Nvidia will always have legion of people who don't understand technology, that will be their main market.
As opposed to all of those other tech companies that market exclusively to consumers who do understand technology?
nvidia = make stock holders happy . If you watched the market share , you can this release was pushed up since AMD has started to make up ground. THe release of the 1080 has nothig to do with consumers.
Where is the logic in this product ? It is dated design and driven by software.
Nvidia will always have legion of people who don't understand technology, that will be their main market.
As opposed to all of those other tech companies that market exclusively to consumers who do understand technology?
nvidia = make stock holders happy . If you watched the market share , you can this release was pushed up since AMD has started to make up ground. THe release of the 1080 has nothig to do with consumers.
Where is the logic in this product ? It is dated design and driven by software.
AMD's stock price jump was due mainly to a deal to sell a bunch of x86 servers in China. That's a market Nvidia doesn't even compete in.
The logic behind both Pascal and Polaris is the same: 14/16 nm process nodes are now available, and that die shrink offers huge gains in performance as compared to 28 nm.
The May 27 date is basically an announcement of when they'll start taking pre-orders. It's not at all similar to a date of when you can actually get a card and play games on it. If you really think that cards with GDDR5X are going to have real commercial availability less than two months after the first GDDR5X chips started sampling, I've got some ocean-front property in Kansas to sell you.
According to the following article some companies (it doesn't mention who but I'm sure nVidia's one of them) have been sampling GDDR5X since at least March
Nvidia will always have legion of people who don't understand technology, that will be their main market.
As opposed to all of those other tech companies that market exclusively to consumers who do understand technology?
nvidia = make stock holders happy . If you watched the market share , you can this release was pushed up since AMD has started to make up ground. THe release of the 1080 has nothig to do with consumers.
Where is the logic in this product ? It is dated design and driven by software.
AMD's stock price jump was due mainly to a deal to sell a bunch of x86 servers in China. That's a market Nvidia doesn't even compete in.
The logic behind both Pascal and Polaris is the same: 14/16 nm process nodes are now available, and that die shrink offers huge gains in performance as compared to 28 nm.
market share, not prices. AMDhas a Soc contract in China, that is the price jump. I'm talking about market shares. It's a little different.
To say nvidia is in panic, isn't true, but one has to question an entire press conference for one dated product.
The May 27 date is basically an announcement of when they'll start taking pre-orders. It's not at all similar to a date of when you can actually get a card and play games on it. If you really think that cards with GDDR5X are going to have real commercial availability less than two months after the first GDDR5X chips started sampling, I've got some ocean-front property in Kansas to sell you.
According to the following article some companies (it doesn't mention who but I'm sure nVidia's one of them) have been sampling GDDR5X since at least March
Sampling means that the earliest Nvidia could request GDDR5X chips and get them was late March. Before that, the chips didn't physically exist, or perhaps rather, the chips that did exist were lab test samples that didn't work right. Sampling is for extremely small volumes. You absolutely can get enough chips from sampling to test several cards and see if they work, and that's the point of sampling.
To launch a card that is actually available for the general public to buy, you don't just need a few DRAM chips, or even a few dozen. You need millions. That's not available until volume production, and Micron says they hope to have GDDR5X start volume production this summer. Not necessarily "you can get delivery of millions" this summer, but "we'll start production of millions".
The GDDR5X chips that Nvidia needs to do even a soft launch of the GeForce GTX 1080 won't physically exist in this universe until this summer at the earliest, and possibly not until fall. And even then, it takes time for wafers to go through a fab, Micron to ship them to Nvidia board partners, board partners to assemble boards, and then ship the completed boards around the world so you can buy one.
And that's assuming that everything goes flawlessly and you're just waiting on memory chips. Which probably won't be the case. GDDR5X sampling started in March. That means it wasn't physically possible for Nvidia to test their GDDR5X memory controller until March. First silicon almost invariably has problems, and the first GDDR5X controller ever built is not at all likely to work flawlessly. But Nvidia couldn't find out what was wrong and start fixing it until March.
Ordinarily, you'd want launch to happen several months after you start volume production of a GPU chip. That can't happen until after you fix the chip, and remember, Nvidia couldn't even find out what was broken in the GDDR5X controller until March. Add two months or so for a typical respin, and if you only need one, that's doing pretty well. I have no idea how many respins GP104 needs, but if we assume one, they could start volume production in May, start getting large numbers of chips back in July, and have time to accumulate enough stock for a hard launch in October--if they're lucky, and if they can get enough GDDR5X chips by then.
But a hard launch on May 27? Not happening. No way, no how, not possible. Or at least not if you mean May 27, 2016. The slide that announced the date didn't give the year.
There's no iron law that that you can't order volume production until you have the results of your latest respin and know that it's good. But if you hope your latest respin will be good and order volume production without waiting to see if it is, and you end up being wrong, you just wasted tens of millions of dollars on wafers you'll have to throw in the garbage. Maybe you'll take that risk if the respin only fixed one little thing and you're highly confident that it will work, but it's typically not a sane thing to do.
It is theoretically possible that Nvidia could have started volume production months ago before testing the GDDR5X controller, and just figure that most or all of the chips will probably have to be binned as GTX 1070s, and then plan on doing a respin to fix the GDDR5X controller once you can test it. (I'd be surprised if they did this, but it's at least theoretically possible.) So we don't know for certain that there can't possibly be a hard launch of the GTX 1070 in June. But the GTX 1080? Nope. Or at least not unless Nvidia changes the specs on it to use GDDR5 instead of GDDR5X.
Nvidia will always have legion of people who don't understand technology, that will be their main market.
As opposed to all of those other tech companies that market exclusively to consumers who do understand technology?
nvidia = make stock holders happy . If you watched the market share , you can this release was pushed up since AMD has started to make up ground. THe release of the 1080 has nothig to do with consumers.
Where is the logic in this product ? It is dated design and driven by software.
AMD's stock price jump was due mainly to a deal to sell a bunch of x86 servers in China. That's a market Nvidia doesn't even compete in.
The logic behind both Pascal and Polaris is the same: 14/16 nm process nodes are now available, and that die shrink offers huge gains in performance as compared to 28 nm.
market share, not prices. AMDhas a Soc contract in China, that is the price jump. I'm talking about market shares. It's a little different.
To say nvidia is in panic, isn't true, but one has to question an entire press conference for one dated product.
My mistake; I misread your post. Regardless, it's highly probable that AMD will gain a lot of market share with Polaris. AMD didn't create a whole new architecture in 2014-2015 like Nvidia did with Maxwell, so for Nvidia to have a new, better architecture competing against an old AMD architecture is why Nvidia's market share went so high. That's not going to be the case with Polaris competing against Pascal.
Nvidia will always have legion of people who don't understand technology, that will be their main market.
As opposed to all of those other tech companies that market exclusively to consumers who do understand technology?
nvidia = make stock holders happy . If you watched the market share , you can this release was pushed up since AMD has started to make up ground. THe release of the 1080 has nothig to do with consumers.
Where is the logic in this product ? It is dated design and driven by software.
AMD's stock price jump was due mainly to a deal to sell a bunch of x86 servers in China. That's a market Nvidia doesn't even compete in.
The logic behind both Pascal and Polaris is the same: 14/16 nm process nodes are now available, and that die shrink offers huge gains in performance as compared to 28 nm.
market share, not prices. AMDhas a Soc contract in China, that is the price jump. I'm talking about market shares. It's a little different.
To say nvidia is in panic, isn't true, but one has to question an entire press conference for one dated product.
My mistake; I misread your post. Regardless, it's highly probable that AMD will gain a lot of market share with Polaris. AMD didn't create a whole new architecture in 2014-2015 like Nvidia did with Maxwell, so for Nvidia to have a new, better architecture competing against an old AMD architecture is why Nvidia's market share went so high. That's not going to be the case with Polaris competing against Pascal.
You are also talking about Nvidia, a company who until they were caught on the 970 series memory issue, wasn't going to say a thing. You have to put in the equation the possiblity of Nvidia being shady and running out a batch of low quality component cards to fill demand.
I will go back logical thinking, nvidia had 1 single card to show today, it is just illogical waste of money .....
also n1080 Maximum GPU Temperature
94 C ; thats is crazy for this point in time. and this : TDP of 180 on one 8 pin rail... have fun replacing power supplies.
also n1080 Maximum GPU Temperature
94 C ; thats is crazy for this point in time. and this : TDP of 180 on one 8 pin rail... have fun replacing power supplies.
Literally everything is wrong with this post, 1080 is going to have ~300mm2 die size, while Fury X has 600mm2, they're not even in the same class, ...
That operating temperature has been the standard maximum operating temperature for gpus from both side for years now, you have no clue what you're talking about.
There's no such thing as the 8 pin rail, nor are there true divided 12v rails in PSUs any more, ...(not that it would matter for all but the worst, ancient PSUs which would die on you regardless, ...)
The 8 pin connector is tested for 150w power delivery and the graphics card gets 75w from the pci-e slot, so the 8 pin connector needs to deliver only 105w which is 2/3 of the wattage it's rated for, plenty of unneeded headroom don't you agree ?
AMD's stock price jump was due mainly to a deal to sell a bunch of x86 servers in China. That's a market Nvidia doesn't even compete in.
The logic behind both Pascal and Polaris is the same: 14/16 nm process nodes are now available, and that die shrink offers huge gains in performance as compared to 28 nm.
market share, not prices. AMDhas a Soc contract in China, that is the price jump. I'm talking about market shares. It's a little different.
To say nvidia is in panic, isn't true, but one has to question an entire press conference for one dated product.
How is going from a 28nm process to a 16nm process dated?
Do you even understand the implications of this and what it means? Just look at the progress of CPU's and incredible gains that have been made in both power efficiency and increased clock speeds. /facepalm
You are just a clueless troll. Never seen so many stupid posts in a single topic by a single person before. Tjeez.
AMD's stock price jump was due mainly to a deal to sell a bunch of x86 servers in China. That's a market Nvidia doesn't even compete in.
The logic behind both Pascal and Polaris is the same: 14/16 nm process nodes are now available, and that die shrink offers huge gains in performance as compared to 28 nm.
market share, not prices. AMDhas a Soc contract in China, that is the price jump. I'm talking about market shares. It's a little different.
To say nvidia is in panic, isn't true, but one has to question an entire press conference for one dated product.
How is going from a 28nm process to a 16nm process dated?
Do you even understand the implications of this and what it means? Just look at the progress of CPU's and incredible gains that have been made in both power efficiency and increased clock speeds. /facepalm
You are just a clueless troll. Never seen so many stupid posts in a single topic by a single person before. Tjeez.
did you even look at the actual die design ? Just curious...... talk to you later when polaris comes out .
Nvidia's answer is moar memory, just not going to cut it for games in 2017.
AMD's stock price jump was due mainly to a deal to sell a bunch of x86 servers in China. That's a market Nvidia doesn't even compete in.
The logic behind both Pascal and Polaris is the same: 14/16 nm process nodes are now available, and that die shrink offers huge gains in performance as compared to 28 nm.
market share, not prices. AMDhas a Soc contract in China, that is the price jump. I'm talking about market shares. It's a little different.
To say nvidia is in panic, isn't true, but one has to question an entire press conference for one dated product.
How is going from a 28nm process to a 16nm process dated?
Do you even understand the implications of this and what it means? Just look at the progress of CPU's and incredible gains that have been made in both power efficiency and increased clock speeds. /facepalm
You are just a clueless troll. Never seen so many stupid posts in a single topic by a single person before. Tjeez.
did you even look at the actual die design ? Just curious...... talk to you later when polaris comes out .
Nvidia's answer is moar memory, just not going to cut it for games in 2017.
How exactly is Pascal a "dated" architecture? It sure looks modern to me; it's not clear if it has top tier resource binding and I don't see anything analogous to GCN's global data store, but I haven't spotted anything else missing. It's certainly miles ahead of Kepler.
also n1080 Maximum GPU Temperature
94 C ; thats is crazy for this point in time. and this : TDP of 180 on one 8 pin rail... have fun replacing power supplies.
Looking at the die designs, Nvidia is thinking backwards as in the old days where adding more memory fixes everything, imho. More power with less power, with the design they chose, makes little sense.
The first of those is an entire chip (GeForce GTX 750 Ti), while the latter is a zoomed in view of a single compute unit. So they're not even comparable. Your Polaris link is so vague as to not mean much. As for Fiji, you're impressed by what, the color scheme?
The first of those is an entire chip (GeForce GTX 750 Ti), while the latter is a zoomed in view of a single compute unit. So they're not even comparable. Your Polaris link is so vague as to not mean much. As for Fiji, you're impressed by what, the color scheme?
Given that the 1070 - which will be using (readily) available GDDR5 won't be available until (we are told) June I suspect it isn't just the "not really available" GDDR5X that is in short supply.
1070 is far less questionable because its cut down chip so even damaged chips can be used for it (as well as non damaged chips) and GDDR5 is in abundance.
1070 is far less questionable because its cut down chip so even damaged chips can be used for it (as well as non damaged chips) and GDDR5 is in abundance.
The GTX 1070 isn't going to be waiting on memory, but they can't launch it before they have lots of working chips. To order a large production run before they can test the GDDR5X controller to see if it even works would be taking a huge risk, but if Nvidia didn't do that, we're not going to see real availability of the GTX 1070 for several months. I don't know what they did, but Nvidia describing the GTX 1070 similarly to the 1080 that is so obviously a paper launch doesn't give me much confidence in a hard launch.
How is going from a 28nm process to a 16nm process dated?
Do you even understand the implications of this and what it means? Just look at the progress of CPU's and incredible gains that have been made in both power efficiency and increased clock speeds. /facepalm
You are just a clueless troll. Never seen so many stupid posts in a single topic by a single person before. Tjeez.
did you even look at the actual die design ? Just curious...... talk to you later when polaris comes out .
Nvidia's answer is moar memory, just not going to cut it for games in 2017.
That's a patently ridiculous claim much like when you said the 970 was 4 years old. Both low end enthusiast cards from AMD and Nvidia (970 and 380X) will be viable for at least a couple of years to come, let alone the 1070 and 1080. While I'm not very impressed with the Nvidia reveal and don't think they'll be hard launching soon to say they won't be relevant for 2017 is just a ludicrous assertion.
Most Steam gamers don't even have a rig comparable to a 970 and 4th gen gaming i5/i7.
Why are you trying so hard to sell the 970's with messed up ram ? If you watch API development and the speed it is being developed, you would see 2017 is a pretty good target year for gpu releases. AMD released the Furyx duo just to release it, because I think their next gen is just that, next gen.
I'm order some crow to cook up for people to eat here in a month or two.
I mean seriously the amount of fanboying on both sides of this issue in this thread is facepalm worthy.
Quiz is still trying to claim you won't see a 1080 in even july, and that Polaris will be available in july, even though we've seen barely any info.
So basically TL:DR is "Nvidia is liars and they're going to epically screw up, and AMD are perfection and will magically release a card that we have almost no info on BEFORE NVidia releases their new cards"
I think I'm going to trust a reputable publication like Anandtech over a bunch of fanboys:
The first two cards out of the gate will be NVIDIA’s high-end cards, the GTX 1080 and GTX 1070. 1080p jokes aside, these are straightforward successors to what has now been NVIDIA’s product stack naming scheme since 2010, with the GTX 1080 representing their new high-end card while the GTX 1070 is a cheaper card meant to hit the enthusiast sweet spot. These cards will be launching over the next month, with GTX 1080 hitting the market May 27th, and GTX 1070 two weeks later, on June 10th. They will be priced at $599 and $379 respectively.
"The surest way to corrupt a youth is to instruct him to hold in higher esteem those who think alike than those who think differently."
"Hitting the market" isnt same as "available". Wait for Titan X, after you ordered it, had been what, 7-8 weeks (maybe even longer in some cases) after it "hit the market". Availability for 1080 might be a huge issue.
Comments
Where is the logic in this product ? It is dated design and driven by software.
The logic behind both Pascal and Polaris is the same: 14/16 nm process nodes are now available, and that die shrink offers huge gains in performance as compared to 28 nm.
Companies can opt to charge a fee if they wish. Charging however doesn't always work. Remember Rambus and RDRAM?
http://www.pcgamer.com/micron-rolling-out-gddr5x-memory-for-new-graphics-cards/
To say nvidia is in panic, isn't true, but one has to question an entire press conference for one dated product.
http://www.overclock3d.net/articles/gpu_displays/amd_gains_gpu_market_share/1
To launch a card that is actually available for the general public to buy, you don't just need a few DRAM chips, or even a few dozen. You need millions. That's not available until volume production, and Micron says they hope to have GDDR5X start volume production this summer. Not necessarily "you can get delivery of millions" this summer, but "we'll start production of millions".
The GDDR5X chips that Nvidia needs to do even a soft launch of the GeForce GTX 1080 won't physically exist in this universe until this summer at the earliest, and possibly not until fall. And even then, it takes time for wafers to go through a fab, Micron to ship them to Nvidia board partners, board partners to assemble boards, and then ship the completed boards around the world so you can buy one.
And that's assuming that everything goes flawlessly and you're just waiting on memory chips. Which probably won't be the case. GDDR5X sampling started in March. That means it wasn't physically possible for Nvidia to test their GDDR5X memory controller until March. First silicon almost invariably has problems, and the first GDDR5X controller ever built is not at all likely to work flawlessly. But Nvidia couldn't find out what was wrong and start fixing it until March.
Ordinarily, you'd want launch to happen several months after you start volume production of a GPU chip. That can't happen until after you fix the chip, and remember, Nvidia couldn't even find out what was broken in the GDDR5X controller until March. Add two months or so for a typical respin, and if you only need one, that's doing pretty well. I have no idea how many respins GP104 needs, but if we assume one, they could start volume production in May, start getting large numbers of chips back in July, and have time to accumulate enough stock for a hard launch in October--if they're lucky, and if they can get enough GDDR5X chips by then.
But a hard launch on May 27? Not happening. No way, no how, not possible. Or at least not if you mean May 27, 2016. The slide that announced the date didn't give the year.
There's no iron law that that you can't order volume production until you have the results of your latest respin and know that it's good. But if you hope your latest respin will be good and order volume production without waiting to see if it is, and you end up being wrong, you just wasted tens of millions of dollars on wafers you'll have to throw in the garbage. Maybe you'll take that risk if the respin only fixed one little thing and you're highly confident that it will work, but it's typically not a sane thing to do.
It is theoretically possible that Nvidia could have started volume production months ago before testing the GDDR5X controller, and just figure that most or all of the chips will probably have to be binned as GTX 1070s, and then plan on doing a respin to fix the GDDR5X controller once you can test it. (I'd be surprised if they did this, but it's at least theoretically possible.) So we don't know for certain that there can't possibly be a hard launch of the GTX 1070 in June. But the GTX 1080? Nope. Or at least not unless Nvidia changes the specs on it to use GDDR5 instead of GDDR5X.
I will go back logical thinking, nvidia had 1 single card to show today, it is just illogical waste of money .....
Logical press conference 2015 :
P.S. Lisa Su is a beast.
Fury x Transistor count : 8.9B
Nvidia 1080 transister count : 7.2 Billion
also n1080 Maximum GPU Temperature 94 C ; thats is crazy for this point in time.
and this : TDP of 180 on one 8 pin rail... have fun replacing power supplies.
That operating temperature has been the standard maximum operating temperature for gpus from both side for years now, you have no clue what you're talking about.
There's no such thing as the 8 pin rail, nor are there true divided 12v rails in PSUs any more, ...(not that it would matter for all but the worst, ancient PSUs which would die on you regardless, ...)
The 8 pin connector is tested for 150w power delivery and the graphics card gets 75w from the pci-e slot, so the 8 pin connector needs to deliver only 105w which is 2/3 of the wattage it's rated for, plenty of unneeded headroom don't you agree ?
How is going from a 28nm process to a 16nm process dated?
Do you even understand the implications of this and what it means? Just look at the progress of CPU's and incredible gains that have been made in both power efficiency and increased clock speeds. /facepalm
You are just a clueless troll. Never seen so many stupid posts in a single topic by a single person before. Tjeez.
Nvidia's answer is moar memory, just not going to cut it for games in 2017.
http://img.hexus.net/v2/graphics_cards/nvidia/m4xw3li750Tqi/Graphs/ArchB.png
Pascal :
http://cdn.wccftech.com/wp-content/uploads/2016/04/NVIDIA-Pascal-SMP.jpg
Basically 2 maxwells on 1 die. What kind of solution is that for future gaming. How does A-sync work ?
we are in the 21st century aren't we ?
Polaris :
http://1u88jj3r4db2x4txp44yqfj1.wpengine.netdna-cdn.com/wp-content/uploads/2016/01/amd-polaris-2.jpg
Easy to see how a-sync will work, as well as applying new die design to make better use of the actual die, instead of just going with moar ram.
Fiji looks like aliens made it :chuffed:
http://cdn.videocardz.com/1/2015/11/fiji-block-diagram.jpg
Looking at the die designs, Nvidia is thinking backwards as in the old days where adding more memory fixes everything, imho. More power with less power, with the design they chose, makes little sense.
Nvidia pascal with HBM
http://images.anandtech.com/doci/10222/gp100_block_diagram-1.png
yet they never showed the card ...... #hmm
Why are you trying so hard to sell the 970's with messed up ram ? If you watch API development and the speed it is being developed, you would see 2017 is a pretty good target year for gpu releases. AMD released the Furyx duo just to release it, because I think their next gen is just that, next gen.
Fury X Pro Duo
http://www.newegg.com/Product/Product.aspx?Item=N82E16814150768
The other thing Nvidia doesn't tell, if the games they showcase are using nvidia gameworks. Nvidia are shady.
nvm. Guess i won't be using unreal egine anytime soon lol
http://www.extremetech.com/gaming/178831-balkanized-gaming-nvidia-gameworks-now-a-core-part-of-ue4-amd-counters-with-mantle-integration-in-cryengine
I'm order some crow to cook up for people to eat here in a month or two.
I mean seriously the amount of fanboying on both sides of this issue in this thread is facepalm worthy.
Quiz is still trying to claim you won't see a 1080 in even july, and that Polaris will be available in july, even though we've seen barely any info.
So basically TL:DR is "Nvidia is liars and they're going to epically screw up, and AMD are perfection and will magically release a card that we have almost no info on BEFORE NVidia releases their new cards"
I think I'm going to trust a reputable publication like Anandtech over a bunch of fanboys:
http://www.anandtech.com/show/10304/nvidia-announces-the-geforce-gtx-1080-1070
The first two cards out of the gate will be NVIDIA’s high-end cards, the GTX 1080 and GTX 1070. 1080p jokes aside, these are straightforward successors to what has now been NVIDIA’s product stack naming scheme since 2010, with the GTX 1080 representing their new high-end card while the GTX 1070 is a cheaper card meant to hit the enthusiast sweet spot. These cards will be launching over the next month, with GTX 1080 hitting the market May 27th, and GTX 1070 two weeks later, on June 10th. They will be priced at $599 and $379 respectively.
"The surest way to corrupt a youth is to instruct him to hold in higher esteem those who think alike than those who think differently."
- Friedrich Nietzsche