How much do you think it will cost?

Your house.
Hope you have insurance.

>>56229923
the housfire meme is for nvidia

>>56229893
Same as the previous generation of Fury GPU's.

>>56229893
Somewhere between a lot and one metric fuckton.
>>56230028
This is a good estimate.

>>56229893
I reckon £420MSRP for the lowest tier Vega (RX490), £550 for the follow up to the Fury X.

>>56230581
no way, it's gtx 1080 tier

>>56230616
As I said. £420MSRP. Actual prices will likely go up to £480, perhaps even £500

>>56229893
It will cost zero because amd is bancrupt

>>56229893
Same MRSP as Fury X.

>>56230638
>>56230581
Your joke money units mean nothing to the rest of the world.

>>56230712
They still value higher than the dollar, even with the Brexit.
So, no.

>>56230638
but £550 for the follow up to the Fury X is way too much. it's how much GTX1080 costs now (the price will definitely drop once Vega comes out) and AMD has always released their top cards for much lower price than the Nvidia equivalents.

>>56230736
I'm calling it now; 490 is aimed at 1080.
Fury X successor is aimed at competing with the Titan X (pascal) or 1080Ti should it release.
If Fury X v2 can knock Titan X (p) to the floor, even a price of £650 would be justified.

>>56230725
>They still value higher than the dollar, even with the Brexit.
Ladies and gentlemen, a person who has no fucking clue what currency value represent. Next youre going to tell me that company X is more valuable than company Y because its shares are listed at a higher price.

>>56229957
Not when Nvidia beats AMD in price, performance, and power consumption all at once.

>>56230866
Of course I fucking wouldn't. Assets, debt, cash on hand, land, influence on the market, IP's and active workforce are just some of the extra things to consider when valuing a company.
But the pound is STILL higher in value to the dollar.
You can't deny it, bitch.
There's a lot more dollars going around, and that only hurts you. It may be the universal trading platform, but once that swings over to the Yen, you'll be fucked.

There's more total asset value in the dollar, but a single dollar to a single pound, the pound is worth more. Deal with it.

Too much, thanks to HBM 2.

>>56230783
>I'm calling it now; 490 is aimed at 1080.
>Fury X successor is aimed at competing with the Titan X (pascal) or 1080Ti should it release.

The GP102 is almost exactly 1.5 GP104s.
If Vega is going to compete with both GTX 1080 and Titan Xp, it will have to be two distinct chips.

>>56231103
Which it would be.
Fiji wasn't Hawaii.
Fiji and Hawaii may be born from the same design concepts, absolutely, but even if you put Fiji and Hawaii at the same FLOPS, Fiji still beats Hawaii.
Don't forget that Fiji takes some lessons from Tonga, which is newer than Hawaii (albeit less powerful by design) and had some changes under the hood.
If AMD repeats this, Vega will be the 490 and about half a year later, maybe a full year later, we'll see the next step. Could be Navi with that said. Navi may very well be Vega on steroids, what with the whole 'scalability' thing.
THAT would challenge TX:P quite handily, actually.
So with this post, I'd like to change my decision.

Vega won't fight Titan X(p). Navi will. And Navi will be coming a lot sooner than Polaris did compared to Hawaii->Fiji

>>56231202
Sorry, got lost with that last bit. Navi will come a lot sooner than Polaris did after Fiji.

>>56231202
>>56231211
Why the fuck would AMD bother with HBM2 (and its delays) for a GPU small enough to not compete with GP102?

even a 2x HBM2 GPU would slightly edge out GP102's bandwidth, so it's just a matter of how much die space you'd want to throw at it. (GP102 = 470mm2)

A 1x HBM2 GPU would only match the RX 480's and GTX 1070's bandwidth while costing a shit ton more to manufacture.

>>56230933
Fiat currency value doesn't matter much, because the central bank may set it whatever.

In fact having a lower currency cost is better, because you get more for exports. Oh wait, you country doesn't export anything except paki terrorists. Nvm.

>>56230930
So, Still nVidia then!

>>56230736
Who cares? I'm getting the best looking card for my build and currently it's EVGA FTW.

>2016
>not picking you hardware by color

>>56231355
Different anon. Nice digits.

This is AMD we're talking about. Who knows why they do anything anymore with the majority of their products being shit

>>56231612
> newfag detected

AMD has been rocky as fuck for several years, but over a longer historical timeframe they've been in the lead maybe a third of the time at least.

Poolaris was a slightly disappointing low-mainstream part, but Zen seems quite promising, and Vega could actually be efficiency competitive with Nvidia again if HBM2 actually delivers.

File: 1469346816186.gif (3MB, 219x238px) Image search: [Google]

3MB, 219x238px

>>56229893
>amd

>>56230783
And I'm going to screenshot this post so I can make thread called "AMD cucks" once 490 turns out to be a 1070 with terrible drivers.

>>56231725
>It matters that my company used to be great

Every football fan ever.

>>56231976
No problem. And I'll screenshot yours so when I'm right, I can post it and you'll be ashamed of yourself - then you'll try to commit sudoku, before failing and shaming your ancestors.
Then you'll commit sepukku.

File: 1471846123551.png (194KB, 349x365px) Image search: [Google]

194KB, 349x365px

>>56231786
>Nvidia

>>56231612
>Majority of their products being shit
Hows that 3.5gb of RAM treating you, Nvidia shill squad? You know most games nowadays say "PLEASE DONT USE THIS OPTION IF YOU HAVE LESS THAN 4GB OF RAM" if you turn it up to Ultra.

>>56232079
I didnt point any specific products out did I fucknugget? The 390 is bretty good for what it is, also the nano if its price is low enough. The 380 really did well againts the 960 and the 970 was obviously gimped as you so pointed out. The 950 stomped the 370 (my card sadly) into the dust but it may change with further optimization and time. The fury and fury x are shit for their price and an overall disappointment, just barely performing at reference 980ti levels.

AMD cpu's well, the only good chip is the 6300. Maybe the 8350 if its used or if it is being used for a cheap render station.
Cmt is utter shit, meming with half cores that act independently.

Amd is good for budget specifics and should be neglected for anything high end.

>>56230930


https://www.youtube.com/watch?v=sRo-1VFMcbc&t=0m16s

File: 1458448817300.webm (773KB, 480x480px) Image search: [Google]

773KB, 480x480px

>>56232060
>had a hd7970
>try linux mint
>nothing work while my old 8500gt work perfectly

>>56229893
about one loo

>>56232428
Not my fault you couldnt install AMD drivers through the terminal

File: 1436658870921.png (155KB, 455x396px) Image search: [Google]

155KB, 455x396px

>>56232428
>linux
>gaming

File: .jpg (37KB, 426x341px) Image search: [Google]

37KB, 426x341px

Tell you what, I had a crazy idea when I was taking a shit earlier.
Navi, the next step after Vega, is touting 'scalability' on the roadmap as its main feature which makes it stand out. 'Next Gen Memory' seems a little bit wishful, since HBM2 is so new.
Then, as I sat there, taking a shit, it hit me.

What if Navi was a 450mm2 die... but wait for it... not just one 450mm2, but two dies on a single interposer. 4/8GB HBM2 each, 8/16GB HBM2 between them. Being on the same interposer, they could potentially work as the same GPU. No MGPU nonsense, just working in parallel, as shaders do anyway.
A (450mm2 x2) GPU.

Scalability.
Single interposer, single GPU scalability.
It would fucking DESTROY Titan X(p)

>Nvidias face when

>>56232564
Pretty certain that is the plan. It completely negates issues with large die yields and makes high end cards MUCH more manufacturable.

>>56232544
I did it...
>>56232549
>gpu are only for gayming

>>56229893
'bout three fiddy

File: ohgod.jpg (2KB, 125x91px) Image search: [Google]

2KB, 125x91px

>>56232630
The more I think about it, the more I see it.
HBM2 memory is the critical requirement.
Without memory that has both extremely high bandwidth and extremely high base speed (which HBM2 should provide), the solution could only ever be akin to dual GPU cards. But with an interposer connecting the GPU's and memory, with enough bandwidth and speed of the actual connections themselves, their parallelism will work in favour of the overall outcome.
Even choosing to make smaller dies (250 - 300mm2) would effectively bring down power usage while maintaining high output performance.
>This is the future we have become

>>56229893
They claim around $500

If thats true and it undervolts like poplaris nvidia will be dropping prices in panic

>>56230930
Glad summer is over.

>>56231372
Yolks on you mate. The remaining industry Brittan has cannot keep up with demand
But the government for the last 20 years has been utter garbage and destroyed infrastructure ontop of promoting customer service jobs over actual skills like sciences.

>>56232905
>over actual skills like sciences.
And literacy, clearly. Your English is terrible.

>>56232960
define English.

>>56232905
>Brittan
>Garbage
>Yolks
I'll bet you're French or Dutch. Probably just jealous of our glorious independence so you're trying to diminish it by shitposting.

>>56232564
>>56232806
AMD has been pretty open in the past about their plans for "die stacking."
Die stacking isn't 3D stacking, or 2.5D stacking when it comes to packaging. Its taking a single ASIC and breaking it down into separate components to be fabbed on a process specifically tailored to each of them. They could have a single die that is the front end, then two, four, eight, any number of dies loaded with CUs, then have several more dies for ROPs.

Cost for production absolutely plummets through the floor. The only limitation is complexity in packaging, which really isn't an issue.

>>56233225
> jealous of our glorious independence

Good for you, guys. Maybe you will grow up to be a real relevant country, like you once were.