Radeon RX 480 "Polaris" Launched at $199

[humbug]

Hopefully the rx490 will be released on the 29th june

Yeah.

 

[shankly1985]

Thats what am rooting for tbh Do AMD have something they not yet showing us?

 

[humbug]

Thats what am rooting for tbh Do AMD have something they not yet showing us?

As he said, its unlikely AMD are using one GPU from the P10 SKU and calling it the plain 480, no X or XT or 5...
just as unlikely its just that one GPU with a 6 pin power connector.
This is also the $200, with "GPU's upto $300"

For more likely there is another version with two 6 pin connectors, higher clocked and with more shaders putting it more in the 1070 performance area for $300.

 

[D.P.]

A 40Cu and higher clocked version is only going to be about 20% faster. A great card for $300, still not 1070 ballpark, but solid performance for $80 cheaper.

 

[triss]

A 40Cu and higher clocked version is only going to be about 20% faster. A great card for $300, still not 1070 ballpark, but solid performance for $80 cheaper.

Will depend on what clock speed its at and just how much of a help GCN4 is . at that price it would be a very popular card

 

[humbug]

A 40Cu and higher clocked version is only going to be about 20% faster. A great card for $300, still not 1070 ballpark, but solid performance for $80 cheaper.

Depends on how you interpret ballpark.

I'll put it another way, +20% would have it 95% the performance of a 980TI.

I guess one would have to consider much faster is the 1070 than the 980TI? before it starts to look like too close to the 1080, are Nvidia doing another Titan-X and then 980TI with no real performance difference at half the cost.

 

[OnionSpark]

I am considering a new build with an AMD 8320e and two 480's.

I have a theory that AMD's new feature GCN 4 feature set may point to better CPU utilization in DX11 games along with the DX12 benefits already available to it.

Any thoughts on that?

Interesting. You work for one of AMD's AIB partners - do you have access to some sort of information there that we do not?

I can't quite see how even a single 480 would not be horribly, horribly gimped by an extremely low-end CPU like the 8320e, let alone dual 480s. It would certainly be interesting if what you're saying is true, as that would fix one of AMD's current gaping problems, and make the 480 appealing to a lot more people. That would be excellent news (for everyone except a friend of mine, who recently had to pick up a 960 for his AMD CPU system! Just a few more weeks and he could have had a 480!).

Are you under an NDA that is preventing you from giving more details? Can you at least give us a hint?

 

[humbug]

Interesting. You work for one of AMD's AIB partners - do you have access to some sort of information there that we do not?

I can't quite see how even a single 480 would not be horribly, horribly gimped by an extremely low-end CPU like the 8320e, let alone dual 480s. It would certainly be interesting if what you're saying is true, as that would fix one of AMD's current gaping problems, and make the 480 appealing to a lot more people. That would be excellent news (for everyone except a friend of mine, who recently had to pick up a 960 for his AMD CPU system! Just a few more weeks and he could have had a 480!).

Are you under an NDA that is preventing you from giving more details? Can you at least give us a hint?

It comes from Raja explaining the new front end he was building for Polaris, some months ago.
It amounted to identifying and rectifying AMD's problem with batching.

 

[OnionSpark]

It comes from Raja explaining the new front end he was building for Polaris, some months ago.
It amounted to identifying and rectifying AMD's problem with batching.

I was actually hoping for something more practical than theoretical! I know a lot of people here simply believe Nvidia and AMD's marketeers and press statements, but I've always preferred independent confirmation where possible.

Wasn't it Raja who also claimed a historic leap in performance per watt? Polaris looks to be far behind Pascal there if we assume that all of AMD's figures are accurate, and behind Maxwell if there's even the slightest exaggeration.

Which hints that AMD is only comparing with AMD, and so "rectifying" batching doesn't therefore mean that there will be a significant improvement compared with the market at large, only with AMD's earlier GPUs.

I don't think I'm being overly skeptical here - a 480 with a low-end AMD CPU? Not outside the realm of possibility - please tell me more. Two 480s with a low-end AMD CPU? That's.... revolutionary.

 

[humbug]

I was actually hoping for something more practical than theoretical! I know a lot of people here simply believe Nvidia and AMD's marketeers and press statements, but I've always preferred independent confirmation where possible.

Wasn't it Raja who also claimed a historic leap in performance per watt? Polaris looks to be far behind Pascal there if we assume that all of AMD's figures are accurate, and behind Maxwell if there's even the slightest exaggeration.

Which hints that AMD is only comparing with AMD, and so "rectifying" batching doesn't therefore mean that there will be a significant improvement compared with the market at large, only with AMD's earlier GPUs.

I don't think I'm being overly skeptical here - a 480 with a low-end AMD CPU? Not outside the realm of possibility - please tell me more. Two 480s with a low-end AMD CPU? That's.... revolutionary.

Thats a bit ironic, given that we have no clue what the performance per watt is of Polaris.

We have less information on that than we do anything.

 

[SiDeards73]

What find Ironic is Onionspark pretty much stalked Ferrari around these and other forums pretty much calling him out and insulting him, and is now asking him for info lol.

Onionspark your a troll mate, you made word and phrase your posts nicely but at the end of the day your a troll.

If you are sincere in your questioning (which always appear to be fairly condescending) I think you should firstly apologise to the guy before asking him for info.

Also it was said a long long time ago DX12 would have the capability to harness more cores and that actual physical cores appear to be better than virtual ones, I will see if I can dig up the info.

Maybe he wants a low end AMD CPU because they are cheap and it's been common knowledge for a while there's very little difference in the AMD CPUs once you over clock them in regards to performance in games

 

[Gregster]

You should have reported the post that you found offensive Si instead of going with the personal attack. That isn't helping this thread at all again, pretty much like this post isn't.

 

[Concrete]

FWIW, I use my 'low end' 8320 over my i5 everyday

 

[andybird123]

Its hardly surprising that with so much unsubstantiated positive rumouring going on, there are going to be people who are a bit skeptical and ask for sources rather than just blindly accepting the most positive spin.

 

[SiDeards73]

You should have reported the post that you found offensive Si instead of going with the personal attack. That isn't helping this thread at all again, pretty much like this post isn't.

Yeah probably Greg, just found it ironic that the guy appeared a few weeks ago, basically trolled and stalked another forum member, then comes back begging for info lol, double standards much?

Anyhow, i was hoping to see some leaks by this weekend, guess not, whens the next show or event that AMD are attending? E3 is it? someone mentioned Hotchips also hopefully get more info then

 

[chroniclard]

Its hardly surprising that with so much unsubstantiated positive rumouring going on, there are going to be people who are a bit skeptical and ask for sources rather than just blindly accepting the most positive spin.

Dont think anyone is blindly accepting anything. Just comparing/speculating on what little info there is so far.

 

[Kaapstad]

Yeah probably Greg, just found it ironic that the guy appeared a few weeks ago, basically trolled and stalked another forum member, then comes back begging for info lol, double standards much?

Anyhow, i was hoping to see some leaks by this weekend, guess not, whens the next show or event that AMD are attending? E3 is it? someone mentioned Hotchips also hopefully get more info then

Who cares about rumours or leaks, all that interests me is actual performance figures from a number of tech sites and user benches.

 

[Kaapstad]

Dont think anyone is blindly accepting anything. Just comparing/speculating on what little info there is so far.

Best thing to do about speculating is

1. Go down the pub

2. Drink 15 pints of Lager

3. Enjoy yourself

4. When you walk out the pub if you still want to speculate it will be just as accurate as when you walked in the pub.

 

[Gregster]

Yeah probably Greg, just found it ironic that the guy appeared a few weeks ago, basically trolled and stalked another forum member, then comes back begging for info lol, double standards much?

Anyhow, i was hoping to see some leaks by this weekend, guess not, whens the next show or event that AMD are attending? E3 is it? someone mentioned Hotchips also hopefully get more info then

Well that seems to have seeped over several forums with Ferrari and others and Ferrari did say he worked for an AIB but has since told us he is a teacher, And I don't care either way in truth. I can see Ferrari is an enthusiast like many of us and wants info like many of us.

End of June for this 480 and there will be many leaks, rumors and speculation in between, which will no doubt spark off more debates lol. The things we have seen though are the 3DMark11 score which looks to be for the 480 and with a clock speed of 1266Mhz (iirc). The score was very respectable for a sub $200 card and if that is indeed the correct score, it looks like it could well suit many mid range gamers. Raja would have been better going with the 3DMark run in my opinion, rather than that weird CF AoTS test lol.

 

[AlamoX]

As he said, its unlikely AMD are using one GPU from the P10 SKU and calling it the plain 480, no X or XT or 5...
just as unlikely its just that one GPU with a 6 pin power connector.
This is also the $200, with "GPU's upto $300"

For more likely there is another version with two 6 pin connectors, higher clocked and with more shaders putting it more in the 1070 performance area for $300.

well that would make it about FuryX performance for 250£ excellent performance per dollar behind 1070 by 10-15%.
but i still dont believe they have another card higher, simply because as he said AMD seem to have worked very hard on efficiency, and would be stupid to throw all that out by pushing another extra 10-15% boost with overclock, because 4CUs should only give 11%.