Tuesday, September 20th 2016

AMD Vega 10, Vega 20, and Vega 11 GPUs Detailed
AMD CTO, speaking at an investors event organized by Deutsche Bank, recently announced that the company's next-generation "Vega" GPUs, its first high-end parts in close to two years, will be launched in the first half of 2017. AMD is said to have made significant performance/Watt refinements with Vega, over its current "Polaris" architecture. VideoCardz posted probable specs of three parts based on the architecture.
AMD will begin the "Vega" architecture lineup with the Vega 10, an upper-performance segment part designed to disrupt NVIDIA's high-end lineup, with a performance positioning somewhere between the GP104 and GP102. This chip is expected to be endowed with 4,096 stream processors, with up to 24 TFLOP/s 16-bit (half-precision) floating point performance. It will feature 8-16 GB of HBM2 memory with up to 512 GB/s memory bandwidth. AMD is looking at typical board power (TBP) ratings around 225W.Next up, is "Vega 20." This is one part we've never heard of today, and it's likely scheduled for much later. "Vega 20" is a die-shrink of Vega 10 to the 7 nm GF9 process being developed by GlobalFoundries. It will feature 4,096 stream processors, too, but likely at higher clocks, up to 32 GB of HBM2 memory running full-cylinders at 1 TB/s, PCI-Express gen 4.0 bus support, and a typical board power of 150W.
The "Vega 11" part is a mid-range chip designed to replace "Polaris 10" from the product-stack, and offer slightly higher performance at vastly better performance/Watt. AMD is expecting to roll out the "Navi" architecture some time in 2019, and so AMD will hold out for the next two years with "Vega." There's even talk of a dual-GPU "Vega" product featuring a pair of Vega 10 ASICs.
Source:
VideoCardz
AMD will begin the "Vega" architecture lineup with the Vega 10, an upper-performance segment part designed to disrupt NVIDIA's high-end lineup, with a performance positioning somewhere between the GP104 and GP102. This chip is expected to be endowed with 4,096 stream processors, with up to 24 TFLOP/s 16-bit (half-precision) floating point performance. It will feature 8-16 GB of HBM2 memory with up to 512 GB/s memory bandwidth. AMD is looking at typical board power (TBP) ratings around 225W.Next up, is "Vega 20." This is one part we've never heard of today, and it's likely scheduled for much later. "Vega 20" is a die-shrink of Vega 10 to the 7 nm GF9 process being developed by GlobalFoundries. It will feature 4,096 stream processors, too, but likely at higher clocks, up to 32 GB of HBM2 memory running full-cylinders at 1 TB/s, PCI-Express gen 4.0 bus support, and a typical board power of 150W.
The "Vega 11" part is a mid-range chip designed to replace "Polaris 10" from the product-stack, and offer slightly higher performance at vastly better performance/Watt. AMD is expecting to roll out the "Navi" architecture some time in 2019, and so AMD will hold out for the next two years with "Vega." There's even talk of a dual-GPU "Vega" product featuring a pair of Vega 10 ASICs.
194 Comments on AMD Vega 10, Vega 20, and Vega 11 GPUs Detailed
And overclock?! ROFL! My Fury is at 1135 MHz - I get 10% more performance. Pascal overclocks usually yield a 7% boost - WOW!
Edit: just noticed a big mistake. The 1070 is actually 2 steps below a fury x as a fury x was meant to compete with the ti model. So things are looking pretty good there.
The only real exception is with Polaris... GPU overclocks aren't worth much alone, but memory overclocks combined with higher GPU clocks can be quite good. Still, even with that, you're only talking about 15% more performance over reference... and I consider a 30% improvement to be the smallest worthwhile improvement to consider spending money on... unless you're just at the cusp of stable frame rates or just barely outside of your monitor's FreeSync range, when even 5~10% more performance can make the difference between stutters and silky-smooth experiences.
For me, though, I usually underclock my video cards. My RX 480 is currently running 640Mhz on the core and 1Ghz on the RAM. I can play most of the simple games and do all of the normal tasks that I want - while consuming only ~35W under load (above idle usage of ~20W).
I then have three other performance presets (using Afterburner).
900/1500 - Somewhat more demanding games (Crysis [Warhead] mostly)
1000/2000 - Demanding games (BF4, Hitman [Absolution], Civ V)
1288/2150 - Seriously Demanding Games (BF1 the only one so far)
Mind you, I have to keep frame rates at or below 146Hz and above 30Hz for FreeSync, so I tune things appropriately. All of my profile have significant under-volting applied and have had a few hours, at least, of games played with them. This is expected, Pascal and Polaris both represent a double-generation jump (well, Polaris anyway... Pascal is basically just a shrunken Maxwell with clock speed improvements enough for two generations of performance increase).
The only effort this generation that should really be applauded, IMHO, is AMD's. GCN4 is ~15% faster per clock, more efficient, brings quite a bit of meaningful new tech, and still clocks 15%+ higher... the fact that it was only used on a mid-range GPU is irrelevant from a technology perspective. Pascal is just the same old stuff with a few tweaks and clock speed improvements... boring stuff, really, albeit the feat is still impressive... it's basically just teams looking for critical paths and optimizing them away.
Jokes aside, AMD does need to make some more serious efforts towards improving architectural efficiency. It mostly comes down to under-utilization, IMHO. If AMD had the same utilization rates as nVidia, they would be pretty close to just as efficient... instead AMD crams 35%+ more computing power to break even.
Per TFLOP, AMD isn't less efficient... but per FPS, they certainly are.
Consider though that by the time Vega launches almost all AAA games will be DX12/Vulkan, and Vega will have at least 12 TFLOP's while the Titan has 11. Vega will also have more bandwidth too...
Vulkan/DX 12 are allowing AMD GPUs to fill in the gaps left by their scheduler windows... gaps which really shouldn't exist as much as they do in the first place. AMD needs a new ABI, scheduler, and driver in order to get rid of more of those gaps using DX11... but they could, conceivably, then see a ~25% boost in performance... without providing more hardware processing power...
Vega will need to come out with a low price, no doubt. This is one of the major reason I think AMD will use a 2048-bit HBM2 bus... two stacks are cheaper than four.. and even if the costs are not so much better, you still have a smaller die/interposer or both to help.
Volta will trash Vega 10, no doubt. The only way this wouldn't happen is if Vega is using the long-rumored new ABI... then we could see Vega 10 being
as much as twice as fast as the Fury X. I have serious doubts about this, though... the same team that created Polaris created Vega just six months later, so it's most certainly a direct a descendant of GCN 4... and there's some indication in my own circle that AMD has abandoned the new ABI altogether as the software is catching up and they anticipate DX11 performance to become irrelevant - as well as their utilization issues... all with no work on their part.
Timing, however, is a major issue for AMD. Unless Vega 10 is much better than anticipated, Volta will wipe the floor with Vega... and AMD's own timeline doesn't have Navi until 2018... giving nVidia the top tiers for nearly two years.
I've got serious doubts nVidia will manage to keep it's speed advantage over AMD (their cards run @ a much higher speed then AMD's) and, if so, that also means efficiency will suffer: time will tell. If they do manage to keep the speed WHILE incorporating async, then i agree with you.
is like a never ending soap tv show when every week a new rumor appear...
"The future has a way of arriving unannounced." George Will
I did see this tidbit from WCCFtech that got me thinking WTF? It's hype to the max when a website is looking at a card to be released in 3 years time and then comparing it against a card out now (not AMD's fault). This is part of an ongoing problem when AMD start to release info, the rest of the web starts to pump it up, unfortunately.
Anyhoo, I've said enough here. Suffice to say let's not diss the 980ti as it still is a very powerful card when it's let loose, especially so if you're still on DX11 and lots of us still are. And in the UK, the main suppliers don't have many (if any) Fury X on sale and they are still £500 (not $300). Conversely, the 980ti's are under £350.
Funny ol' world.
ps- can people stop double and triple posting? Use multi-quote.
EDIT: I had to come back
Given this was a year ago and I checked the only 2 games left in TPU suite (Witcher 3 and BF4) both give gains to the 980ti a year on (as does Fury X). To understand at least my point and that of enthusiasts with overclocked 980ti's. My card runs pretty much where the OC Lightning is. You need to understand that performance to see why people defend their beloved 980ti's so much. A stock 980ti is meaningless when looked at alone. That was why I never bought a Fury X - it released at an awful OC scenario. Even with voltage it pales in comparison to a 980ti.
So for some performance perspective, this is why you shouldn't write of a 980ti. In that example below it's >30% above stock. That's where my card games.