390X coming soon few weeks

Either way, AMD is missing the boat big time. It's tax refund season, and many people are already out spending. With no information when the launch is, probably a good 1/3 of people looking to spend their refund money will not have ready cash available for a high end purchase.
 
I'm just glad that in this generation, we don't have every NV fanboy running around claiming the 980 was supposed to be the 960 but AMD had nothing so they sold it as the 980!

You seem confused about the progression of GPU SKU's from NVIDIA?



GK104-425-A2 (GTX 770 (294 mm²)) -> GM204-400 (GTX970/980 (398 mm²))
GK110-425-B1 (GTX780 (561 mm²)) - >GM200 (GTX990? (~550-600 mm²))

The SKU numbers and die-sizes indicates you are wrong.

If we go a generation further back it look like this:
GK104-400-A2 (GTX680 (294 mm²)) -> GK104-425-A2 (GTX 770 (294 mm²)) -> GM204-400 (GTX970/980 (398 mm²))

Notice anything in pacticulear?
Might help if we go a generation further back:
GF114 (GTX 560 Ti(332 mm²)) -> GK104-400-A2 (GTX680 (294 mm²)) -> GK104-425-A2 (GTX 770 (294 mm²)) -> GM204-400 (GTX970/980 (398 mm²))

Every GPU with the moniker Gxxx4 has been a "midrange" SKU from NVIDIA
Every GPU with the moniker Gxxx0 has been the "highend" SKU from NVIDIA
Every "highend" SKU from NVIDA the last generations have been +500 mm²

I remember the same claims when the GTX680 launched...a lot of noise that suddenly got all quiet with the release of Titan/GTX780Ti.

TL;DR:
You are wrong.
 
For the 400 and 500 series, Gxxx4 was reserved for midrange cards, while the big die Gxxx0 were for high end cards. Every card that was x70 and above used the Gxxx0 big die, while x60 and below used the Gxxx4 die. (except for the 465 based on GF100 and the 560 Ti 448 core based on GF110)

But for the 600 series, nVidia changed tactics, and never released big die GK110, instead the flagship 680 was based on a full GK104. The 600 series skipped GK110 altogether, and it wasn't until the 700 series that GK110 was seen. And even then the trend broke once again because the 770 was still based on a full GK104, albeit with a factory overclock on both the core and memory to boost performance relative to the 680. The only GK110 in the 700 series lineup were the Titan cards, 780 and 780 Ti.

He's saying we're seeing a repeat of history -- nVidia is once again releasing the mid-range Gxxx4 cards first and branding them as if they were high end cards by giving the x80 moniker to the full GM204 chip (much like what they did with GK104 and 680). And then they'll release the big die cards 6 months later and slap on Ti after the x80 or whatever they want to call it this round to milk even more money.

So no he's not wrong, you just didn't comprehend what he said correctly.
 
Last edited:
You seem confused about the progression of GPU SKU's from NVIDIA?



GK104-425-A2 (GTX 770 (294 mm²)) -> GM204-400 (GTX970/980 (398 mm²))
GK110-425-B1 (GTX780 (561 mm²)) - >GM200 (GTX990? (~550-600 mm²))

The SKU numbers and die-sizes indicates you are wrong.

If we go a generation further back it look like this:
GK104-400-A2 (GTX680 (294 mm²)) -> GK104-425-A2 (GTX 770 (294 mm²)) -> GM204-400 (GTX970/980 (398 mm²))

Notice anything in pacticulear?
Might help if we go a generation further back:
GF114 (GTX 560 Ti(332 mm²)) -> GK104-400-A2 (GTX680 (294 mm²)) -> GK104-425-A2 (GTX 770 (294 mm²)) -> GM204-400 (GTX970/980 (398 mm²))

Every GPU with the moniker Gxxx4 has been a "midrange" SKU from NVIDIA
Every GPU with the moniker Gxxx0 has been the "highend" SKU from NVIDIA
Every "highend" SKU from NVIDA the last generations have been +500 mm²

I remember the same claims when the GTX680 launched...a lot of noise that suddenly got all quiet with the release of Titan/GTX780Ti.

TL;DR:
You are wrong.

That is not fact. More like a stab in the dark.

If you go back further than that, GTX 465 used GF100.

GTX 560Ti had multiple variations, two of them included GF110.

The GTX 680 used GK104 which was their flagship GPU for 11 months.

The GTX Titan release was actually the new flagship "High end" card when it released and it was the flagship card for about 8 months till the GTX 780Ti came out.

http://en.wikipedia.org/wiki/List_of_Nvidia_graphics_processing_units

Processor code name and die size have a lot less correlation than you are giving them.
 
The point is that Nvidia released the best they had at the time, making it their "highend" GPU for that time period.
GK100 was scrapped and GK110 was barely able to make the Q3 '12 deadline for Oakridge and still needed 2-3revisions before it was ready for a consumer market launch.

In the overall scheme of things when looking at the generational GPU family lineup, yes the Gxxx4 is typically "midrange."
 
That is not fact. More like a stab in the dark.

If you go back further than that, GTX 465 used GF100.

GTX 560Ti had multiple variations, two of them included GF110.

The GTX 680 used GK104 which was their flagship GPU for 11 months.

The GTX Titan release was actually the new flagship "High end" card when it released and it was the flagship card for about 8 months till the GTX 780Ti came out.

http://en.wikipedia.org/wiki/List_of_Nvidia_graphics_processing_units

Processor code name and die size have a lot less correlation than you are giving them.

That is why I combined the SKU name with the diesize.
The only GTX 560 Ti that used diffrent SKU was a limited OEM version.

The GTX680 is a bad example to use, as the GK104 moniker told me (and others) that this wasn't the highend Kepler SKU...and hence I went from a GF110 to a GK110 (and not a sidegrade to GK104)

Even if forums was full of people saying there were no "highend" GKxx0 SKU in the works, that GF104 was the high end.

If you notice, Nvidia seems to have kinda made their own version of Intel's "Tick-Tock" now.

Midrange SKU (GK104 - Tick) - Highend SKU (GK110 - Tock) - Midrange SKU (GM204 - Tick) - Highend SKU (GM200- Tock)

I will wager that we will see the same trend with Pascal, so the trend will look like this:
Midrange SKU (GK104 - Tick) -
Highend SKU (GK110 - Tock) -
Midrange SKU (GM204 - Tick) -
Highend SKU (GM200- Tock) -
Midrange SKU (GPxx4 - Tick) -
Highend SKU (GPxx0) - Tock) -
(repeat with next gen)

I think I will revisit this topic when Pascal launches, to see if my prediction was right
 
That is why I combined the SKU name with the diesize.
The only GTX 560 Ti that used diffrent SKU was a limited OEM version.

The GTX680 is a bad example to use, as the GK104 moniker told me (and others) that this wasn't the highend Kepler SKU...and hence I went from a GF110 to a GK110 (and not a sidegrade to GK104)

Even if forums was full of people saying there were no "highend" GKxx0 SKU in the works, that GF104 was the high end.

If you notice, Nvidia seems to have kinda made their own version of Intel's "Tick-Tock" now.

Midrange SKU (GK104 - Tick) - Highend SKU (GK110 - Tock) - Midrange SKU (GM204 - Tick) - Highend SKU (GM200- Tock)

I will wager that we will see the same trend with Pascal, so the trend will look like this:
Midrange SKU (GK104 - Tick) -
Highend SKU (GK110 - Tock) -
Midrange SKU (GM204 - Tick) -
Highend SKU (GM200- Tock) -
Midrange SKU (GPxx4 - Tick) -
Highend SKU (GPxx0) - Tock) -
(repeat with next gen)

I think I will revisit this topic when Pascal launches, to see if my prediction was right

The limited OEM and the CORE 448, were the two models of the GTX 560Ti with GF110.

High-end is what is considered the top contender at the time. The performance crown. You are not using that correctly... It's considered the top tier of performance for it's time in that generation. Not for the duration of the architecture.

A new generation comes out about once every year, with a few revisions and modified cards in the middle. A new micro architecture comes out once about every two years. This isn't anything new. We will see a revision of Maxwell in Q4 2015 - Q1 2016.
 
Either way, AMD is missing the boat big time. It's tax refund season, and many people are already out spending. With no information when the launch is, probably a good 1/3 of people looking to spend their refund money will not have ready cash available for a high end purchase.

... and this is why AMD are not doing very well financially... always wrong place at wrong time and also poor marketing.
 
Exactly. China and India are the top of the fastest growing superpower economies.
India actually just overtook China 2 quarters ago.

This was due to that.



Nope.

That's the entire Asia Pacific share vs North America (basically USA). In essence one country is 20% of their AIB dollars. That's very significant. And as for India, it had a short good run recently but it's still mired in corruption and red tape with disproportionate poverty and lack of infrastructure. I wouldn't try comparing it to China. China has a huge industrial base, India has call centers and tea stands.
 
India has call centers and tea stands.
Nope.


India has lots of problems but none of them have anything to do with a un-diversified economy. To call any of the BRICS superpower economies is a bit of a stretch though. Growth looks to be leveling out too in some of the bigger ones (ie. China, maybe India, Russia isn't looking too hot either ATM) and at least in China it appears as if their housing bubble is in the process of popping which is going to have interesting consequences to say the least for them and the global economy.

Of course none of that really matters too much for AMD with Fiji. It is after all going to be a very high end low volume part when its first released. They don't need to sell volume to make money off of them. And the PC enthusiast market never really cared too much about holidays or tax seasons before when making their buying decisions. There is no reason to believe it'll suddenly matter now.
 
Nothing like a good change of subject...

The BRICs are growing fast indeed, but their economies can't be compared to those of the west yet.

The biggest reasons for their growth have been because they are comparatively "low cost" places to do business. The sizes of their economies are simply because there are so many people, not because their individual standards of living and as such abilities to spend on technology.

Wake me when the GNP/capita gets anywhere close to US/Europe/Japan standards.
 
Yeah, probably shouldn't have gone with "super-power"...

Anyway, just pointing out that they are major emerging markets with the potential to get much bigger.
 
Nope.


India has lots of problems but none of them have anything to do with a un-diversified economy. To call any of the BRICS superpower economies is a bit of a stretch though. Growth looks to be leveling out too in some of the bigger ones (ie. China, maybe India, Russia isn't looking too hot either ATM) and at least in China it appears as if their housing bubble is in the process of popping which is going to have interesting consequences to say the least for them and the global economy.

Of course none of that really matters too much for AMD with Fiji. It is after all going to be a very high end low volume part when its first released. They don't need to sell volume to make money off of them. And the PC enthusiast market never really cared too much about holidays or tax seasons before when making their buying decisions. There is no reason to believe it'll suddenly matter now.


You know my response wasn't supposed to be taken LITERALLY right? Read the rest of what I wrote, especially the lack of an industrial base in India. Your graph just backs up my statement more than anything else with industrials only comprising 5.93% of it's total economy. Now go fetch a Chinese one and compare, bet they will be world's apart. China makes and exports stuff to the world, India has call centers, cheap IT guys, some cheap cotton shirts/pants and..nothing else worth talking about. Even Modi recognizes this and that's why he's pushing the "Made in India" campaign because he realizes India lacks a strong industrial base. The US economy has been weakened because it relied too much on the service sector bs and exported it's industry to cheaper labor countries.

So getting back on topic after having established the US does make up a huge portion of AIB income, part of AMD hurting so much could very well be their timing. Some people are willing to spend more money at certain times of the year (tax refund season, holidays) and AMD consistently misses those targets with anything exciting on the market. Regardless, AMD better do something quick and knock it out of the ballpark with the 390X because things aren't looking too good right now: http://www.statista.com/statistics/...aphics-card-shipments-since-3rd-quarter-2010/ look at that huge drop in market share in just one quarter - I bet Q4 2014 will show an even bigger decline.

Edit: Yep I was right, things are worse than I thought -

Capture4(1).JPG


Finally I'll close with some personal anecdotal evidence. While I typically only buy NVIDIA hardware (was burned by AMD mobile xfire), I would still have considered R390X a month ago when I was building a new computer. But since the only thing on the market was the old 290X and no freesync displays on the market, I went with a G-Sync display + 2 NVIDIA graphics cards. That's a lot of money lost for AMD from just one sale.
 
Last edited:
The limited OEM and the CORE 448, were the two models of the GTX 560Ti with GF110.

High-end is what is considered the top contender at the time. The performance crown. You are not using that correctly... It's considered the top tier of performance for it's time in that generation. Not for the duration of the architecture.

A new generation comes out about once every year, with a few revisions and modified cards in the middle. A new micro architecture comes out once about every two years. This isn't anything new. We will see a revision of Maxwell in Q4 2015 - Q1 2016.

There is already a new revision of Maxwell out.

Maxwell V1 = GM107
Maxwell V2 = GM204

We could try adding one more parameter, that support my claim and goes against your claim:



Midrange SKU (GF114 - Tick) - 332 mm² - 256 bit / 128.26 GB/s
Highend SKU (GF100 - Tock) - 520 mm² - 384 bit / 192.384 GB/s
Midrange SKU (GK104 - Tick) - 294 mm² - 256 bit / 192.256 GB/s
Highend SKU (GK110 - Tock) - 561 mm² - 384 bit / 336.4 GB/s
Midrange SKU (GM204 - Tick) - 398 mm² - 256 bit / 224 GB/s
Highend SKU (GM200- Tock) - 550 mm² - 384 bit / 336 GB/s
Midrange SKU (GPxx4 - Tick) - ~300 mm² - (stacked memory, details not public yet)
Highend SKU (GPxx0) - Tock) - ~550 mm² - (stacked memory, details not public yet)

My upgradepath:
GF100 -> GK110 -> GM200 (true highend upgrade path)
A buyer on more restricted bugdet should go:
GF114 -> GK104 -> GM204 (midrange upgrade path)

The only reason Nvidia can get away with selling Gxxx4 SKU's as "highend" is because Nvidia's midrange SKU's can compete with AMD's highend SKU's.

Again, look at the metrics.
 
Last edited:
I'm hoping for a June release of the new AMD lineup though if it gets delayed to September then it'll be two years since the release of the 290X. And it was about two years to the 7970 before that. AMD seems to be keeping to a roughly even timetable with their high-end releases. Looking forward to building the new system when the parts do finally become available sometime this year though.
 
The only reason Nvidia can get away with selling Gxxx4 SKU's as highend is because Nvidia's midrange SKU's can compete with AMD's highend SKU's.

Again, look at the metrics.

"Highend"

I'm hoping for a June release of the new AMD lineup though if it gets delayed to September then it'll be two years since the release of the 290X. And it was about two years to the 7970 before that. AMD seems to be keeping to a roughly even timetable with their high-end releases. Looking forward to building the new system when the parts do finally become available sometime this year though.
Q3 is BS.
 
There is only one new card ?

Can you link to the press release from AMD saying that they will launch in a few weeks..

If they will not paperlaunch, or at least show it on GDC, then screw this, I'm getting 980. Won't be waiting on 290 till Fall.
 
If they will not paperlaunch, or at least show it on GDC, then screw this, I'm getting 980. Won't be waiting on 290 till Fall.

980 is only ~34% better than a 290? That's almost in the neglible difference zone for me. But to each their own! :)

I also realize I am a hypocrite because I went for the 980 over 970 which is only 18% faster for 58% most cost. bahahah
 
You know my response wasn't supposed to be taken LITERALLY right?
Since you gave no indication you were being sarcastic, nope. You can't fault people for reading things as they written. Otherwise it just turns into a cluster-fuck of 'reading between the lines' and tone arguments which shits up the thread for pages at a time.

Your graph just backs up my statement more than anything else with industrials only comprising 5.93% of it's total economy.
You're not making any sense or backing up your original post in any way shape or form.

Just because industry only accounts for ~6% of their market doesn't mean their economy is un-diversified or made up of 'call centers and tea stands'. You're ignoring the other ~94% of which call centers only make up around ~2.4%.

A economy heavily based around any given sector is fragile and to be avoided, even if that sector happens to be industry.

The US economy has been weakened because it relied too much on the service sector bs and exported it's industry to cheaper labor countries.
Nope.


The US was only relatively recently surpassed by China in manufacturing. Everyone else is a distant 3rd or 4th compared to the US and China in this respect. Its true that less labor is involved in manufacturing now but this is due to vastly improved automation and so called 'dark factories' which are operated with virtually no manpower on site for weeks or months at a time. This is why output has improved even though labor has dropped massively in industry.


The fact of the matter is the move to a service economy was a requirement, not a option, as the manufacturing jobs were being eliminated or outsourced en masse by the early to mid 90's.

What has trashed the US economy is 2 consecutive bubbles (.com and the Global Credit Crisis, the latter mostly manifested as a housing bubble in the US) which has destroyed any savings by old and younger generations and declining and/or stagnant wages while the cost of living (housing, college, medical care in particular) rose.

So getting back on topic after having established the US does make up a huge portion of AIB income
You never established this at all. Making claims isn't the same as proof. What has been shown in earlier links is the US makes up a significant portion but a huge portion of AIB shipments.

Also AIB shipments by country have nothing inherently to do with product launch timing. That is a whole other issue entirely. BTW:

Market share is graphed here but sales would show up too and there is little to no sign of seasonality or market timing having an impact.

PC enthusiasts in general don't buy based on the tax season or holiday's. They buy based on bang for the buck when they upgrade or because they want the latest greatest item. This is especially true of the sort of PC enthusiast who is capable and willing to spend the $500+ that is often required for a new top end card at launch.

Finally I'll close with some personal anecdotal evidence.
No one cares about them on this issue. Don't bother posting if that is all you've got.
 
980 is only ~34% better than a 290? That's almost in the neglible difference zone for me. But to each their own! :)

I also realize I am a hypocrite because I went for the 980 over 970 which is only 18% faster for 58% most cost. bahahah

It depends what resolution you run in.

If you run at 4k, or use DSR, then basically the 290x and 980 GTX are neck and neck.

AMD has 4k+ resolution figured out. Nvidia is a lil bit behind in that category.
 
It depends what resolution you run in.

If you run at 4k, or use DSR, then basically the 290x and 980 GTX are neck and neck.

AMD has 4k+ resolution figured out. Nvidia is a lil bit behind in that category.

At stock I believe that's about correct @ 1080P. OC vs OC the 980 was 24.x% ahead of the 290x.

Then of course [H] noted the 290x crossfire was smoother than 980 SLI at 4k.

Pretty impressive really for how old the 290x is and price point.
 
There is already a new revision of Maxwell out.

Maxwell V1 = GM107
Maxwell V2 = GM204

We could try adding one more parameter, that support my claim and goes against your claim:



Midrange SKU (GF114 - Tick) - 332 mm² - 256 bit / 128.26 GB/s
Highend SKU (GF100 - Tock) - 520 mm² - 384 bit / 192.384 GB/s
Midrange SKU (GK104 - Tick) - 294 mm² - 256 bit / 192.256 GB/s
Highend SKU (GK110 - Tock) - 561 mm² - 384 bit / 336.4 GB/s
Midrange SKU (GM204 - Tick) - 398 mm² - 256 bit / 224 GB/s
Highend SKU (GM200- Tock) - 550 mm² - 384 bit / 336 GB/s
Midrange SKU (GPxx4 - Tick) - ~300 mm² - (stacked memory, details not public yet)
Highend SKU (GPxx0) - Tock) - ~550 mm² - (stacked memory, details not public yet)

My upgradepath:
GF100 -> GK110 -> GM200 (true highend upgrade path)
A buyer on more restricted bugdet should go:
GF114 -> GK104 -> GM204 (midrange upgrade path)

The only reason Nvidia can get away with selling Gxxx4 SKU's as "highend" is because Nvidia's midrange SKU's can compete with AMD's highend SKU's.

Again, look at the metrics.

I never understood this logic. If you look purely at gaming performance, does it matter even one bit if a card is a Gxx04 or a Gxx00 based chip? Nvidia gets away with selling 04 sku's as "high end" because the 04 sku's beat their own previous "high end" sku's at gaming. If you need douple precision flops, it'd be reasonable to argue that an 04 sku is inferior to an 00 sku, but this idea that the 980 isn't a "true" high-end card just boggles the mind. What is a "high end" GPU if not the fastest gaming GPU in the market? If it walks like a duck, quacks like a duck.....
 
I never understood this logic. If you look purely at gaming performance, does it matter even one bit if a card is a Gxx04 or a Gxx00 based chip? Nvidia gets away with selling 04 sku's as "high end" because the 04 sku's beat their own previous "high end" sku's at gaming. If you need douple precision flops, it'd be reasonable to argue that an 04 sku is inferior to an 00 sku, but this idea that the 980 isn't a "true" high-end card just boggles the mind. What is a "high end" GPU if not the fastest gaming GPU in the market? If it walks like a duck, quacks like a duck.....

that depends on if the duck is green, then it's really a phoenix.
 
I never understood this logic. If you look purely at gaming performance, does it matter even one bit if a card is a Gxx04 or a Gxx00 based chip? Nvidia gets away with selling 04 sku's as "high end" because the 04 sku's beat their own previous "high end" sku's at gaming. If you need douple precision flops, it'd be reasonable to argue that an 04 sku is inferior to an 00 sku, but this idea that the 980 isn't a "true" high-end card just boggles the mind. What is a "high end" GPU if not the fastest gaming GPU in the market? If it walks like a duck, quacks like a duck.....

I've been debating this to myself. I've been looking forward to "big maxwell" but in reality it almost makes sense to get a 04 sku for TDP reasons. If every tick and tock is 30% higher who really cares if it's a tick or a tock.
 
I never understood this logic. If you look purely at gaming performance, does it matter even one bit if a card is a Gxx04 or a Gxx00 based chip? Nvidia gets away with selling 04 sku's as "high end" because the 04 sku's beat their own previous "high end" sku's at gaming. If you need douple precision flops, it'd be reasonable to argue that an 04 sku is inferior to an 00 sku, but this idea that the 980 isn't a "true" high-end card just boggles the mind. What is a "high end" GPU if not the fastest gaming GPU in the market? If it walks like a duck, quacks like a duck.....

I do agree with you to some extent. A rose by any other name and so forth. That said, this is in part the reason we no longer see the massive performance jump between generations e.g. 7800 to 8800 back when nV used to kick off with their first string tech.
 
I do agree with you to some extent. A rose by any other name and so forth. That said, this is in part the reason we no longer see the massive performance jump between generations e.g. 7800 to 8800 back when nV used to kick off with their first string tech.

You're probably right, as is Factum when he says that one reason for nV's strategy is AMD's inability to really compete across all metrics, so they can live off the evolutionary improvements in their designs. I also think a big (maybe the biggest) reason is the slowing of process shrinks. Its easy to throw transistors at an application and get huge benefits when you're shrinking every couple years, but when you're stuck at one node like we've been stuck on 28nm its a different story. Then you get this kind of "tick-tock" cadence where you're putting out a new arch (i.e. 00/10 sku's) then refining certain aspects (i.e. 04 sku's) without ever really getting generational leaps.
 
I never understood this logic. If you look purely at gaming performance, does it matter even one bit if a card is a Gxx04 or a Gxx00 based chip? Nvidia gets away with selling 04 sku's as "high end" because the 04 sku's beat their own previous "high end" sku's at gaming. If you need douple precision flops, it'd be reasonable to argue that an 04 sku is inferior to an 00 sku, but this idea that the 980 isn't a "true" high-end card just boggles the mind. What is a "high end" GPU if not the fastest gaming GPU in the market? If it walks like a duck, quacks like a duck.....

It's really simple.
NVIDIA's new midrange SKU's have always matched/surpassed the previous highend SKU's., nothing new here.

So is the progression of SKU's, as far back as I can remember.
Look at the specs again.

I know a lot of people were in denial wtih the GTX680 (GK104) and claimed that it WAS the high end "Kepler", based on the performance compared to AMD's highend SKU.
But it wasn't.
Highend "Kepler" was the GK110.

I was on a GF100, hence I did NOT bo for the GK104.
The SKU name (GK114), the memory buswith (265 bit) and the dieszie (294 mm²) told me that.
I knew when the GTX680 launched that this was NOT the highend SKU to replace GF100.
That would come later with a GKxx0 SKU nad I was right.

So I didn't do a "sidegrade", but I got a true upgrade.

The same reason why I didn't go for the GM204, it would not be a true upgrade for a GK110.
That comes with the GM200.

So, once again, this is a highend upgrade path:
GF100 -> GK110 -> GM200

The midrange upgradepath is this:
GF114 -> GK104 -> GM204

NIVIDA even shows this, as they chosse the GTX680 as the "obvious" upgradepath for a GTX980.

http://www.pcper.com/reviews/Graphi...and-GTX-970-GM204-Review-Power-and-Efficiency

If you cannot look beyond the relative performance between AMD/NVIDIA and ignore the technical specifications of the SKU's themself then you will never understand it.

Again, look at the specs.
NVIDIA midrange SKU's: (332 mm² - 256 bit) -> (294 mm² - 256 bit) -> (398 mm² - 256 bit)
NVIDIA highend SKU's (520 mm² - 384 bit ) - > (561 mm² - 384 bit) -> (550 mm² - 384 bit)

I can explain it too you, but I can not understand it for you, sorry.

TL;DR
Don't look at the performance of AMD vs NVIDIA (it only speaks to the relative relationship between AMD and NVIDA at that specific time)...look at the technical specs between NIVIDA SKU's

I predict this "tick-tock" will continue with "Pascal":
GF114 -> GK104 -> GM204 -> GPxx4
GF100 -> GK110 -> GM200 -> GPxx0
 
If they do not release a higher end SKU then the highest performing is their high-end SKU.
Atop of that when the GK104 came out it was obvious that the GK110 would replace it, but we had no idea of knowing if that card was even ready or not.
If GK110 was not launched then due to being incomplete then that makes the GK104 Nvidias high-end GPU.
That is reality, reality and technical road maps typically differ.
 
Last edited:
If they do not release a higher end SKU then the highest performing is their high-end SKU.
Atop of that when the GK104 came out it was obvious that the GK110 would replace it, but we had no idea of knowing if that card was even ready or not.
If GK110 was not launched then due to being incomplete then that makes the GK104 Nvidias high-end GPU.
That is reality, reality and technical road maps typically differ.

You seem to think this is a marketing debate, not a technical one..

The GK110 launched as a QUADRO cards first (and before that a K20X tesla...more profit in those markets than the Geforce market...hence the "long" wait.):
http://www.anandtech.com/show/7166/nvidia-announces-quadro-k6000

In NVIDIA’s product hierarchy, the Quadro 6000 cards hold the position of NVIDIA’s most powerful products. They’re not just the flagship cards for the Quadro family, but really the flagship for the entire generation of GPUs, possessing the compute functionality of Tesla combined with the graphics functionality of GeForce/Quadro, and powered by what’s typically the single most powerful GPU configuration NVIDIA offers. They’re unabashedly high end – and have a price tag to match – but in many ways they’re the capstone of a generation. It should come as no surprise then that with the Quadro K6000, NVIDIA is looking to launch what will become the king of the Keplers.

It's really not that hard, when you lok at the technical data and forget about PR or AMD's performance.

If I do a search on your posts, will I find you posting at the launch of the GTX680 that is was the "highend" kepler"?
Even if we all know that to be false?
 
Since you gave no indication you were being sarcastic, nope. You can't fault people for reading things as they written. Otherwise it just turns into a cluster-fuck of 'reading between the lines' and tone arguments which shits up the thread for pages at a time.

Really? You took "Tea Stands" literally as a foundation for an economy? :rolleyes:


You're not making any sense or backing up your original post in any way shape or form.

Just because industry only accounts for ~6% of their market doesn't mean their economy is un-diversified or made up of 'call centers and tea stands'. You're ignoring the other ~94% of which call centers only make up around ~2.4%.

Doesn't matter if it's diversified or not, that wasn't my point. Point is India cannot be compared to China as it doesn't have the infrastructure or manufacturing industry that China does. This is something India itself recognizes but I guess you know better (or pretend to).

Nope.


The US was only relatively recently surpassed by China in manufacturing.
The fact of the matter is the move to a service economy was a requirement, not a option, as the manufacturing jobs were being eliminated or outsourced en masse by the early to mid 90's.

Too funny, it was a "requirement" to shift to a service based economy due to loss of manufacturing jobs yet you try to argue that it doesn't matter because output is fine. :confused:

http://www.usitc.gov/research_and_a...ecline of U.S. Manufacturing Employment_0.pdf
This paper finds a link between the sharp drop in U.S. manufacturing employment after 2001 and the elimination of trade policy uncertainty resulting from the U.S. granting of permanent normal trade relations to China in late 2000. We find that industries where the threat of tariff hikes declines the most experience greater employment loss due to suppressed job creation, exaggerated job destruction and a substitution away from low-skill workers. We show that these policy-related employment losses coincide with a relative acceleration of U.S. imports from China, the number of U.S. firms importing from China, the number of Chinese firms exporting to the U.S., and the number of U.S.-China importer-exporter pairs.

Also this: http://www2.itif.org/2012-american-manufacturing-decline.pdf

Even in these sophisticated areas, U.S. manufacturing leadership is in peril. Correcting for biases in the official data, ITIF finds that from 2000 to 2010, U.S. manufacturing labor
productivity growth was overstated by a remarkable 122 percent.

...


Lamentably, the state of American manufacturing—and by extension the American
economy—has been seriously misdiagnosed. In fact, the idea that “all is well” is faulty on
two counts. First, even when relying on official U.S. government data, it is clear that
manufacturing output growth has lagged this decade, particularly in a number of key
sectors. Second, and more importantly, it is increasingly clear that there are substantial
upward biases in the U.S. government’s official statistics and that real manufacturing
output and productivity growth is significantly overstated. The most serious bias relates to
the computers and electronics industry (NAICS 334)—its output is vastly overstated.
Correcting for these statistical biases, we see that the base of U.S. manufacturing has eroded faster over the past decade than at any time since WWII, when the United States began compiling the statistics. In other words, the massive loss of jobs is not due to productivity alone.
uW8TKMQ.jpg

d51W4aX.jpg

7sphL51.jpg



What has trashed the US economy is 2 consecutive bubbles (.com and the Global Credit Crisis, the latter mostly manifested as a housing bubble in the US) which has destroyed any savings by old and younger generations and declining and/or stagnant wages while the cost of living (housing, college, medical care in particular) rose.

That's only part of the problem, not all of it like you attribute.


You never established this at all. Making claims isn't the same as proof. What has been shown in earlier links is the US makes up a significant portion but a huge portion of AIB shipments.

Right, so you agree with me that the US is a significant portion of the AIB market. So what exactly are you arguing again? :rolleyes:

Also AIB shipments by country have nothing inherently to do with product launch timing. That is a whole other issue entirely. BTW:

Market share is graphed here but sales would show up too and there is little to no sign of seasonality or market timing having an impact.

Look at the quarter to quarter AIB changes, they typically peak at Q3-Q4 when AIBs stock up for seasonal sales:

GPU-AIB-shipments-JPR.png


This doesn't break it down by vendor but seeing individual company results, we can clearly see AMD contracting. In fact, given the terrible attach rates and eroding AIB market, it makes sense that there's only really room for one player.

Apologies to everyone for the OT discussion, this is my last post on the subject as not to drag it out.
 
Last edited:
Back
Top