.
 
INVIDIA treated the 680 in every way, shape and form as their high end product

That is true simply because they have the interest in doing so. And there is no bigger damn lie than marketing.

The 680 behaves just as the new mid range which has lower, on par or a little bit higher performance than the old high end, simply due to the process shrink and architecture optimisations. But those factors don't really transform it into a proper high-end, just into a make-up one.
 
There's no excuses here for "it was only midrange"; NVIDIA made nothing higher than this, it is therefore defacto their high end card.

Wrong. The GK110 aka Titan was made and is higher than the GK104 aka GTX680.

Early reports suggest Nvidia has a real winner on its hands and that Kepler is such a strong performer, even Nvidia's mid-range cards will give AMD's high-end GPUs a run for their money.

http://www.maximumpc.com/article/ne...clear_winner_against_amds_tahiti_architecture
 
I'm not going to get into the pissing match of 680 versus 7970, however I am going to state this:

NVIDIA marketed the 680 as their high end card, they numbered it as their high end card, and they put a price tag on it commensurate with their high end card. For more than a year, it was their defacto high end card.

There's no excuses here for "it was only midrange"; NVIDIA made nothing higher than this, it is therefore defacto their high end card.


Without saying we have absoluletely no idea if GK100 = GK110 or not.. as we have never seen the spec of the GK100... We dont even know if it was possible to product it or not.. If it was working..
 
Last edited by a moderator:
NVIDIA treated the 680 in every way, shape and form as their high end product.

Yes because that made the most marketing sense given the competitive landscape, still it doesn't change the characteristics of the GK104 chip itself, that's all they needed and now they can sell their actual high end card for $1000. Titan is in now way a more special card than what a 3GB 580 was two years ago, much less so actually if you compare it to something like Evga 580 Classified.

Your logic has holes.
 
So we're just going to discount the 7970 "OC" as you put it because...?

What happens when you OC a 7950? Does it play in the same league as the 680? I'm sure it does.

And what happens, when you OC a 680? The point was and is that the smaller and less complex GK104 is playing in the same league as Tahiti. Not more, not less.

Yep I agree with most of it, but the problem is you continue to believe that Nvidia is somehow a mile ahead because their - in your own words - "beefed up midrange card" is - in my words - "15% slower than AMD's fastest card".

It's ridiculous to claim that the 7970 GHz doesn't count because it's "overclocked" - why not assume the 7970 was underclocked when most analysts realise that AMD was playing it extremly safe? You can still continue to overclock GHz editions, by far...and they will pull away from OC'd 680's. If Nvidia had enough good 680's to make an official overclocked SKU out of them they'd have done so months ago in order to compete with the GHz edition btw.

Please don't lie. I never said what you claim.
I simply corrected Mintmaster by saying and proving that the 680 and 7970 are equally fast. And by explaining why the 680 wasn't highend, given Nvidia's history as of late. That's it!
Btw, the 7970 GHz uses about 50W more than the 680. You could also say Nvidia has been playing it safe by keeping power consumption of the 680 in check. It's a matter of perspective.
 
What happens when you OC a 7950? Does it play in the same league as the 680? I'm sure it does.

And what happens, when you OC a 680? The point was and is that the smaller and less complex GK104 is playing in the same league as Tahiti. Not more, not less.
Of course it is, it's packing almost as much transistors in pure gaming chip, while Tahiti is much more than pure gaming chip
 
. But those factors don't really transform it into a proper high-end, just into a make-up one.

Again, this has nothing to do with anything. There was nothing else they made that was higher than the 680 for an entire year and then some. It was, defacto, their high end card.

A1xLLcqAgt0qc2RyMz0y said:
Wrong. The GK110 aka Titan was made and is higher than the GK104 aka GTX680
This method of argument is exactly the same as claiming the 480 wasn't a high end card because the 580 replaced it. The 680 was the high end card for a year. The Titan is now the high end replacement. Done and done.

Dr Evil said:
Yes because that made the most marketing sense given the competitive landscape, still it doesn't change the characteristics of the GK104 chip itself
Nothing you said has anything to do with anything. Please show me the HIGHER card than the 680 for that entire year, and you have proven your argument. There was none, defacto, 680 was the high end card.


Want to prove that the 680 isn't their high end card for all of 2012?
Show me their high end card, then. Until you can demonstrate the higher card, the 680 was as high as it got. This is, by absolute pure definition, their high end card. This simply cannot be argued any other way.
 
What happens when you OC a 7950? Does it play in the same league as the 680? I'm sure it does.

And what happens, when you OC a 680? The point was and is that the smaller and less complex GK104 is playing in the same league as Tahiti. Not more, not less.

Simply because AMD started out with an underclocked Tahiti chip and Nvidia took advantage of it by overclocking their GK104.

The 7970 GHz edition was 3 months later than the 680, 21% bigger die size and 15% faster. But that's a "win" for Nvidia right?

Same as the 8 month later Titan at 55% bigger and 30% faster than the 7970 GHz was a "win" for Nvidia as well. The double standards is just staggering.

Please don't lie. I never said what you claim.
I simply corrected Mintmaster by saying and proving that the 680 and 7970 are equally fast. And by explaining why the 680 wasn't highend, given Nvidia's history as of late. That's it!
No what you claimed was that the 680 was not high end, while suggesting that the 7970 non-GHz edition was AMD's best effort and the GHz edition was simply an overclocked version.
 
Nothing you said has anything to do with anything. Please show me the HIGHER card than the 680 for that entire year, and you have proven your argument. There was none, defacto, 680 was the high end card.


Want to prove that the 680 isn't their high end card for all of 2012?
Show me their high end card, then. Until you can demonstrate the higher card, the 680 was as high as it got. This is, by absolute pure definition, their high end card. This simply cannot be argued any other way.

What the hell are you talking about? I haven't said that it wasn't their high end card, I said why it WAS/turned out to be one.

However with high enough probability it wasn't originally designed to be one. Full process node shrink and it has what 18% more transistors, only 55% of the die size and same amount of memory BW as the 580 had. Compare that to past events.

You seem to only understand what the sticker says and are unwilling or unable to comprehend any factors behind the issue.

edit: The response to Mintmaster was now that Titan is already out, but also includes my thoughts about the origin of the 680. Obviously it is their high end single GPU card if there is nothing else on the market :)
 
Last edited by a moderator:
Back
Top