AMD Polaris architecture – GCN 4.0

Stanners · 23 May 2016 at 15:10

Ferrari100 said:
One thing is sure, AMD are super - secretive with this gen.
I actually wonder if Nvidia know what they are up against. If AIB's have no clue then Nvidia's spies might be struggling to get facts as I am sure thats where many leaks come from.

Its nice to see that AMD's engineers and employees are just getting there heads down, working hard and not hyping anything up (or allowing the media to hype them up).

Why would AMD let Nvidia have sales if they knew they had something special, Saying that with AMD's past PR it would not surprise me.

andybird123 · 23 May 2016 at 15:27

SiDeards73 said:
So basically "Primitive Discard Accelerator" is AMD's Gameworks Bleach? because its no secret Nvidia loves tessellation as it has a negative impact on AMD hardware and a not so negative impact on their own.

So what has AMD effectively done here? found a way at hardware level to just discount all this unnecessary Tessellation? If thats the case would we see any gains on Polaris Hardware running games that use a lot of Gameworks tessellation etc?

the only rumour I've seen is that PDA is AMD's name for Conservative Rasterisation, if you read the description of what CR does, it discards primitives and its a hardware implementation that's faster than a software only algorithm, its also something that was missing from previous AMD cards

its not do to with culling tesselation though, as an example its to do with not culling parts that are needed to make up a detailed run time shadow, the old fast way of doing it in software cause artefacts, the new fast hardware method results in better image quality without a corresponding increase in GPU time

AMD appear to have also added tesslation parts to Polaris, but I don't think this is directly related to PDA

Lokken86 · 23 May 2016 at 15:42

Fact is nvidia and amd influence each other. If someone wants to talk about how an NVIDIA competing product could impact the Polaris launch or whatever,as long as it's in context, I don't have a problem with it. Just so long as it's not obvious trolling.

Ferrari100 · 23 May 2016 at 15:46

Lokken86 said:
Fact is nvidia and amd influence each other. If someone wants to talk about how an NVIDIA competing product could impact the Polaris launch or whatever,as long as it's in context, I don't have a problem with it. Just so long as it's not obvious trolling.

+1

nashathedog · 23 May 2016 at 15:47

Almost tea-time here, Time for a Twix....

Admiral Huddy · 23 May 2016 at 16:01

Lokken86 said:
Fact is nvidia and amd influence each other. If someone wants to talk about how an NVIDIA competing product could impact the Polaris launch or whatever,as long as it's in context, I don't have a problem with it. Just so long as it's not obvious trolling.

Agreed.

Let me make this perfectly clear, any obvious trolling or attempts to bait others will be dealt according the warning system laid out here.

https://forums.overclockers.co.uk/showthread.php?t=18585963

Ferrari100 Considering you've been here for a while yet only have a few hundred posts, you seem to have only returned to wind people up so have a short holiday to cool off.

Thread re-opened.

erazor51 · 23 May 2016 at 16:08

Link: http://videocardz.com/60253/amd-radeon-r9-480-3dmark11-benchmarks

Lokken86 · 23 May 2016 at 16:10

erazor51 said:
Link: http://videocardz.com/60253/amd-radeon-r9-480-3dmark11-benchmarks

Told you my Fury was looking nervous

. That performance would be great if priced right.

Stanners · 23 May 2016 at 16:13

Lokken86 said:
Told you my Fury was looking nervous . That performance would be great if priced right.

These could be nice if priced correctly.

erazor51 · 23 May 2016 at 16:13

WhyCry said:
One more thing, according to Sisoft database, C7 has 2304 cores, there is a possibility that there's another P10 chip with 2560 cores (full silicon), even faster than those. Just wanted to make it clear.

Rroff · 23 May 2016 at 16:15

What they do need is a card at that CF performance point that can beat the 1070 on price.

Lokken86 · 23 May 2016 at 16:17

If they do have a 2560 shader card that would leapfrog the Fury x. Maybe with some oc headroom to boot.

tavtavtav · 23 May 2016 at 16:17

There's absolutely no way the 480x will be that fast at stock...
Edit: oh crossfire? Nvm.

old gamer · 23 May 2016 at 16:20

Lokken86 said:
Told you my Fury was looking nervous . That performance would be great if priced right.

mine too, I'm suddenly getting the urge to crossfire again

Mauller · 23 May 2016 at 16:21

tavtavtav said:
There's absolutely no way the 480x will be that fast at stock...

And why not? AMD made some large improvements, they could even be faster for all we know.

Lokken86 · 23 May 2016 at 16:22

tavtavtav said:
There's absolutely no way the 480x will be that fast at stock...

It's best case scenario as far as I'm concerned. If true it's exactly what they need for a good priced mid range card.

Not impossible either. The 6870 was mid range and it was just as fast as the 5870. So was the 970 compared to the first Titan.

tavtavtav · 23 May 2016 at 16:24

I was looking at the crossfire results

flopper · 23 May 2016 at 16:24

Lokken86 said:
Told you my Fury was looking nervous . That performance would be great if priced right.

seems in line with everything expected.
driver + hz = price/performance king

D.P. · 23 May 2016 at 16:24

erazor51 said:
Link: http://videocardz.com/60253/amd-radeon-r9-480-3dmark11-benchmarks

Impressive if true, could be. Fingers crossed.

EDIT, I didn't see the top result was crossfire (CF). These numbers make sense with the current rumours. Videocardz benchmarks were accurate with the 1080.
It all comes down to price, Fury performance at half the price? Will sell very well.

h4rm0ny · 23 May 2016 at 16:31

erazor51 said:
Link: http://videocardz.com/60253/amd-radeon-r9-480-3dmark11-benchmarks

From the same article: "Chart shows the best possible scenario based on those results". It's just a projected on rumours and leaks.

That said, if it were true...