Poll: ** The AMD VEGA Thread **

[Rannoch]

So will there be proper benchmarks in 8 days then?

 

[Jono8]

Surprised theres been nothing from that PDXLAN yet.

I'm surprised there has been nothing. Not one leaked bench, picture, price on the RX Vega cards.

 

[AlamoX]

wth a 484mm² die vega shouldn't be too far from the Ti, i for once think AMD might be sandbaging.

 

[RSR]

Its a myth I tell ya... Its NEVER comin!

Nooooooooooooooooooooooooooooooooooo......

 

[Gerard]

wth a 484mm² die vega shouldn't be too far from the Ti, i for once think AMD might be sandbaging.

If its not then they done something majorly wrong.

 

[Gerard]

Surprised theres been nothing from that PDXLAN yet.

Couple of pics doing the rounds of people playing on vega systems but not much else, no card pics or anything so far.

 

[TNA]

8 days remain till we find out about RX Vega.

Yeah, really looking forward to this still for some reason

I am expecting at least GTX 1080 performance for around £400-450 in which case I might pull the trigger. Still got the £375 sitting around from when I sold my 1080 (the same price I purchased it for) so as long as it is not a noisy bitch I would be getting a brand new card and ability to use Freesync with it. Not bad when I think of it like that

If they try and charge £500 for 1080 performance however, they can keep their card.

 

[akarypid]

From what I've seen there are 3 different pipelines - traditional, some kind of quasi tile based deferred immediate mode which is basically a compatibility hack without full performance benefits and the full tiled based implementation with maximum performance - from the looks of it drivers can force the first 2 but its upto the application developer to support the highest performance variant.

AMD seem to have gone for a more "elegant" solution rather than nVidia's brute force implementation with better ability to flexibly handle different loads but downside is it seems it needs more input from developers to get the same or better levels of performance from it so....

EDIT: From reading up on it the approach AMD is using is more about power (snigger) and bandwidth saving than all out performance while nVidia uses a brute force approach that can also have bigger implications for performance.

Hey Rroff, care to share any references? I like reading about this stuff (have the tech background to understand enough of it).

 

[old gamer]

Yeah, really looking forward to this still for some reason

I am expecting at least GTX 1080 performance for around £400-450 in which case I might pull the trigger. Still got the £375 sitting around from when I sold my 1080 (the same price I purchased it for) so as long as it is not a noisy bitch I would be getting a brand new card and ability to use Freesync with it. Not bad when I think of it like that

If they try and charge £500 for 1080 performance however, they can keep their card.

Pretty much the same, I think £450 is my max for 1080 performance

 

[WeSbO]

The new geometry pipeline is the primitive shaders part of things, nothing to do with the DSBR. The primitive shaders as with the rest of then can be taken advantage of in drivers on a per game basis, but for best performance they need to be directly coded for.

Yes but with some of the improvements in the Geometry pipeline needing to be implemented in game, if those improvements aren't available in game would the pixel engine, later in the render pipeline, still take advantage of DSBR (maybe the reason for requiring the driver to choose the preferred rasterization method?), because from what I've read (and quite possibly misunderstood) a lot of the optimizations are intertwined, one earlier stage in the rendering pipeline improving the next stage, improving the next stage and so on, so is it still efficient if one of those optimisations isn't used, hell could it even hinder performance, a latency of some sort or something even if it's benefits of the L2 cache ?
So if DSBR did depend on other optimisations that need to be implemented in game, even though the driver decides rasterization and not needing to be optimized in game, it would still need those earlier in game optimizations, so ultimately dependent on developers

 

[RSR]

Yeah, really looking forward to this still for some reason

Agreed!

I have been hovering over the buy button on a FTW3 for the last month though, so it better be good.

 

[nashathedog]

It is a hardware feature that is independent to the game, it is something AMD will sort in the driver itself. It is not something that is opened to a game through an API.

Where did you read that? I'd be interested in having a read up on it too if you can point me in the right direction please.

 

[nashathedog]

Yeah, really looking forward to this still for some reason

I am expecting at least GTX 1080 performance for around £400-450 in which case I might pull the trigger. Still got the £375 sitting around from when I sold my 1080 (the same price I purchased it for) so as long as it is not a noisy bitch I would be getting a brand new card and ability to use Freesync with it. Not bad when I think of it like that

If they try and charge £500 for 1080 performance however, they can keep their card.

I'm hoping it doesn't go beyond 500. If it does it'll be hard to swallow.

 

[Mtom]

So will there be proper benchmarks in 8 days then?

NDA is few days after Siggraph, but can change, also it seems AIB cards in September.

 

[AllBodies]

Navi doesn't really need 7nm, it needs whatever their scaling mechanism will be to work and scale well.

Process nodes don't work like that. They've designed Navi on 7nm, so they can't just 'zoom in' and put it on 14nm with less cores.

It looks like Linus is gong to rip AMD a new one next week. skip to 21:38ish

Hasn't he implied there that there are features not turned on with FE Vega?

And even that they might not be able to be turned on for some reason?

His whole analogy of early-access, and having to buy a DLC to get the full features even though you already paid?

 

[r7slayer]

Developers wont use it, as they have no market share, thats what they need first, like i said earlier.

They ain't going to bother with TrueAudio, if only about 10 people have it, we saw it with Vulkan, and everything else they've done, Vulkan, its a piece of **** to do, one bloke, couple of hrs, yet no one was interested in it, they had to get rid of it in the end, then when Kronos took it, and Nvidia jumped on board (as they wouldn't touch it when it was under AMD), who have the market, it came back to life, and now the devs are all over it, and the games are a coming.

Dx12s the same, its only being patched into Dx11 laters in the odd game here, there, as its AMD only at present, yes Nvidia can do Dx12, but not very well, as they are mainly Dx11, but that's supposed to be changing with Volta, that's supposed to be geared for 12, and if it is, then we'll see Dx12 explode into life, just like Vulkan, as the devs will be all over it, and we'll be able to finally say goodbye to the odd Dx11'er, patched 12 laters, but until then, may as well just forget about it.

What i see here is. Devs won't touch AMD unless nvidia are involved. True audio has been available since 2013 in devices so i doubt it's 10 people but nicely going making out it being very insignificant. Doesn't make you come across sour at all. Mantle's problem was that purely AMD cards initially could only use the API and whilst more people had nvidia cards yes it made sense for devs not to bother. Now that Vulkan is here and is not only capable of running on nvidia hardware its capable of being on other OS too as well as other hardware there is no excuse other than devs who CBA because they would rather stick with the old stuff because it's easier. Don't make out that nvidia made Vulkan popular lol, without mantle it would be nothing. Weve heard people talk about a couple of people to code vulkan into games in a few days, that's no time at all. DX12 works better on AMD hardware yea due to its architecture which is why we don't see great uplift on DX12 because once again devs don't want to push something that benefits the few aka AMD even though if done right it is a better API to use. Wasn't pascal supposed to change things with DX12 for nvidia with it having the "Highest" support lol.

Either way i'm getting fed up of lazy developers and the bias in this gaming industry which is just leaving us and leading us into stagnation. Crappy ports and crappy DX12 patches. Unfinished games and you tend to need to buy DLC or even expansions to get the full experience. Don't do demo's anymore to see how the game runs etc to try before you buy. Origin do that 1 hour thing which is good but thats not devs. Not touching DX12 because it works better on AMD hardware even though if done right can boost nvidia hardware. Too much work apparently. Gotta keep that backwards compatibility for all those still stuck with DX11 hardware only or even Win7/8. Then there is Vulkan which is just as good doesn't get touched either.
Ryzen gota a lot of hate early on just purely because its AMD. Now tables are turning because they are actually good and good value for money so even nvidia are using them in their battleboxes.

 

[AlamoX]

Navi doesn't really need 7nm, it needs whatever their scaling mechanism will be to work and scale well.

navi blueprint is done on 7nm, making another one for 14nm will cost them a lot of money and time they dont have.

 

[opethdisciple]

What i see here is. Devs won't touch AMD unless nvidia are involved. True audio has been available since 2013 in devices so i doubt it's 10 people but nicely going making out it being very insignificant. Doesn't make you come across sour at all. Mantle's problem was that purely AMD cards initially could only use the API and whilst more people had nvidia cards yes it made sense for devs not to bother. Now that Vulkan is here and is not only capable of running on nvidia hardware its capable of being on other OS too as well as other hardware there is no excuse other than devs who CBA because they would rather stick with the old stuff because it's easier. Don't make out that nvidia made Vulkan popular lol, without mantle it would be nothing. Weve heard people talk about a couple of people to code vulkan into games in a few days, that's no time at all. DX12 works better on AMD hardware yea due to its architecture which is why we don't see great uplift on DX12 because once again devs don't want to push something that benefits the few aka AMD even though if done right it is a better API to use. Wasn't pascal supposed to change things with DX12 for nvidia with it having the "Highest" support lol.

Either way i'm getting fed up of lazy developers and the bias in this gaming industry which is just leaving us and leading us into stagnation. Crappy ports and crappy DX12 patches. Unfinished games and you tend to need to buy DLC or even expansions to get the full experience. Don't do demo's anymore to see how the game runs etc to try before you buy. Origin do that 1 hour thing which is good but thats not devs. Not touching DX12 because it works better on AMD hardware even though if done right can boost nvidia hardware. Too much work apparently. Gotta keep that backwards compatibility for all those still stuck with DX11 hardware only or even Win7/8. Then there is Vulkan which is just as good doesn't get touched either.
Ryzen gota a lot of hate early on just purely because its AMD. Now tables are turning because they are actually good and good value for money so even nvidia are using them in their battleboxes.

This is partly why I never bought a 10 series card. Even before Ryzen released and people started saying that the Nvidia drivers where not delivering in DX12, I felt like there was an issue with DX12 and Nvidia.

We will see I imagine a change in developer behaviour once Volta launches which I expect to be a DX12 class GPU (10 series are really a DX11 class GPU with some DX12 capability).

Once Nvidia have the full hardware they will be the one's to champion DX12 and Vulkan and we will start seeing game design changing amongst the devs.

This should* help AMD as well, but it will be on Nvidia's terms.

A bit like the Mayweather vs McGregor fight. Yea sure they'll fight, but it'll be a boxing match rather than MMA.

----

Although saying that I am going to contradict my self as I have been eyeing up 1080's over the last few weeks. No other options....

 

[r7slayer]

This is partly why I never bought a 10 series card. Even before Ryzen released and people started saying that the Nvidia drivers where not delivering in DX12, I felt like there was an issue with DX12 and Nvidia.

We will see I imagine a change in developer behaviour once Volta launches which I expect to be a DX12 class GPU (10 series are really a DX11 class GPU with some DX12 capability).

Once Nvidia have the full hardware they will be the one's to champion DX12 and Vulkan and we will start seeing game design changing amongst the devs.

This should* help AMD as well, but it will be on Nvidia's terms.

A bit like the Mayweather vs McGregor fight. Yea sure they'll fight, but it'll be a boxing match rather than MMA.

Yea seems to be this way but will game devs start making games with DX12 in mind and leaving backwards compatibility for DX11 for people with other OS? As DX12 is Win 10 only. It makes Vulkan far more attractive in my eyes. I still can't wait to see what or how RX Vega cards do in DX12 and Vulkan although supposedly was more optimised for running better in a DX11 pipeline.
AFAIK it seems Pascal is just shrunk Maxwell with added support for DX12. Hopefully Volta will bring DX12 to the table more.

 

[r7slayer]

navi blueprint is done on 7nm, making another one for 14nm will cost them a lot of money and time they dont have.

Im quite excited to see what Navi will be like. 7nm and its the architecture which Raja had most involvement in. Not to mention is using "Next gen memory". Also the potential to make a big GPU out of higher salvageable smaller dies could be quite interesting.