Cloudfire777
Golden Member
- Mar 24, 2013
- 1,787
- 95
- 91
No, that's MCDRAM. The difference is that HBM will be built on a silicon interposer, while the memory chips on your image are on an organic substrate. The Intel solution will likely be cheaper and have better yields, but it can't quite match the kind of bandwidth or latencies that HBM can provide.
HBM will be stacked even closer.
I just realized something. if 380x is as fast or faster than 980, that means russian's assertion that 980 is just a mid range is 100% true!
that means nvidia fans has been buying a mid range card like the 460/660ti for 550$, woah. I am surprise nv fans aren't mad about it
damn, this is the longest 3 weeks ever! I want more info!
I just realized something. if 380x is as fast or faster than 980, that means russian's assertion that 980 is just a mid range is 100% true!
that means nvidia fans has been buying a mid range card like the 460/660ti for 550$, woah. I am surprise nv fans aren't mad about it
damn, this is the longest 3 weeks ever! I want more info!
Sorry, let me put my angry face on and throw a fit on a forum!
Subyman, your explanation makes sense, but who should boozzer really be mad at? Nvidia for trying to maximize profits, or AMD for allowing them to do it due to lack of competition?
Nice postIt doesn't require a mid-range card to beat a 980 for this to be the truth.
It is all based on how Nvidia used to operate, and what we see today.
The x80 cards used to be "the big chip", the full-fat beastly dies. Up until the 680, the top-end cards were all over 500mm, then the 680 was under 300mm. This was easily achievable due to a lack of pressure by AMD at the time.
The 780 returned to form with 561mm^2 chip of various configurations, but they demonstrated that was but a quick reprise, as the 980 is just under 400mm.
Various people deny this is how you define mid-range or top-end, but ultimately, Nvidia's methodology is very rooted in history, and their top-end releases were always mammoth chips. They've been cutting back and saving the big chips for other things because it can make them far more money selling the big chips to large compute customers and professional GPU users. When the competition lets them get away with it, what would you do? They can make far more money with those big chips by not selling them at consumer profit margins, or waiting until the process has matured significantly before they start selling consumer parts.
This is sort of where the Titan chips fit in: they milk the consumer market and smaller compute users for higher profit margins, and then they finally release a true consumer numbered-series GPU at lower prices.
These prices are still higher than big chips had ever been, when you could get a 280/285, 480, and 580 for around $500, and had the top-end performance short of dual-GPU solutions. Now top-end single chip performance, if Nvidia has their way, is over $700... $1000 if you count the Titan releases.
Here's to hoping the AMD price structure of their 300 series helps reign in Nvidia's pricing. This is what the consumer market needs, strong competition to prevent price gouging for the sake of shareholder-demanded revenue increases.
The 380X is full Tonga. It'll be dramatically slower than the 980.
That doesn't change the fact that the 980 is midrange, though.
Nice post
Just wanted to chime in:
How far do you think Nvidia may even take this if given the chance? I mean, if the gap between Nvidia and AMD just keeps growing, would they even try placing GP106 as the x80 of its generation if it's competitive with AMD's best at the time and edges out GM200? I don't want to see the market when such a thing is even a concern, but it's possible isn't it?
Regarding GM204/the 980 being mid-range, well by literal definition even:
GM206 - 1024 shaders
GM204 - 2048 shaders
GM200 - 3072 shaders
Seems pretty middle of the range to me, smack dab in between GM206 and GM200.
In for one 390X 8GB, if that's true!
People, remember that 360/370/380 are OEM chips. Keep your eyes open for 370X, 380X and 390/390X.
Here are the latest price ive heard from Korea (rumor per moderators request):
R9 370X - $229
R9 380X - $349
R9 390 - $429
R9 390X 4GB - $549
R9 390X 8GB -$599
You may notice that R9 380X cost $100 more than R9 290X.
Lets just say GTX 980 will get some serious competition.
Its a reason why AMD only announced and listed the non-X chips
I am speaking of what used to be mid-range is now sold as high end, and what used to be high end is now sold as ultra luxury. It is semantics, but nvidia managed to move what used to be a $500-600 card into an exclusive $1000 card and then replace the large die $500-600 price bracket with a small die card that used to be in the $200-300 range. They managed to shift their technology upwards in price. This was mainly done because it could be. The small die cards competed well with AMD's top end cards, which left room for nvidia to shift into ultra luxury category.
As the generation proceeds, Nvidia simply steps a version of the large die GPU down into lower price brackets and slots the fastest small die in as a mid-range card. We saw this with the 680 -> 770 and the Titan -> 780.
Its relative to their previous strategy. The real definition of mid-range/high-end is simply pricing structure and number of units moved.
So the only >4GB card will be the $599 390x?
That looks like lit might be G1/4 threaded! :awe:
Day one purchase if it is and the thing is a beast.
That's how they increased their margins to over 50%. It's a short term solution though, IMO. It's not like discrete GPU's are always going to be around.
People, remember that 360/370/380 are OEM chips. Keep your eyes open for 370X, 380X and 390/390X.
Here are the latest price ive heard from Korea (rumor per moderators request):
R9 370X - $229
R9 380X - $349
R9 390 - $429
R9 390X 4GB - $549
R9 390X 8GB -$599
You may notice that R9 380X cost $100 more than R9 290X.
Lets just say GTX 980 will get some serious competition.
Its a reason why AMD only announced and listed the non-X chips. Announcements should come eventually![]()
Yeah 8GB FTW. Remember that AMD is facing 6GB 980Ti and 12GB TitanX. That probably resulted in a little lower price than whats ideal for AMD for 8GB 390X.Seems strange to me. Who is going to buy 390 and 390x 4gb if a 390x 8gb is just 50 bucks more?
Heck everyone and his brother is going for the sweet watercooled 8gb version. Man that gfx is just like some well looking chickits the s6 edge returned to gfx land. Lol.
Seems strange to me. Who is going to buy 390 and 390x 4gb if a 390x 8gb is just 50 bucks more?
Heck everyone and his brother is going for the sweet watercooled 8gb version. Man that gfx is just like some well looking chickits the s6 edge returned to gfx land. Lol.
Yeah 8GB FTW. Remember that AMD is facing 6GB 980Ti and 12GB TitanX. That probably resulted in a little lower price than whats ideal for AMD for 8GB 390X.
I didnt say the $599 was the WCE version though. My contact didnt either. I dont know the price difference between the air and wce model![]()
12gb titan x is useless vs 8gb hbm. If nv had that card in their hands they would charge 999usd and market it for what it probably is - the fastest gpu. Who cares if 980ti is just 15% slower. Amd need to make a premium of the situtation and boost their brand value. Dont let nv hype the 12gb. Stay firm - and the 12gb will fall flat.
![]()
Here is the picture without wccftechs logo
![]()
Of course 8GB HBM is clearly superior to 12GB GDDR5. 12GB is major overkill for gamers.
But Im talking abou capacity. 4GB anything today would suffer in reviews for games that require 4GB+. And we are talking a card for 4K gaming. AMD cover all bases and make sure reviews dont put their new card in bad light.
$50+ for 4GB HBM more and $599 is excellent value for us gamers.
