AMD Polaris architecture – GCN 4.0

D.P. · 31 May 2016 at 15:05

ICDP said:
http://www.hardocp.com/article/2016...x_1080_founders_edition_review/6#.V02WxrBRE-U

No difference between GTX1080 at DX11 vs DX12 in actual gameplay. Fury X shows an increase.

We can cherry pick all day but for me the facts point to GTX 1080 showing no real improvement going from DX11 to DX12.

DX12 reduces CPU overhead, so what exactly do you expect to see? The benchmarks prove that Pascal DX11 drivers are not CPU bound.

Do you want Nvidia to purposely gimp DX11 drivers in order to show a bigger gain in DX12?

Rroff · 31 May 2016 at 15:06

Toss3 said:
I don't see how AMD's biggest, and only, competitor being mentioned in a thread about an upcoming AMD product would be irrelevant? We were discussing benefits gained from DX12, as that is an area where AMD have had the biggest advantage (and still do imo), and one way for Polaris to be more attractive than the GTX 1070 for customers. I think a lot of people are waiting for Polaris until deciding which product to settle for, so everything you'll be seeing tomorrow form AMD is going to be compared to their competitors' products (i.e. Intel and Nvidia).

I suspect when there is any actual Polaris news the theme of the thread will switch to that anyhow, there isn't exactly much else for people to talk about at the moment.

Seanspeed · 31 May 2016 at 15:06

ICDP said:
http://www.hardocp.com/article/2016...x_1080_founders_edition_review/6#.V02WxrBRE-U

No difference between GTX1080 at DX11 vs DX12 in actual gameplay. Fury X shows an increase.

We can cherry pick all day but for me the facts point to GTX 1080 showing no real improvement going from DX11 to DX12.

Holy crap, ok.

For like the 3rd or 4th time, the point is that the *difference* between a 980Ti and a 1080 is *bigger* in DX12 implementations than it is in DX11 implementations. Look across all the benches shown of a variety of different DX12 and DX11 games. That shows an absolutely undeniable improvement in DX12 instructions for Pascal. Undeniable, cannot be argued, even the other guy finally agreed with it. Even the bench you linked there shows it.

Stop with this man. Myself and everybody else wants to move and stop talking about this, but you absolutely INSIST on spreading this misinformation in the face of all the actual facts. This is boring for me and for everybody else. Give it up.

Boomstick777 · 31 May 2016 at 15:11

Seanspeed said:
Holy crap, ok.

For like the 3rd or 4th time, the point is that the *difference* between a 980Ti and a 1080 is *bigger* in DX12 implementations than it is in DX11 implementations. Look across all the benches shown of a variety of different DX12 and DX11 games. That shows an absolutely undeniable improvement in DX12 instructions for Pascal. Undeniable, cannot be argued, even the other guy finally agreed with it. Even the bench you linked there shows it.

Stop with this man. Myself and everybody else wants to move and stop talking about this, but you absolutely INSIST on spreading this misinformation in the face of all the actual facts. This is boring for me and for everybody else. Give it up.

He's not spreading misinformation. He said The 1080 has no gains in DX12 VS DX11 in that game which is true.

So you are spreading misinformation about him. Reported.

Gregster · 31 May 2016 at 15:13

To sum up what we know about Polaris for those who might have missed it among the dx12 discussions....

It will be called a 490 or 490X, so we assume it will be the 390/X replacement.
It will compete with the Nano
Clocks are not stated or overclocking potential.
Memory isn't stated but I expect 8GB of 512 bit as standard.
Price is expected to be low, so expect $350 ish

Vega potentially coming pre Chrimbo but could be as early as October and will be the big card that is expected to give a 1080 a beating and should be competing with the 1080Ti.

Seanspeed · 31 May 2016 at 15:17

Boomstick777 said:
He's not spreading misinformation. He said The 1080 has no gains in DX12 VS DX11 in that game which is true.

So you are spreading misinformation about him. Reported.

Reported? Holy hell, we're in Kindergarten here.

And he picked out the one benchmark that proved the little thing that he wanted to show even though I showed TWO other benchmarks that shows entirely opposite. NOT THAT THAT WAS EVEN WHAT THE ARGUMENT WAS ABOUT. It was about whether Polaris has improved DX12 performance over Maxwell. That and that alone. Which even the benchmark he provided demonstrated.

Maybe read through the discussion before going nanny on us and telling the teacher when you clearly dont understand what's going on?

FFS

Confused Stu · 31 May 2016 at 15:19

Gregster said:
To sum up what we know about Polaris for those who might have missed it among the dx12 discussions....

It will be called a 490 or 490X, so we assume it will be the 390/X replacement.
It will compete with the Nano
Clocks are not stated or overclocking potential.
Memory isn't stated but I expect 8GB of 512 bit as standard.
Price is expected to be low, so expect $350 ish

Vega potentially coming pre Chrimbo but could be as early as October and will be the big card that is expected to give a 1080 a beating and should be competing with the 1080Ti.

Cheers Greg, genuinely appreciated from someone who doesn't spend the time trawling through the thread looking for the occasional glittering nugget.

Roll on more info/samples/reviews when they turn up.

Toss3 · 31 May 2016 at 15:23

Gregster said:
To sum up what we know about Polaris for those who might have missed it among the dx12 discussions....

It will be called a 490 or 490X, so we assume it will be the 390/X replacement.
It will compete with the Nano
Clocks are not stated or overclocking potential.
Memory isn't stated but I expect 8GB of 512 bit as standard.
Price is expected to be low, so expect $350 ish

Vega potentially coming pre Chrimbo but could be as early as October and will be the big card that is expected to give a 1080 a beating and should be competing with the 1080Ti.

Some small errors here - according to the latest rumors the cards will be called RX 480 and RX 480X. It will also probably be 8Gb of 256-bit GDDR5 memory (like the GTX 1070). Price should also be lower approx. $299. And not sure where you got the nano-part from, but according to the latest leaks (if they are indeed leaks), the RX 480X is going to go head-to-head with the GTX 1070 (20k 3DMark GPU score).

beany_bot · 31 May 2016 at 15:25

so no 490/490x?

If the 480x is competing with the 1070 at a lower price then I might grab one and a 1440p monitor to boot.

Or do I save up and wait for vega and a 4k monitor.....argh....what to do.

Gregster · 31 May 2016 at 15:29

Confused Stu said:
Cheers Greg, genuinely appreciated from someone who doesn't spend the time trawling through the thread looking for the occasional glittering nugget.

Roll on more info/samples/reviews when they turn up.

Toss3 said:
Some small errors here - according to the latest rumors the cards will be called RX 480 and RX 480X. It will also probably be 8Gb of 256-bit GDDR5 memory (like the GTX 1070). Price should also be lower approx. $299. And not sure where you got the nano-part from, but according to the latest leaks (if they are indeed leaks), the RX 480X is going to go head-to-head with the GTX 1070 (20k 3DMark GPU score).

No worries and yep Toss, my mistake about the naming. As for the memory, if it is 512 bit, it would pull in the faster cards at 4K and could well be a good selling point but maybe I am just being overly optimistic on that. I was also going on a videocardz rumor about the speed. I know we had the steps meter thingy but not sure if too much is being read into that or not

Seanspeed · 31 May 2016 at 15:30

beany_bot said:
so no 490/490x?

If the 480x is competing with the 1070 at a lower price then I might grab one and a 1440p monitor to boot.

Or do I save up and wait for vega and a 4k monitor.....argh....what to do.

I think we'll see a baby Vega take the 490 series moniker and big Vega get its own flagship naming ala Fury.

As for Vega and 4k, well I'd say that to get the same performance that P10 would get at 1440p, but in 4k, Vega would likely need to be about 100% more powerful(in actual demonstration, not just on-paper flops or whatever). Most benches show that framerates are halved going from 1440p->4k.

I'd say we're still a generation away from 4k/60fps being a truly reasonable target to attain from a single card solution.

Toss3 · 31 May 2016 at 15:35

Gregster said:
I was also going on a videocardz rumor about the speed. I know we had the steps meter thingy but not sure if too much is being read into that or not

Don't you dare trample on my hopes and dreams!

Still weird to have a goal of 7783 (goal is usually an even number like 10k), and actual walked steps being basically 20k which is what most modern gpus are getting in 3dmark.

But, only ~12hours to go!

bru · 31 May 2016 at 15:36

Gregster said:
To sum up what we know about Polaris for those who might have missed it among the dx12 discussions....

It will be called a 490 or 490X, so we assume it will be the 390/X replacement.
It will compete with the Nano
Clocks are not stated or overclocking potential.
Memory isn't stated but I expect 8GB of 512 bit as standard.
Price is expected to be low, so expect $350 ish

Vega potentially coming pre Chrimbo but could be as early as October and will be the big card that is expected to give a 1080 a beating and should be competing with the 1080Ti.

I'm sorry have I missed something, I thought it was going to be the 480/x series.
a lot of the rumours point to a 256bit bus and 8GB ram performance is unknown, you cannot take that chart from PC gmaer seriously as they said they just put the line on there next to the Nano as they think that is where the performance will fall.
as for price AMD have said they want to bring VR capability to the mainstream price point.

Of course if there has been any actual announcement or concrete leaks to the contrary then I apologise but I must have missed then in amongst all the moaning about Pascal or talk of Zen and the many other non Polaris related things.

Nutella33 · 31 May 2016 at 15:37

Power colour unveiling new water cooling blocks for new single and dual cards is interesting!

Yrs I know wffc but looks officialish.

http://wccftech.com/powercolor-liquid-cooler-amd-polaris-amd-radeon-pro-duo/

Sounds like we might even get proper details. Rumours of Zen news as well elsewhere.

The event will feature in-depth details of the Polaris GCN 4.0 architecture along with their new CPU lineups on the AM4 and FP4 platforms.

beany_bot · 31 May 2016 at 15:39

Seanspeed said:
I think we'll see a baby Vega take the 490 series moniker and big Vega get its own flagship naming ala Fury.

As for Vega and 4k, well I'd say that to get the same performance that P10 would get at 1440p, but in 4k, Vega would likely need to be about 100% more powerful(in actual demonstration, not just on-paper flops or whatever). Most benches show that framerates are halved going from 1440p->4k.

I'd say we're still a generation away from 4k/60fps being a truly reasonable target to attain from a single card solution.

I dunno, the GTX1080 is getting mighty close to 4k. I was kind of thinking that the 1080Ti might do it. And big Vega too.

Sure I could happily game at 1440p for a few years but I am desperate to go full blown 4K. I want the monitor to watch 4K youtube and amazon anyway but it would be a crime not to game on it at 4k. Don't want any blurriness when downscaling to 1440p. Argh I dunno. Its a headache.

Jono8 · 31 May 2016 at 15:48

Toss3 said:
Some small errors here - according to the latest rumors the cards will be called RX 480 and RX 480X. It will also probably be 8Gb of 256-bit GDDR5 memory (like the GTX 1070). Price should also be lower approx. $299. And not sure where you got the nano-part from, but according to the latest leaks (if they are indeed leaks), the RX 480X is going to go head-to-head with the GTX 1070 (20k 3DMark GPU score).

By leaks, you mean some guys pedometer reading...

I really think people are clutching at straws with that one.

Gregster · 31 May 2016 at 15:57

Toss3 said:
Don't you dare trample on my hopes and dreams! Still weird to have a goal of 7783 (goal is usually an even number like 10k), and actual walked steps being basically 20k which is what most modern gpus are getting in 3dmark.

But, only ~12hours to go!

I sure hope you are right but....

Jono8 said:
By leaks, you mean some guys pedometer reading...

I really think people are clutching at straws with that one.

I do think this. But you never know.

@bru. Yer, I was corrected but thanks anyways

Jono8 · 31 May 2016 at 16:01

I hope it is true as well but im not getting my hopes up

I suspect the videocardz leaked benches are probably closer to the truth as they were bang on with the 1080 and 1070.

Freddie1980 · 31 May 2016 at 16:07

D.P. said:
Do you want Nvidia to purposely gimp DX11 drivers in order to show a bigger gain in DX12?

If it meant flogging more graphics cards then I wouldn't put it past them.

Mauller · 31 May 2016 at 16:08

Jono8 said:
By leaks, you mean some guys pedometer reading...

I really think people are clutching at straws with that one.

He has earlier pics of him talking about steps he has taken throughout the day, the goal value keeps changing and the Steps taken and goal on the final pic he posts line up far too well with 3dMark 11 P and X scores that something cheeky could easily be going on. Although i don't think they are the exact values if he has seen them, since it would be too obvious, they could be +-10% of the actual values which still puts it around furyx - titanx performance.

Tweet with the pedometer and values.
https://twitter.com/killyourfm/status/736836910954381312

His gallery of the trip with earlier pedometer pics on it.
https://www.instagram.com/thegametechnician/