AMD VEGA confirmed for 2017 H1

JediFragger · 19 Nov 2016 at 13:01

SiDeards73 said:
So tempted to just buy a card today to tide me over til Vega, been looking to replace my 290 for WoW, PoExile, D3, Division, Elder Scrolls Online and im thinking either a Fury, or the GTR Black Edition 480... also and i hate to say it, the Zotac 1070 looks appealing too, i'd lose Freesync with that option though.

There's a Nano fs in the MM if you have access...

Clock it up and sorted :cool:

DarrenM343 · 19 Nov 2016 at 13:09

nashathedog said:
The top Vega cards need to be a good bit faster than the Fiji cards, like 50% or it won't have the wow factor of a good release. If it's beating a 1080 that'd be great if not it'll still be my upgrade option but it will be a bit meh.

You all have low expectations.
If big Vega does exceed the 1080 by a huge amount it'll be a failure too. By the time Vega comes out the 1080 would have been on the market for close to a year. R&D gets cheaper to hit a performance level with time so has to exceed the 1080. BY then Nvidia might have released more cards or be very close to their next gen.

I'm hoping AMD are using this time to do a lot of R&D and their Vega ends up being close to Nvidia's next gen, not current gen, so that means blowing away the current TX too. If AMD can't do this then it'll be another rather poor showing IMO , even taking their lower R&D budget into consideration.

Griffildur · 19 Nov 2016 at 13:27

DarrenM343 said:
You all have low expectations.
If big Vega does exceed the 1080 by a huge amount it'll be a failure too. By the time Vega comes out the 1080 would have been on the market for close to a year. R&D gets cheaper to hit a performance level with time so has to exceed the 1080. BY then Nvidia might have released more cards or be very close to their next gen.

I'm hoping AMD are using this time to do a lot of R&D and their Vega ends up being close to Nvidia's next gen, not current gen, so that means blowing away the current TX too. If AMD can't do this then it'll be another rather poor showing IMO , even taking their lower R&D budget into consideration.

that'll be quite a jump in performance, you've basically gone the other way by expecting too much. Nothing wrong with it, but still ... quite a jump.

TNA · 19 Nov 2016 at 16:16

DarrenM343 said:
You all have low expectations.
If big Vega does exceed the 1080 by a huge amount it'll be a failure too. By the time Vega comes out the 1080 would have been on the market for close to a year. R&D gets cheaper to hit a performance level with time so has to exceed the 1080. BY then Nvidia might have released more cards or be very close to their next gen.

I'm hoping AMD are using this time to do a lot of R&D and their Vega ends up being close to Nvidia's next gen, not current gen, so that means blowing away the current TX too. If AMD can't do this then it'll be another rather poor showing IMO , even taking their lower R&D budget into consideration.

This is what I have been saying for a long time. The top end Vega needs to be at least matching the Titan PX when it comes out or it will be a failure IMO.

Can anyone explain how new architecture + HBM2 + 14nm + extra time to develop = 1070 performance? Just does not compute.

It was not that long ago that top end AMD (Fury X) and NVIDIA (Titan X) were like not hugely apart in performance. How did AMD suddenly go from that to it will be surprising if they are able to beat a 1070/80?

AMD have had an extra year to develop Vega so it needs to be beating Titan XP and bringing that tier of performance to Fury X price range at the least.

AllBodies · 19 Nov 2016 at 21:19

TNA said:
This is what I have been saying for a long time. The top end Vega needs to be at least matching the Titan PX when it comes out or it will be a failure IMO.

Can anyone explain how new architecture + HBM2 + 14nm + extra time to develop = 1070 performance? Just does not compute.

It was not that long ago that top end AMD (Fury X) and NVIDIA (Titan X) were like not hugely apart in performance. How did AMD suddenly go from that to it will be surprising if they are able to beat a 1070/80?

AMD have had an extra year to develop Vega so it needs to be beating Titan XP and bringing that tier of performance to Fury X price range at the least.

In other words, Vega should beat Nvidia's cards on a performance per mm2 basis, otherwise they're inferior relative to their release time.

JediFragger · 19 Nov 2016 at 21:33

Let's hope that Vega smashes nVidia a new one tbh!! We need options/competition dammit!!

Lokken86 · 19 Nov 2016 at 22:21

JediFragger said:
Let's hope that Vega smashes nVidia a new one tbh!! We need options/competition dammit!!

It won't. Up to 30% faster than a FuryX with a price tag to match is my guess.

LoadsaMoney · 19 Nov 2016 at 22:21

JediFragger · 19 Nov 2016 at 22:26

Lokken86 said:
It won't. Up to 30% faster than a FuryX with a price tag to match is my guess.

Well I hope it's 50% plus at least (although that's probably a lot to wish for!!)

Lokken86 · 19 Nov 2016 at 23:10

JediFragger said:
Well I hope it's 50% plus at least (although that's probably a lot to wish for!!)

Yea but you can tell they are struggling with something if it's taking this long to get it out the door.

theforce · 19 Nov 2016 at 23:12

I would expect 50% faster as a minimum, considering they are moving from 28nm to 14nm, new architecture and the extra capacity of HBM should help achieve that.

Lokken86 · 19 Nov 2016 at 23:14

theforce said:
I would expect 50% faster as a minimum, considering they are moving from 28nm to 14nm, new architecture and the extra capacity of HBM should help achieve that.

Well either they are struggling with clock speeds or something or working on something truly special but I doubt the latter

Rroff · 19 Nov 2016 at 23:17

Lokken86 said:
Yea but you can tell they are struggling with something if it's taking this long to get it out the door.

I believe they only just got upto speed with a revision of 14nm with optimisations for high performance products.

bru · 19 Nov 2016 at 23:43

Well Polaris was 40% faster than Tonga (probably only 25% if you actually take the 470 2048 cores vs the 380x 2048 cores) So maybe a 50% boost over the FuryX is asking a bit much.

There have been a couple of rumours now that have listed its processing power at 9.5TFlops ish which is about twice that of a 470. So that probably gives us a starting point, then with the bonus of HBM it could be a bit better, maybe putting it between a 1070 and 1080.

Bonjour · 19 Nov 2016 at 23:47

if it's priced £600+, will that be enough?

humbug · 20 Nov 2016 at 00:04

bru said:
Well Polaris was 40% faster than Tonga (probably only 25% if you actually take the 470 2048 cores vs the 380x 2048 cores) So maybe a 50% boost over the FuryX is asking a bit much.

There have been a couple of rumours now that have listed its processing power at 9.5TFlops ish which is about twice that of a 470. So that probably gives us a starting point, then with the bonus of HBM it could be a bit better, maybe putting it between a 1070 and 1080.

Your making the classic mistake of basing your equations off how the Fury-X does when bottlenecked.

when not bottlenecked it sits between the GTX 1070 and 1080, as one would expect for an 8.6 TFlop GPU.

Its actually really really simple, 4096 Shaders at 1050Mhz = 8.6TFlops.

@ RX 470 speed (1206Mhz) = 9.9TFlops
@ RX 480 Speed (1266Mhz) = 10.3TFlops

GTX 1070 = 6.5TFlops
GTX 1080 = 9TFlops

So what AMD need to do is un-bottleneck the shaders.

Kaapstad · 20 Nov 2016 at 00:11

Dave2150 said:
If HBM2 wasn't 'useful', NVIDIA wouldn't have used it on their most powerful GPU to date, the GP100.

HBM2 enables far higher memory bandwidth and far lower power usage, which enable a significantly faster GPU overall.

HBM2 in its current state on the Tesla cards will throttle cards capable of very high fps like the Pascal Titan.

Latency is more important than huge bandwidth when dealing with very high fps.

humbug · 20 Nov 2016 at 00:13

Kaapstad said:
HBM2 in its current state on the Tesla cards will throttle cards capable of very high fps like the Pascal Titan.

Latency is more important than huge bandwidth when dealing with very high fps.

The Latency with HBM is lower than GDDR.

Its not that.

Kaapstad · 20 Nov 2016 at 00:14

humbug said:
The Latency with HBM is lower than GDDR.

Its not that.

It is not lower than GDDR5X

Comparing HBM2 to GDDR5X is like a lorry to a sports car.

You will get a lot more work done with the lorry but the sports car will get you from A to B quicker.

CAT-THE-FIFTH · 20 Nov 2016 at 00:29

humbug said:
The Latency with HBM is lower than GDDR.

Its not that.

Yep,the physical traces are more closer to the GPU unlike GDDR5X. Fiji was bottlenecked by not having enough ROPs and probably CPU software bottlenecks under DX11. Under DX12 and Vulkan it does far better.

In fact if you look at Polaris 10,the improved command processor and primitive discard accelerator has improved performance under DX11 a decent amount. This is why in newer games a RX480 with significantly lower bandwidth is pushing ahead of even a R9 390 or R9 390X.