New Upcoming ATI/AMD GPU's Thread: Leaks, Hopes & Aftermarket GPU's

My guess is
Radeon RX Vega 4GB(Fury with Vega GPU)
Radeon RX Nano 4GB/8GB(Second Gen with Vega GPU downclock core)
Radeon RX Vega X 8GB(Fury X with Vega GPU)

 

Busted!
https://www.reddit.com/r/Amd/comments/68toj2/welcome_to_the_vega_hype_train_3dc_forum/

Still refusing to post relevant links, making me search for myself.
Still seamlessly mixing content pulled from different articles with with your own comments, again without referent links. So rude...

 

And where you put HBM2?

Big Vega will have HBm2.
The other Vega types,if will be,they will put with Gddr5x/6.

 

SRY @Noisiv :bang: (Always investigate :nerd: )

So, we are at square one -> besides all of this could be BS
As i said (many times) we need to wait for ATI/AMD to throw a bone at Us :thumbup:

 

AMD Radeon RX Vega Specs Leaked In Linux Patch:
AMD Radeon RX Vega Features 4 Shader Engines, 64 NCUs, 4 Render Back-Ends & 256 Texture Units



So first things first, the core count.
Vega 10 features 64 Next generation compute unit, each containing 64 GCN stream processors. The entire chip has a total of 4096 next generation GCN stream processors divided into four divisions each make up a single Shader Engine.
Each 1024 sp shader engine has two Asynchronous Compute Units, one render back-end and 4 texture blocks.
Each render back-end is comprised of 16 render output units, for a total of 64 ROPs (ROPs are with SMT so it has 128ROPs ?)
Each texture block is comprised of 16 texture mapping unit, for a total of 256 TMUs.
Vega 10 also supports 8 independent work threads simulatenously.

By comparison, Polaris 10 Pro the GPU powering the RX 470 has exactly half as many stream processors, half as many texture mapping units and half as many render output units. Polaris 10 Pro is a slightly cut back variant of Polaris 10 XT which powers the RX 480.
Vega 10 has double the render output units of Polaris 10 XT, can execute double the number of work threads and features 78% more stream processors and texture mapping units.

Spoiler

case CHIP_VEGA10:
adev->gfx.config.max_shader_engines = 4;
adev->gfx.config.max_tile_pipes = 8;
adev->gfx.config.max_cu_per_sh = 16;
adev->gfx.config.max_sh_per_se = 1;
adev->gfx.config.max_backends_per_se = 4;
adev->gfx.config.max_texture_channel_caches = 16;
adev->gfx.config.max_gprs = 256;
adev->gfx.config.max_gs_threads = 32;
adev->gfx.config.max_hw_contexts = 8;

PR:

– 4x Power Efficiency
– 2x Peak Throughput/Performance Per Clock
– High Bandwidth Cache
– 2x Bandwidth per pin
– 8x Capacity Per stack (2nd Generation High Bandwidth Memory)
– 512TB Virtual Address Space
– Next Generation Compute Engine
– Next Generation Pixel Engine
– Next Generation Compute Unit optimized for higher clock speeds
– Rapid Packed Math
– Draw Stream Binning Rasterizer
– Primitive Shaders

UPD. Whats this?
-> https://fccid.io/MSIP-REM-ATI-102-D08601

 

No, its still 64 ROPs.

But you can't compare them directly on different arch, and say "ewww why only 64"
Because ROPs on Polaris seem more efficient then on previous GCN revisions.
I doubt AMD engineers would make an unbalanced product which is strongly ROP limited.
My biggest hopes are in perf/Watt. I dont want 250W sitting under my desk in the middle of summer. Good riddance Hawaii

TLDR;
Vega > 2x 470/570, due to higher clocks.
Should be stronger than GTX 1080. 1080 being slightly less than 2x GTX 1060.

 

Now what we don't know, how the ATI New Tek (The so called Next-Gen) will behave in real world scenarios.

I can tell that it will be ~1080Ti (not OC one) up to new Xp in 4k scenarios in DX11 of course....

He He :idea: in DX12/Vulcan it will be unbeatable for some time IMO...

June is near now :banana:
Hmm i think we can Run the Not'Over-Hype Train now :nerd:

Besides:
Direct3D 11[46] was released as part of Windows 7. It was presented at Gamefest 2008 on July 22, 2008 and demonstrated at the Nvision 08 technical conference on August 26, 2008.[47][48] The Direct3D 11 Technical Preview has been included in November 2008 release of DirectX SDK.[49] AMD previewed working DirectX11 hardware at Computex on June 3, 2009, running some DirectX 11 SDK samples.[50]


lol it's 9 years of DX11 now ----> :behead:

 

All it matters now perf wise is... CLOCKS!

And maximum clocks are directly corelated to products intrinsic perf/Watt.
Theoretically speaking(!), if perf/Watt is good, they can clock the Vega sky high and beat the 1080Ti noproblemo!

I wanna find all those perf/Watt naysayers and beat them over the head with my new Vega :banana:

 

Maybe I'm greedy, but I hope it's more than 8gb of vram

 

Not likely, especially in the short term. From what I have read there is still a shortage of HBM2 at this stage. The price of the card would likely be astronomical. Maybe a bit down the track when it is more readily available.

 

HBM and the memory controller in Fiji at least is a little different to traditional graphics cards, the ram is used more like a cache which is why the Fury X is still able to produce decent frames at high resolutions, that and the latency is significantly lower.

 

They also seem to have expanded a lot on the memory controller tech.

 

Where is my Big Vega ? I'm waiting for you ...

 

Yes... aren't we all

 

3dcenter.org has proposed upcoming Vega architectural changes. Source speculation links can be found at the article. Apologies for the Google translation below.
http://www.3dcenter.org/news/was-sind-die-bekannten-architektur-verbesserungen-von-amds-vega

 

I think it's time for AMD to release more performance info on Vega, considering it is supposed to release sometime in the next several weeks. C'mon AMD give us patient people some news. Please....

 

yeah, some people are getting somewhat impatient, especially on https://www.reddit.com/r/Amd/comments/69ocxc/why_you_should_take_vega_leaks_with_salt/ and https://forum.beyond3d.com/posts/1979935/

 

VEGA will be Great if:

They can do
~200FPS average in 1080p
~144FPS average in 1440p
~90FPS average in 4k

Across the Board :approval:
so in DX9/10/11 and in Next-Gen DX12/Vulcan...

 

That is alot to ask for especially at 4k. If Vega can manage to match the 1080Ti, then AMD has got it right.

 

I'm being realistic here. With AMD's small R&D budget compared to Nvidia, I just can't see them getting near 90FPS @4K. The Ti can't even do that and Nvidia has pretty much been alone at the top for almost a year now. I too would love to see AMD lay the smack down on.

Hopefully with the success of RyZen, AMD will finally be able to give the RTG the money and resources they desperately need to bring a killer lineup. That probably won't happen in the near future though.