The 970/980 are going to be EOL'ed once the 1070 ships most likely. The 980Ti stopped production a month ago last I heard.
Maybe the 970 will get renamed to a 1060, but I doubt it. I think a 1060 will be missing until GP106 launches later this year.
A 300$ GPU like this is bad for AMD. This is a big issue for them. I hope that it is cheaper because the brand struggles overall.
My thinking is simple: the majority of AMD buyers are brand loyalists, just like the majority of nvidia's buyers. They(AMD) are betting too much on efficiency alone. This is not the right time to do this(market share wise): expect the 970 to sell even better.
While nvidia strengthens its brand further, increases the price and drops a bomb at the high end in order to sell over-priced mid to low end hardware, AMD tries to rise from the bottom with better price/$ and efficiency. This will not gain them market share today.
Increasing the price at the high end this day and age is the smarter choice, if you get the irony.
This is the natural way to go: nvidia is smarter, AMD is dumber. AMD should have never let go the high end. This is not the time to do this.
Why is GDDR5X Nvidia only? I haven't really been looking at that being exclusive.
Eventually I expect GDDR5x to replace GDDR5 in it's entirety, unless cost is a major roadblock, but that's still at least a couple of years away.It's not that it's Nvidia exclusive, it's that AMD isn't using it. It's in short supply right now and it doesn't make a lot of sense to use it for what's being branded as a mainstream part.
Also AMD has been developing HBM and is far more likely to use that in their high-performance part instead of GDDR5X. I imagine eventually that GDDR5X supply will increase and prices will drop and at that point you'll see more of it on low-end parts, especially once the clock speeds increase enough to allow GPUs to have smaller bus width and save die space.
... When the launch price of the r9 380 / 380x was $199 / 229 why in the world are people thinking this is a $300 card?!?
huh? There has been no launch price announced. But the "mainstream" market suggest an upper end price of $300.
It's not replacing 380/x but rather the 390/x's & thus will be priced accordingly.huh? There has been no launch price announced. But the "mainstream" market suggest an upper end price of $300.
It's not replacing 380/x but rather the 390/x's & thus will be priced accordingly.
A 300$ GPU like this is bad for AMD. This is a big issue for them. I hope that it is cheaper because the brand struggles overall.
My thinking is simple: the majority of AMD buyers are brand loyalists, just like the majority of nvidia's buyers. They(AMD) are betting too much on efficiency alone. This is not the right time to do this(market share wise): expect the 970 to sell even better.
While nvidia strengthens its brand further, increases the price and drops a bomb at the high end in order to sell over-priced mid to low end hardware, AMD tries to rise from the bottom with better price/$ and efficiency. This will not gain them market share today.
Increasing the price at the high end this day and age is the smarter choice, if you get the irony.
This is the natural way to go: nvidia is smarter, AMD is dumber. AMD should have never let go the high end. This is not the time to do this.
It's not replacing 380/x but rather the 390/x's & thus will be priced accordingly.
Why do people keep saying that it will have similar perf to 390 while still selling near $300?
AMD wants to lower the entry point for VR, to bring it to the masses
https://youtu.be/p010lp5uLQA?t=962
TAM = 7.5m for VR
So they want to bring 970/390 type of performance for much less.
"We can now produce GPUs which can run the minimum spec for lower cost, higher volume, consuming less power and running faster"
... When the launch price of the r9 380 / 380x was $199 / 229 why in the world are people thinking this is a $300 card?!?
Vega was pushed up to this year according to several rumors. So if it launches is October, thats not far off. Sooner than GP100/102 is currently expected to launch.
We're talking about price tier, at least initially, & not absolute performance, the latter always moves up a notch (or two) on a new node. So IMO we'll have, at launch ~In the 40 to 28nm transition the HD6970 was "replaced" by the HD7850 performance-wise at a lower price. I will be truly disappointed if the 480/X replaces the 390/X with similar performance and price.
AMD just updated their investor presentation, it clearly shows Vega in 2017.
We're talking about price tier, at least initially, & not absolute performance, the latter always moves up a notch (or two) on a new node. So IMO we'll have, at launch ~
Polaris 11 ~ 750(Ti) price bracket with performance between GTX 960 & R9 380/x
Polaris 10 ~ 970 price bracket with performance between 390/x & Fury(X)
You can add +/- 10% to the launch price of the Polaris' equivalent, price bracket, cards. If they perform lower, they will likely be priced lower than my estimates.
That's a reasonable price for the end user but not AMD. The Polaris' price will come down eventually, but no one (realistically) should expect the fully uncut Polaris to debut anywhere near 200$ as you'll set yourself up for some major disappointment.Ok, imagine you're amd. You have two costs: development costs, which are fixed regardless of the number of cards you sell and the actual cost of producing each card in terms of fabrication costs.
Right now AMD is losing money because they don't have enough market share - that is to say while they make enough money per card, their fixed cost of developing a new gpu is too high relative to the number of cards sold.
They have one goal right now: TAKE SHARE
They way they do that is by producing a card that is cheap to manufacture ( 232mm2) and hitting the mass market at a reasonable price (200 USD). Taking a 50% price hike from 380 to 480 would be idiotic and counterproductive when they're trying to remain relevant in the gpu space.
... When the launch price of the r9 380 / 380x was $199 / 229 why in the world are people thinking this is a $300 card?!?
This is fury performance on gen 0 drivers for well under 250 USD. This is a winner.
If 14nm wafers cost $5K.I dont know. Also, the 380 was $200 9 months ago. But it is based off the 285 which is damn near 2 years old and even back then it was only $250. In terms of today's value, its really only $150, max. Which means it couldnt possibly cost AMD more than $80 to buy those Tonga chips from the foundry. So now we move to 14nm and all the sudden these chips are double the price? I doubt it. I think AMD is paying $120 tops for Polaris 10 dies. And I think that is seriosuly highballing it, because Tonga was WAY bigger than Polaris 10, yet still lauched at $250. And they are not so greedy as to take a ~$100 chip and try to turn it into a $300 product. That's just not how AMD operates.
That's a reasonable price for the end user but not AMD. The Polaris' price will come down eventually, but no one realistically should expect the fully uncut Polaris to debut anywhere near 200$ as you'll set yourself up for some major disappointment.
Also why you bringing up the 380 again wrt P10's pricing? The P11, with performance up to 380x will be priced closer to 750Ti's launch price whilst the bigger Polaris may cost 2x as much. There will obviously be cut down variants occupying the vacant $ spots.
AMD just updated their investor presentation, it clearly shows Vega in 2017.
Do you have a link? The most recent one I have seen (That was released to everybody) was from Q1, which was almost two months ago..
Me three.I'd like to see it too.
If 14nm wafers cost $5K.
P10 yields [full die]
83/wafer @ 0.5 defects/cm^2 $60
101/wafer @ 0.4 defects/cm^2 $50
123/wafer @ 0.3 defects/cm^2 $41
152/wafer @ 0.2 defects/cm^2 $33
And you get all the cut die for free. If you assign a value to the cut, partially defective
die, this cost goes way down. The room certainly exists for $199 - $ 249 P10 models.