Intel Kaby Lake-G might have Polaris graphics instead of Vega.

[Bharat Makhija]

Source : https://www.techquila.co.in/intel-kaby-lake-g-polaris-gpu-vega/

 

Intel Kaby Lake-G, (the Intel Haydes Canyon NUC specifically) might not have Vega graphics in it as advertised.

Quote

AIDA64, a third-party benchmark utility, identifies the Radeon GPU on the NUC as Polaris 22.

However, SiSoftware Sandra lists the Intel Haydes Canyon NUC to have the integrated GPU as Radeon RX Vega M GH. 

Quote

While replying on third party applications like AIDA64 and Sandra for identifying a GPU architecture is not very reliable, there are multiple other pointers that bolster the claim. The DirectX 12 feature level being limited to 12.0 being the most prominent one. Third-party utilities use PCI device IDs for identification. 694C and 694E are the device IDs for Kaby Lake-G Radeon graphics. These IDs also scream Polaris 22.

To summarize, the factors that help the argument that we might be getting the older Polaris instead of Vega are:

• DX12.0 instead of DX12.1.
• No Rapid Packed Maths support introduced with Vega.
• Uncertainty whether these can be enabled via a software patch.

Giving Intel/AMD the benefit of the doubt, the graphics solution packed with Intel Core i7-8809G in the NUC was custom designed for Intel, so accordingly, architecture modification to a certain extent is certainly possible. However, one thing can be said for sure:

Quote

Intel Core i7-8809G Kaby Lake-G is not the same Vega as the one in the Ryzen 5 2400G.

No comment from Intel/AMD, as of writing this post.

[H4ckbraten]

well it makes sence to sell your opponent a product that is not as good as ur own but better than his own. additionaly they named it not just vega m, what would lead to the point that its a mobile vega gpu, they named it rx vega m, more like the rx series which is based on the polaris architecture.

[DocSwag]

They should test to see if it has tile based rendering. That would be better proof IMO than rapid packed math, which could be driver related.

[Mira Yurizaki]

This would make sense depending on the time frame that Intel actually asked AMD for a partnership. It takes time for hardware to develop and I wouldn't be surprised if all AMD had to offer was Polaris.
 

Of course, the real question is who gets the blame?

[Sauron]

Well... define what "vega" means and then we can have a discussion... how AMD brands their own products is up to them. We should care about performance and features.

[Valentyn]

4 minutes ago, Sauron said:

Well... define what "vega" means and then we can have a discussion... how AMD brands their own products is up to them. We should care about performance and features.

Also custom Intel version; so likely made to their specifications.

[Bcat00]

Nothing more than a naming scheme, nothing new really.

 

nvdea's been doing that with their mobile chips for a long time.

[Dylanc1500]

15 minutes ago, Sauron said:

Well... define what "vega" means and then we can have a discussion... how AMD brands their own products is up to them. We should care about performance and features.

The question I suppose then becomes, which gen of GCN are they using?

[Sauron]

Just now, Dylanc1500 said:

The question I suppose then becomes, which gen of GCN are they using?

And the answer is: it really doesn't matter when we can benchmark it. After all, GCN is as much of a marketing label as Vega.

[Dylanc1500]

2 minutes ago, Sauron said:

And the answer is: it really doesn't matter when we can benchmark it. After all, GCN is as much of a marketing label as Vega.

That is true, but it is what they use to seperate the different generations of archs at least, with the code names falling under that, such as "raven ridge".

[SpaceGhostC2C]

A bit late to the party?  

 

 

[Guest]

Continuing the "what defines Vega?" well the APUs lack some features from the dGPU too.

Quote

Another interesting thought comes when analyzing this debate with the Ryzen 7 2400G and Ryzen 5 2200G products, both of which claim to use Vega GPUs as a portion of the APU. However, without support for HBM2 or the high-bandwidth cache controller, does that somehow shortchange the branding for it? Or are the memory features of the GPU considered secondary to its design?

https://www.pcper.com/news/Graphics-Cards/GPU-Intel-Kaby-Lake-G-More-Polaris-Vega

[Razor01]

https://www.pcper.com/news/Processors/Intel-Announces-New-CPUs-Integrating-AMD-Radeon-Graphics

 

They actually did an article about this last year stating in it looking more like Polaris than Vega.

 

Quote

The GPU looks to be based on the Polaris architecture which is a slight step back from AMD’s cutting edge Vega architecture.  Polaris does not implement the Infinity Fabric component that Vega does.  It is more conventional in terms of data communication.  It is a step beyond what AMD has provided for Sony and Microsoft, who each utilize a semi-custom design for the latest console chips.  AMD is able to integrate the HBM2 controller that is featured in Vega.  Using HBM2 provides a tremendous amount of bandwidth along with power savings as compared to traditional GDDR-5 memory modules.  It also saves dramatically on PCB space allowing for smaller form factors.

 

 

It makes sense AMD agreed to Intel with an older generation product just for competition sake (advantage for them).

[Mira Yurizaki]

2 hours ago, Razor01 said:

It makes sense AMD agreed to Intel with an older generation product just for competition sake (advantage for them).

I don't really believe that. Considering that Intel/AMD already had something resembling a finished product (or an engineering sample), it was likely that the decision to partner up had to be made way before then. My WAG is about a year from nothing to something. If coming from November of 2017, Vega was still an on- paper GPU as far as anyone was concerned.

[Razor01]

21 minutes ago, M.Yurizaki said:

I don't really believe that. Considering that Intel/AMD already had something resembling a finished product (or an engineering sample), it was likely that the decision to partner up had to be made way before then. My WAG is about a year from nothing to something. If coming from November of 2017, Vega was still an on- paper GPU as far as anyone was concerned.

 

 

They were showing Vega off a year before it was released   It was taped out mid 2016.  June or July if I remember correctly.

[Mira Yurizaki]

5 minutes ago, Razor01 said:

 

 

They were showing Vega off a year before it was released   It was tapped out mid 2016.  June or July if I remember correctly.

I can't seem to find anything regarding the GPU being taped out in 2016. This is the earliest my five minutes on Google has gotten me: https://segmentnext.com/2017/02/23/amd-vega-gpu-2/

 

On the flip side it could've been an executive decision by Intel to not go with the latest and greatest.

 

EDIT: There was a Tweet in mid 2016 about a major development milestone being achieved but no mentioning of a tape out (https://www.techpowerup.com/223672/amd-vega-10-gpu-crosses-a-development-milestone)

[Razor01]

twitter Raja Kaduri

 

 

That was when it taped out.

 

It takes about 6 months to year from that to getting risk production back and then finally final silicon, which is what we have, 6 months after they started showing off Vega.  Then final release 6 months after that.

 

December was the when they showed it off first, in live demos/games

 

[W-L]

This was previously reported upon here, let's keep discussion collectively in one thread:

 

 

 -Locked-