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.
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
For example, it doesn't take into account, at all, the impact of memory bandwidth. It doesn't take into account things such as primitive discard, improved scheduling, and so on... It's just an accumulation of the processing power in the shaders.
12 TFLOPs tells us only two two things that are likely: GCN4 SPs are still likely in use. 1.266GHz is the target clock-speed:
+15% - GCN SPs vs Fiji SPs
+20% - Higher clock speed than Fury X (1.26GHz vs 1.05GHz)
That takes Fury X's 8.6T to right about 12T. A few minor improvements (and some rounding in the rumor) will cover the minor gap.
Given how constrained performance on Polaris 10 is by memory bandwidth, we should expect an additional gain in FPS from that - 15% is fair game (you can nearly get that just by overclocking RX 480's memory... gains are effectively linear). Vega includes the new primitive discard and other benefits seen in Polaris 10 as well, so this will push effective performance compared to Fury X even more.
If you add 60% more performance to Fury X, where do you land? Bingo. Still, from 40% to 60% over Fury X is a pretty broad range - and only the top of this will catch up to the top Pascal GPU.
Hence the 475 and 485 will be 4GB HBM cards, and the RX Furies will be 8GB HBM cards. The 465 may even be a 2GB HBM card.
I have determined that Nvdia will maintain the performance crown.....at a hefty price and Amd will design something ridiculously advanced that won't be fully
utilized for years to come but be at a great price.
/thread.
I definitely wish Vega was here as an alternative to the Titan, but I would rather AMD wait and release a monster card with HBM2 than rush out a half-baked GDDR5X card in the Ultra-High end.
Also, When Fury X was released it was meant to be awesome but it was a bit of a disappointment. When you say it matches a 980ti and beats it (game environment dependent, massive 20% OC of 980ti not withstanding) it actually should match it - they were price point stable mates. They were last years top gen. A Fury X was meant to take on Nvidia. Now it is, a year later. The 1070 replaces the 980 ti and the 1080 takes the 2nd top crown in the vast majority of titles. 1080 even beats Fury X in AotS, Hitman, ROTR and Warhammer (all DX12). Remember, generationally the 1080 replaces the 980 (which Fury X pisses on).
No arguing Fury X is a great card but it's hard to listen to people saying how poorly NVidia cards age when my 980ti is still trouncing framerates. I'm on DX11 of course so I don't get to see the fuss but so far, DX12 hasn't delivered better graphics.
So, to recap, when you sayIt came out after it and was 'meant' to beat it back then. It's only now performing as it was meant to back in 2015 (and mainly in specific, not all, DX12 and one Vulkan title) but it's still weaker than Nvidia's 2nd top card (1080).
And to clarify - I want Vega to beat the Titan X.
The 480 is fantastic, but just not quite strong enough for me. Thus I ended up buying a Fury Nitro and overclocking it to 1135/518. I wish I had more, but for $300 I can't complain.
But that's also mean another thing: people are going to wait to get their mid-rang hbm gpu, meaning that polaris gpu will still get a lot of stock when the refresh launch. I have a bad feeling about the pricing of the refresh.
I'm already looking forward to Volta.
www.techpowerup.com/forums/threads/amd-vega-10-vega-20-and-vega-11-gpus-detailed.226012/page-5#post-3526293
On average the Fury X is beating it in every resolution but 1080p (And even then it loses by 4%). Furthermore the best indicators of this fall and next year's game performance are DOOM and Deus EX where the Fury X does in fact TROUNCE the 980 Ti by 20%.
I will agree it would have been nice if the Fury X performed this well last year, but at the end of the day who really cares?! Both the 980 Ti and the Fury X destroyed all games last year, and it never mattered the 980 Ti was like 5% stronger. Yes it overclocked WAY better at launch, but then again the Fury series overclocked perfectly fine once voltages were unlocked (Although of course Maxwell overclocks better on average still).
Yeah, I see the Fury somewhere on that list. Somewhere very far away from the 1070 stock, let alone OC. I guess you'll retort that all the recent AMD Gaming Evolved (totally not GamWorks guyz) games have been doing pretty well on the Fury series, too bad that comes with crazy frame spikes on AMD cards. Curious that only AMD sponsored games have had such differing performance, when AotS, TW: Warhammer, Tomb Raider, etc have been doing just fine on Nvidia.
techreport.com/review/30639/examining-early-directx-12-performance-in-deus-ex-mankind-divided/3
Who knows, maybe vendor-neutral DirectX 12 and Vulkan games will continue to help the Fury reach up near 1070 performance. All I know is that the vast majority of games being played today perform 20% better on the 1070 and don't suck up 300 watts while doing it.
All future poofing means for me is that the product will continue to provide the same experience, or even a better experience for 2-3 years. People who bought an i7 5 years ago continue to get 60-100 FPS in the latest games while the old i3's need to be replaced (SB i3's can't really do 60 FPS anymore). Of course nothing is completely future proof, but I expect my products to maintain their relative value for 3 years, and frankly they should never lose performance to their launch competition.
We can talk about AoTS and Warhammer all you want lol:
www.guru3d.com/articles-pages/total-war-warhammer-directx-12-pc-graphics-performance-benchmark-review,6.html
The way it's meant to be played: At a lower framerate!
EDIT: Actually, Techpowerup themselves just did that bench again and the stock 1070 beats the Fury X at all resolutions. Looks like your argument is crumbling away. Yikes, check out those No Man's Sky benches too.
www.techpowerup.com/reviews/MSI/GTX_1070_Gaming_Z/23.html
They will also tell you that their 4GB fury X is more future proof than a 8GB 1070 "because shaders/vulcan/HBM".
In this case, though, I really do hope AMD isnt too far below that 1080... I have the 1080 and it's 100% artificially crippled... nvidia already holding back and gouging.10-15% slower but yes - that game.