WHat do you think the full gpu lineup will look like at the end of summer

[Alatar]

perf - card - price

100% - GTX Titan X 12GB - $999
95% - GTX 980Ti 6GB - $650
90% - Radeon Fury (something watercooled reference Fiji) 4GB - $650
83% - Fiji Pro (dunno about the name) 4GB - $549
75% - GTX 980 4GB - $499
70% - Radeon R9 390X 8GB - $399
65% - GTX 970 4GB - $329
65% - Radeon R9 390 8GB - $299
50% - Radeon R9 380X 4GB (full tonga) - $219
45% - Radeon R9 380 2GB - $179
45% - GTX 960 2GB/4GB - $179
...

 

[maddie]

8GB isnt coming. Even AMD said so.

You are into Q2/Q3 2016 before you see any 8GB or higher cards. And it will be based on HBM2.

Proof? Where is the statement by AMD?

You often make statements that are wrong.

 

[ShintaiDK]

Proof? Where is the statement by AMD?

http://arstechnica.co.uk/informatio...hbm-why-amds-high-bandwidth-memory-matters/2/

AMDs CTO confirmed 4GB limit. This is also why AMD is so busy trying to upsell 4GB via PR nonsense.

Not that its any surprise for anyone that isnt entirely lost in the hype. Just like the rejection of the almost complete rebrand top to bottom. Both Hynix and AMD was way too ambitious to what could be delivered in time with HBM. And its quite some time since Hynix downgraded the ambitions.

As said before, HBM2 is where the fun starts. HBM1 is more like a...concept.

 

[maddie]

http://arstechnica.co.uk/informatio...hbm-why-amds-high-bandwidth-memory-matters/2/

AMDs CTO confirmed 4GB limit. This is also why AMD is so busy trying to upsell 4GB via PR nonsense.

Not that its any surprise for anyone that isnt entirely lost in the hype. Just like the rejection of the almost complete rebrand top to bottom. Both Hynix and AMD was way too ambitious to what could be delivered in time with HBM. And its quite some time since Hynix downgraded the ambitions.

As said before, HBM2 is where the fun starts. HBM1 is more like a...concept.

This is not proof. You are wrong. He never confirmed 4 Gb. In fact he said that you can have any amount of stacks with the only limit being 1 GB/ stack.

You're not limited in this world to any number of stacks, but from a capacity point of view, this generation-one HBM, each DRAM is a two-gigabit DRAM, so yeah, if you have four stacks you're limited to four gigabytes

 

[ShintaiDK]

Ok despite of what AMD is saying you just keep believing in 8GB HBM1 cards with 8192bit MC then, good luck with that

"You're not limited in this world to any number of stacks, but from a capacity point of view, this generation-one HBM, each DRAM is a two-gigabit DRAM, so yeah, if you have four stacks you're limited to four gigabytes. You could build things with more stacks, you could build things with less stacks. Capacity of the frame buffer is just one of our concerns. There are many things you can do to utilise that capacity better. So if you have four stacks you're limited to four [gigabytes], but we don't really view that as a performance limitation from an AMD perspective."

"If you actually look at frame buffers and how efficient they are and how efficient the drivers are at managing capacities across the resolutions, you'll find that there's a lot that can be done. We do not see 4GB as a limitation that would cause performance bottlenecks. We just need to do a better job managing the capacities. We were getting free capacity, because with [GDDR5] in order to get more bandwidth we needed to make the memory system wider, so the capacities were increasing. As engineers, we always focus on where the bottleneck is. If you're getting capacity, you don't put as much effort into better utilising that capacity. 4GB is more than sufficient. We've had to go do a little bit of investment in order to better utilise the frame buffer, but we're not really seeing a frame buffer capacity [problem]. You'll be blown away by how much [capacity] is wasted."

Its not for fun he makes excuses about 4GB is plenty later on.

 

[maddie]

Ok despite of what AMD is saying you just keep believing in 8GB HBM1 cards with 8192bit MC then, good luck with that




Its not for fun he makes excuses about 4GB is plenty later on.

Not an 8192 bit memory controller, but (2) 4096 bit memory controllers. They have to have at least (1) 4096 bit controller, so we can be sure that has been done. Double it, space for I/O is not a problem with microbumps. This is not like a PCB.

A bet. I say 8GB HBM, you say 4GB.

I win, you admit I was right.

You win, I admit you were right.

 

[Subyman]

I don't think that will happen as it hasn't been that way historically. While generally the fastest single-GPU card has been Nvidia, at lower tiers the AMD card is almost always faster at similar prices. People shop by price not by product stack and at the same price AMD has almost always been faster.

By uneducated guess:

Titan X ($1000) ~= Fiji WCE ($800-850)
980Ti ($650-700) ~= Fiji Pro ($550-600)
980 ($450-500) ~= Respun 290X (GCN 1.2?) ($400)
970 ($300-329) ~= Respun 290 ($280-300)
960Ti ($250) ~= optimized Full Tonga chip ($200-225)
960 ($180-200) ~= optimized 285 ($160-180)

Too many rumors flying around to know if everything below Fiji will be respun at GF, but that would be the best outcome IMO to help lower power consumption and add more performance

I think your line up looks spot on. :thumbsup:

 

[maddie]

Im not going to say AMD will disappoint you, but amd will disappoint you.

If this is meant for me, I won't be disappointed either way. Out of my price range. Lower tier for me. Just trying to show that we might be making unjustified assumptions.

 

[yacoub]

perf - card - price

100% - GTX Titan X 12GB - $999
95% - GTX 980Ti 6GB - $650
90% - Radeon Fury (something watercooled reference Fiji) 4GB - $650
83% - Fiji Pro (dunno about the name) 4GB - $549
75% - GTX 980 4GB - $499
70% - Radeon R9 390X 8GB - $399
65% - GTX 970 4GB - $329
65% - Radeon R9 390 8GB - $299
50% - Radeon R9 380X 4GB (full tonga) - $219
45% - Radeon R9 380 2GB - $179
45% - GTX 960 2GB/4GB - $179
...

Probably close to right on the order and prices, not sure about the performance percentages. AMD needs 980Ti competitor at same or better price as 980Ti. Also needs that 2nd Fiji model that comes in around $500 and bests the rest of the NVidia lineup below 980Ti. That will be their main seller for the enthusiast market and if it performs well, would be what I'd consider instead of a $650 980Ti. Saving $100-$150 for ~90% of a 980TI's performance would be worth it to me. Else, I'd just consider the 980Ti.

 

[Stormflux]

Given the recent 980ti launch, considering all the rumours and speculation and finally paying heed to the all important marketshare and mind share; this is where I think AMD has to price their cards at. On top of a continuous marketing push to educate and restablish themselves as a brand. 290x two years ago? Who's AMD?

Whether these AMD skus (8GB HBM vs. 4GB HBM) exist or not is not the point.

[Luxury] $999 Titan-X 12GB (103%)
[Luxury] $849 8GB HBM Fury XT OC WaterCoolingEdition (130%)

[Ultra-End] $749 8GB HBM Fury XT (120%)

[High-End] $699 8GB HBM Fury Pro (110%)
[High-End] $649 980Ti 6GB (100%) |Base Line Performance|

[Mid-Range] $549 4GB HBM Fury Pro (90%)
[Mid-Range] $499 980 4GB (80%)

[Mid-Low-Range] $399 390X 8GB GDDR5 (70%)
[Mid-Low-Range] $319 970 3.5GB (60%)

[Low-Range] $299 380X 4GB GDDR5 (50%)
[Low-Range] $269 960 2GB GDDR5 (40%)


Now if the 8GB HBM SKUs do not exist, keep the performance but have 4GB HBM, scrap the redundant Mid-Range card and hope AMD can market them well enough. TDP is an important factor and if AMD doesn't have it under control this round... idk. Either way, going to be an exciting month.