AMD Bristol/Stoney Ridge Thread

[Abwx]

I can't see anything missing, unless you mean the parts which turn into dust when the dies are sliced apart. Besides UVD blocks are located below the CPU CUs.

Although this debate is useless since it s not the point, wich was VP9 being supported :

I hope that you can see that the shots are flipped top to bottom but not left to right...

As to know the UVD capabilities by looking at a die, i also hope that you are joking...

 

[The Stilt]

Still can't see any difference. If you look at the file name of Bristol Ridge it says recolor. Color filters can fade some details or cause artifacts.

So since Bristol Ridge doesn't have additional HW for VP9 decoding (over Carrizo), it means that Carrizo should have the hardware too.

I can tell you that it doesn't. I know the UVD hardware used in Carrizo pretty well, since I worked on the software support quite a bit. UVD 6 doesn't have anything VP9 related in it, since Google finalized their distributed HW macros after Carrizo was basically fully laid out.

I know as a fact that Stoney Ridge has many changes to the hardware, including PSP, GPU and UVD. It is entirely possible that Stoney Ridge does have VP9 HW in die, however none of the technical documents mention it. The only UVD related change documented for ST is support for 10-bit HEVC. The other changes are related to PSP, which has been updated to the same level which Zen will use.

 

[dark zero]

So no 6W TDP model? Disappointing.

That was supposed to be the E2. Even more. The E2 was supposed to be TDP configurable like that and had that clocks:

Base (Lowest TDP) / Base (Max TDP) / Turbo
6W / 10W / 15W
1.0 Ghz/ 2.0 Ghz / 2.2 Ghz

Source: The initial leaks of Stoney Ridge.
Sadly seems that AMD discarded that.

 

[Abwx]

Still can't see any difference. If you look at the file name of Bristol Ridge it says recolor. Color filters can fade some details or cause artifacts.

Actually the red shot lacks top, bottom, left and right sides of the other shot die, if you cant see this "detail" what is the point of comparing dies.?..

So since Bristol Ridge doesn't have additional HW for VP9 decoding (over Carrizo), it means that Carrizo should have the hardware too.

That s a mysterious statement, so if i follow you if the most recent chip doesnt have a feature then the older one should have it...?..

I can tell you that it doesn't. I know the UVD hardware used in Carrizo pretty well, since I worked on the software support quite a bit. UVD 6 doesn't have anything VP9 related in it, since Google finalized their distributed HW macros after Carrizo was basically fully laid out.

The only thing that you can tell me is that you think that BR is the same thing at 100% as CZ, wich is not the case.

I know as a fact that Stoney Ridge has many changes to the hardware, including PSP, GPU and UVD. It is entirely possible that Stoney Ridge does have VP9 HW in die, however none of the technical documents mention it. The only UVD related change documented for ST is support for 10-bit HEVC. The other changes are related to PSP, which has been updated to the same level which Zen will use.

Why was BR launched after Sridge if it s the same thing as Carrizo..?.

Notice that AMD state VP9 support for both BR and Sridge, so any discussion on this point using Carrizo as reference is irrelevant, what you know about this latter doesnt obviously apply to the formers.

 

[cbn]

If AMD is supporting VP9 decode in Bristol Ridge via OpenCL acceleration, will they eventually add this to Carrizo as well?

 

[VirtualLarry]

If AMD is supporting VP9 decode in Bristol Ridge via OpenCL acceleration, will they eventually add this to Carrizo as well?

You would think, especially if there's nothing technical preventing it. But AMD's done this before, with their "Steady Video" support only running on certain newer products. (I forget, was it APUs, or dGPUs that had that? Or both).

 

[Abwx]

Look like it s not OpenCL based, otherwise they would have added CZ to the list below in respect of BR features :

http://www.amd.com/en-us/products/processors/laptop-processors#

It s possible that they use a non programmable (through microcode) fixed functions decoding hardware...

 

[monstercameron]

Look like it s not OpenCL based, otherwise they would have added CZ to the list below in respect of BR features :

http://www.amd.com/en-us/products/processors/laptop-processors#

It s possible that they use a non programmable (through microcode) fixed functions decoding hardware...

its definitely shader based just on the fast that the a9 only supports 720p(HD) with the beefier gpu models supports 1080p(FHD)

 

[Abwx]

its definitely shader based just on the fast that the a9 only supports 720p(HD) with the beefier gpu models supports 1080p(FHD)

I think that it has more to do with segmentation.

Stoney Ridge is for rock bottom priced items wich wont use 1080p screens, if it was shader based power comsumption would be too high if 192 SPs were not enough for 1080p, not counting that Carrizo would have the same feature from day one if it was the case...

 

[monstercameron]

I think that it has more to do with segmentation.

Stoney Ridge is for rock bottom priced items wich wont use 1080p screens, if it was shader based power comsumption would be too high if 192 SPs were not enough for 1080p, not counting that Carrizo would have the same feature from day one if it was the case...

video segmentation makes no sense on the decoding side, it just makes it look weak against its competitor.

 

[Abwx]

video segmentation makes no sense on the decoding side, it just makes it look weak against its competitor.

Certainly, but as pointed several times Carrizo has the same shaders count as BR, isnt it, so it s clear that it s not a question of shaders...

 

[dark zero]

Certainly, but as pointed several times Carrizo has the same shaders count as BR, isnt it, so it s clear that it s not a question of shaders...

That is worse... it looks that is a Freagin rebrand instead a new brand chip.

 

[DrMrLordX]

And will it be compatible with Win7 64-bit? (PLEASE!)

I would not expect Win7 compatibility from anything new. Sorry, just the truth.

 

[The Stilt]

VP9 decode is OpenCL based on both Carrizo (UVD 6.0) and Stoney Ridge (UVD 6.1).
Even the initial GCN4 based designs won't support VP9 HW decoding.

Most likely a highly optimized 3rd party SW codec (such as LibAV) will provide significantly better experience than the OpenCL solution. That was the case with Strongene HEVC OpenCL decoder at least.

I asked AMD which softwares currently support the acceleration, let's see what they say. Hopefully it is a driver feature at least.

 

[cbn]

Consumers also asking for 35W APU only sub $500 with Dual Channel DDR-4 and 14-15.6" but not AMD nor OEMs are lessening.

Can we have a A10-9630P APU only with dual Channel DDR-4 2400MHz at 14" format at sub $500 ???? That is what consumers want and are willing to spend.

That would sell like hotcakes, if performance was there.

Ideally at least one of AMD's partners would release a Mini-ITX board (or thin Mini-ITX, ideally with an option for riser and cut-out in the included tall Rear I/O shield) with one of these processors on it.

If this happens, many people will have quick access to the 35W level of performance for evaluation.

If evaluators in the wild like the performance (and I suspect they will based on the frequency scaling results reported by Stilt for Carrizo ) then this might encourage OEMs to invest in dedicated 35W/45W laptop motherboards.

 

[ET]

I just checked HP's site for the x360, the laptop AMD advertised the 7th generation with. The only memory configuration is single channel 8GB. I had really hoped that we're over with the Carrizo fails, but looks like AMD can't get the OEM's in line.

 

[The Stilt]

Single stick isn't a disaster, unless the same motherboard is used for Carrizo-L systems. 8GB 2400MHz DDR4 SO-DIMM module costs 32€. But I agree, shipping a APU based system with single channel memory is nothing short of retarded.

 

[Abwx]

I just checked HP's site for the x360, the laptop AMD advertised the 7th generation with. The only memory configuration is single channel 8GB. I had really hoped that we're over with the Carrizo fails, but looks like AMD can't get the OEM's in line.

Could you provide the link.?.

Because the Pavillions, wich are low cost, are dual channel :

http://h10032.www1.hp.com/ctg/Manual/c05122652

As for the X360 it s 16GB maximum wether it s AMD or Intel inside :

http://www.tomshardware.com/news/hp-envy-premium-laptop-update,31548.html

 

[monstercameron]

Could you provide the link.?.

Because the Pavillions, wich are low cost, are dual channel :

http://h10032.www1.hp.com/ctg/Manual/c05122652

As for the X360 it s 16GB maximum wether it s AMD or Intel inside :

http://www.tomshardware.com/news/hp-envy-premium-laptop-update,31548.html

http://store.hp.com/us/en/pdp/Lapto...ch laptop-t5y05av-1?jumpid=Laptops_Finder_PDP

single dimm...sigh

 

[dark zero]

Plain stupid movement... expected from HP. So there is no real BR product by now.

 

[Abwx]

As said max RAM is 16GB, for the single delivered DIMM it apply to Intel as well (at a higher cost..), so i dont see where s the drama..

http://store.hp.com/us/en/pdp/Laptops/hp-spectre-x360---15t-touch-laptop-p0a58av-1

 

[monstercameron]

As said max RAM is 16GB, for the single delivered DIMM it apply to Intel as well (at a higher cost..), so i dont see where s the drama..

http://store.hp.com/us/en/pdp/Laptops/hp-spectre-x360---15t-touch-laptop-p0a58av-1

who cares about intel? we are discussing amd.
also to top it off, intel's imc would probably extract more bw out of a single dimm vs amd'simc.

 

[dark zero]

And there goes again... OEMs ruining AMD moment to shine as always.

 

[Doom2pro]

Apparantly Wccftech has an OEM AM4 Bristol Ridge board image leak...

http://wccftech.com/amd-am4-motherboard-bristol-ridge-apu-spotted/

 

[Dresdenboy]

Apparantly Wccftech has an OEM AM4 Bristol Ridge board image leak...

http://wccftech.com/amd-am4-motherboard-bristol-ridge-apu-spotted/

This is making its rounds now.

I saw it first linked in the Heise article, mentioned here:
http://forums.anandtech.com/showthread.php?p=38264864#post38264864

Monstercameron already posted it in this thread, too.
http://forums.anandtech.com/showpost.php?p=38269289&postcount=107

So if you read the forums, you get it first.