It means than Vega architecture will be used in Zen APU's.http://cdn.wccftech.com/wp-content/uploads/2016/03/AMD-Vega-10-GPU-Leak.jpg
It says Whole new soc v15 arch.What does this mean?
In server market. Small Vega can be RX 490, big Vega - RX Fury.why would Vega 11 be targeted at the same price range and as a direct replacement for 480? This makes no sense whatsoever.
I don' think so as engineer id says only about one chip greenland which has 4096SP.It means than Vega architecture will be used in Zen APU's.
It means they can't get there without hbm ?In server market. Small Vega can be RX 490, big Vega - RX Fury.
http://videocardz.com/62250/amd-vega10-and-vega11-gpus-spotted-in-opencl-driverI don' think so as engineer id says only about one chip greenland which has 4096SP.
What I meant was that in Server/Professional market replacing Polaris with Vega as mainstream is good idea, because you have high enough margins, to justify increased production costs.It means they can't get there without hbm ?
In this industry nothing happens without a reason. Ask yourselves why NVidia would refresh Pascal GPUs, when they dont have to. Do they, really? Or maybe AMD will also refresh their GPU lineup.
Well obviously AMD is giving some competition, but even without AMD they'd do it anyway as they are *new* sells.
Fury x did it with same TDP as 290x then why not this one?They didn't refresh GTX950, 960, 970, 980 though. Refreshing this time makes sense:
- They already have 2048 CC mobile 1070.
- There will be faster GDDR5X
- Almost all 1060/1070/1080 cards hit 2000mhz. With a more mature process, they could up the GPU clocks 10%.
Besides, after pricing the reference 1070 $100 higher than 970 and 1080 $150 higher than 980, they can also lower prices. If they release a 10% faster 2070 with 2048 CC & G5X and drop the price to $349, that would result in card with ~90% of 1080's performance for 1/2 1080's launch MSRP 1080. For 2080, they could drop the price to $499-$549, while adding 12Gbps G5X and upping GPU clocks. That would give them room to drop 2080Ti for $749-799, and release a fully unlocked Titan XP Black for $1200.
Looking at rumoured specs of Vega 11, it may not even reach the current GTX1070. Right now AMD may be planning their entire launch around the idea that most 1080's cost $630-700, and some are expecting $849-899 1080Ti. But if NV releases a $349 10% faster 2070, and drops the price of a 10% faster 2080 to $549, AMD's entire strategy could be mitigated.
---
It's a bit disappointing that Volta 10's TDP is only 230W. Unless AMD improves performance/watt dramatically from RX 480, I don't see how a 230W Volta 10 can compete with AIB GP102 1080Ti. The constant cries about power usage means we likely won't get a 280-300W AIO-CLC Vega 10 monster.
I mean they can't get up to nv efficiency without hbm.http://videocardz.com/62250/amd-vega10-and-vega11-gpus-spotted-in-opencl-driver
Raven Ridge APU is codename for Zen APU. And it is the same GPU family as Greenland and Vega architectures.
What I meant was that in Server/Professional market replacing Polaris with Vega as mainstream is good idea, because you have high enough margins, to justify increased production costs.
I don't think we will see Vega architecture without HBM2 chips.
Also, Maybe WCCFTech has a point in saying that small Vega can have 130W TDP. If all of their information, and Videocardz, and FUDZilla information comes from the same source it can mean that it comes from professional branch, not consumer.
Apple Mac Pro GPUs have 129W TDP
3072 GCN core, with 2 stacks of HBM2 and 8 GB of memory Total, 1.25 GHz, and we get pretty good, 7.7 TFLOPs GPU that can account for Radeon Pro DX700, for a bargain of 1000$ addition to base price of the computer .
I agree, but rumored 2048SP GDDR5x 1070 refresh will be more than 10% faster than current 1070.I think 15-20% performace is possible.Mobile version is already 10% faster at same clock and its still using 8Ghz memory.They didn't refresh GTX950, 960, 970, 980 though. Refreshing this time makes sense:
- They already have 2048 CC mobile 1070.
- There will be faster GDDR5X
- Almost all 1060/1070/1080 cards hit 2000mhz. With a more mature process, they could up the GPU clocks 10%.
Besides, after pricing the reference 1070 $100 higher than 970 and 1080 $150 higher than 980, they can also lower prices. If they release a 10% faster 2070 with 2048 CC & G5X and drop the price to $349, that would result in card with ~90% of 1080's performance for 1/2 1080's launch MSRP 1080. For 2080, they could drop the price to $499-$549, while adding 12Gbps G5X and upping GPU clocks. That would give them room to drop 2080Ti for $749-799, and release a fully unlocked Titan XP Black for $1200.
Looking at rumoured specs of Vega 11, it may not even reach the current GTX1070. Right now AMD may be planning their entire launch around the idea that most 1080's cost $630-700, and some are expecting $849-899 1080Ti. But if NV releases a $349 10% faster 2070, and drops the price of a 10% faster 2080 to $549, AMD's entire strategy could be mitigated.
More so, imho, the closer we get to the mid-life of the current generation, the less it makes sense to spend a lot of $$$ on higher end cards. I am stating the obvious but after barely a year, it was possible to buy a $650 980Ti/Fury X for $380. Today, it's possible to find open-box 1070 for $325-340 and new Fury X for $325. If flagship Vega/2080Ti launch in 2017 for $650-750, why buy that knowing Volta 3070 will wipe the floor with them for $450? Outside of small fraction of PC gamers on 144Hz 1440p, 3440x1440 and 4K users who must have the latest and fastest, it's actually the games which dictate if the GPU is worth some asking price. 95%+ of games today can be easily played maxed out on a 1070 at 1080p or even 1440p 60Hz.
By missing on this entire holiday season when some highly anticipated games are coming, AMD then has to count on some major AAA games launching in 1H of 2017 to sell their high-end cards. That is because unlike many NV loyalists, AMD buyers won't buy a flagship AMD just to have the latest and greatest. In a nutshell, unless next gen DX12 games are coming out in waves in 2017, many gamers who waited out on 1070/1080 are likely to just skip this entire generation and wait for Volta, or until more demanding PC games come out. History often shows that by the time next gen demanding PC games come out that warrant buying a $650-700 AMD flagship, that card can easily drop to 1/2 its launch price -- and let's face it, in a year or slightly more it will be when 2018 Volta beats it.
I am very pessimistic about Vega launch timing and expected performance in light of Pascal refreshes and the gargantuan performance gap that exists between a cut-down flagship Titan XP and RX 480/Fury X.
Next generation PC games are nowhere to be found, and the mid-range 1070 OC crushes almost everything already. I am starting to even question that outside of the resolutions I outlined, why not find a great deal on a 1070 and just skip all AMD and NV cards of 2017 and instead just spend it on Volta/Navi? We already know that AMD/NV bi-furcate their generations now so there is no more future-proofing with flagships -- more like overpaying upfront for bragging rights.
1070 OC at 1080p/1440p 60Hz smashes through Forza Horizon 3, Gears of War 4, Civ 6, BF1 (sure you may need to adjust 1-2 settings). It seems this entire generation's issue isn't lack of GPU power but lack of true next gen PC games that would under normal circumanstances make one salivate at the prospects of next gen flagship GPUs. Maybe it's just me, but due to lack of PC exclusives of the past, 2017 GPUs are almost sitting in no-man's land between the perfect stop-gap 1070 OC and Volta (when hopefully we may actually get a wave of true next gen 2018-2019 PC games, not these Xbox One console ports).
---
It's a bit disappointing that Volta 10's TDP is only 230W. Unless AMD improves performance/watt dramatically from RX 480, I don't see how a 230W Volta 10 can compete with AIB GP102 1080Ti. The constant cries about power usage means we likely won't get a 280-300W AIO-CLC Vega 10 monster.
Polaris looks like HD4770, a 14nm LPP pipe cleaner for ZEN and VEGA.
But we had to listen on one lie after the other for over a half a year. Only to more or less end up as something Nvidia already had on 28nm.
.
You have done the same test yourself or anyone else and proved to be a lie ??
Nobody that have tested Polaris 11 came even remotely close to what AMD claimed.
In this industry nothing happens without a reason. Ask yourselves why NVidia would refresh Pascal GPUs, when they dont have to. Do they, really? Or maybe AMD will also refresh their GPU lineup.
Dont try to invade the question, i repeat
Did you or anyone else have done the same test and proved to be a lie ??? 60fps cap in Starwars Beggars canyon ??
They're competiting with themselves. By Spring next year most who would have brought Pascal at its current prices/performances will have done so. So they'll need to bump things up a bit.
For very year for multiple years now NV have yielded ~30% top speed increases year on year. The fairly inevitable 1080ti/derivatives should do that next year. There is a chance that - like with the 97/80 - they'll just slot this stuff in above the current top chips but it seems very plausible that they'll refresh and drop things down a bit price wise at the same time.
The 1080 will instantly lose its 'fastest GPU' premium for instance.
Seeing it before the end of the year doesn't mean it'll launch technically.
And with the same setup and what not? No, because AMD never said what they actually used with good reason
So how they lied then ??
Why dont you show its possible then, since its nowhere near review sites.