Radeon RX 480 "Polaris" Launched at $199

[Lokken86]

Do we know anything about the top end 480 (X is it?) at all, or is it mostly speculation at this stage? I've been a bit blind to AMD lately, but this 1070 pricing fiasco now has me seeing RED!

We know the rough price and that it looks to perform somewhere from 390 all the way to Fury.
That's pretty much all we know haha.

Edit: Sorry I didn't read that properly. You meant the rx480x. No we know nothing. Not even if it actually exists.

 

[Legend]

Its all speculation for the $300 card,no one knows apart from Amd.

When is more info expected, or don't we know that either lol?

 

[karik666]

When is more info expected, or don't we know that either lol?

Well the nda lifts on the 29th june,which is the release date for the rx 480,so we should hear something then,lol

 

[kieran_read]

You never know all this silence from AMD could just be a very sophisticated marketing ploy.

Maybe they want people to buy Nvidias very expensive cards so when they release their cards at such a ridiculously cheap price with great performance and under cut the crap out of Nvidia. Those who bought the FE editions and expensive cards will actively rant and rave and promote AMD cards for them while at the same time capitalising on the burnt loyalty of those who were once Nvidia.

or...

They just have nothing and have failed miserably.

 

[ICDP]

If your strategy is to release mid-range cards first and the high-end later, how could you NOT bring out a new GPU that matches the performance and price of an existing GPU? Obviously if you release the mainstream first then there's going to be an existing high-end that probably matches it on performance. And if that existing card matches it on performance, why the Hell would anyone continue to sell it at substantially more than the new card that it is no more powerful than?

Logically if a company releases mainstream first then what you end up with is a down-pricing of the existing high-end cards alongside new cards that can be matched by the old but are more efficient, have new features, etc., and a void at the high end waiting to be filled by the new generation high-end. And the previous generation vanish as stocks deplete.

You're asking for something that doesn't make logical sense if a company releases mainstream first.



Not "most". Just the minority enthusiast segment that makes up places like the OCUK forums. If it delivers that performance at this price, what it will be doing is turning high-end graphics into the mainstream.

Your missing my point. Some performance numbers leaked suggest RX 480 could be similar to R9 390 and if these rumours are true we have the following scenario.

R9 390 - 8GB, ~£230+ with performance around 25% slower than Fury X.
RX 480 - 8BG, ~£230+ with performance around 25% slower than Fury X.

R9 390 and GTX970 have been this rough price for a long time now. Bringing out a new GPU that does not improve on already existing price/perf at that price point is utterly idiotic. Imagine if all AMD or Nvidia did was bring out a GPU at the same price and perf every year but say "hey this one uses less power". This is called stagnation.

1080 is around 6% higher than GTX980Ti prices but also ~30% faster.
1070 is around R9 Fury prices but also ~15% faster.

I think both are laughably overpriced but at least there is a performance increase over previous gen cards.

Now if RX480 ends up at Nano/Fury speeds for ~£230 then yes its an excellent increase in price/perf and a great GPU. But matching existing price and performance of R9 390 would be stagnation, consumers would be paying the same price for the same performance.

 

[humbug]

Firstly, 150w is actually. Min "max" power form 6pin and pcie, you can actually typically draw far more than that, not that I think the card does.

Secondly, form AMDs official figures and the estimated performance it will be closer to 150w. They claims twice the performance per watt, so 390x performance at half the power. 275/2 >> 100 w. 130w is more realistic.

Not this yet again.

Fury-X 375 Watts board power. actual use 275 Watts
290 300 Watts board power actual use 275 Watts

RX 480 150 Watts Board power.

AMD stated performance per watt increase = 2.8x, so 390X @ 275 Watts ~ by 2.8 = 98 Watts.

 

[AlamoX]

You never know all this silence from AMD could just be a very sophisticated marketing ploy.

Maybe they want people to buy Nvidias very expensive cards so when they release their cards at such a ridiculously cheap price with great performance and under cut the crap out of Nvidia. Those who bought the FE editions and expensive cards will actively rant and rave and promote AMD cards for them while at the same time capitalising on the burnt loyalty of those who were once Nvidia.

or...

They just have nothing and have failed miserably.

you know thats not how it works, ppl do not go from a 200£ budget to 600£ on a whim, these are 2 different targets, AMD didnt release more info because they dont want to give Nvidia more info they might need to finish the last tweaks on the 1060 the real competitor for their card, the more they wait the more nvidia waits to show their 1060, if nvidia cannot wait and launch it, they might end up slower or have the wrong price point.
if the 480 turn out to be good, it would out sell the 1080 10 to 1, on a market share percpective that is much more impactful
.

 

[brawl3r]

Could be an honest mistake. But they should take the video down.

Totally agree, gamernexus article called them out.

 

[Mauller]

Not this yet again.

Fury-X 375 Watts board power. actual use 275 Watts
290 300 Watts board power actual use 275 Watts

RX 480 150 Watts Board power.

AMD stated performance per watt increase = 2.8x, so 390X @ 275 Watts ~ by 2.8 = 98 Watts.

Its up to 2.8, so more than likely in situations where new features boost performance over shader utilisation improvements alone. But there are also reductions in power from smaller memory bus and using half the number of ram packages. So power is likely around 80-110watt dependant on clocks and situation considering it has fury nano performance, fewer shaders and higher clocks than 390x.

 

[humbug]

Its up to 2.8, so more than likely in situations where new features boost performance over shader utilisation improvements alone. But there are also reductions in power from smaller memory bus and using half the number of ram packages. So power is likely around 80-110watt dependant on clocks and situation considering it has fury nano performance, fewer shaders and higher clocks than 390x.

Whatever it is its half the power consumption or less for the same performance or more.

Can't grumble at that and be taken seriously.

 

[Mauller]

Whatever it is its half the power consumption or less for the same performance or more.

Can't grumble at that and be taken seriously.

Yeah, those figures are for the core alone. Memory system is another beast, fury nano gives a good example of power differences the memory system can make. Even at 1ghz the card uses less power with a bigger gpu core compared to the 390x. Etc etc gddr5 and larger bus uses more power with 16 packages compared to a faster 256bus with 8 chips.

 

[D.P.]

https://en.wikipedia.org/wiki/PCI_Express#Power

150W from PCI-E + 6pin is the maximum power consumption.

They don't claim twice performance per watt, they claim 2.8x.

150w is the minium that a mobo and psu 6pin can provide,not the max. Just no guarantee you can get more but would be rare not to. I don't think AMD would push 150w but it is important to u derstand the power ratings are mininums, not maximums.

You have to take the 2.8 claim under the same light and Nvidias 3.0x. It applies to specific theoretical scenarios under VR. And claim themselves 1.7x from the 14nm process. You can expect 20-30% from architecture improvements, that gives you the double.performance per watt which is the most recent official information AMD gave investors. This is the similar real world improvement.

The missing puzzle piece is single.pass mutliprojettison do VR. Nv claim 70% savings, AMD will have similar.

You add all that up and you can makes claims like 2.8x to 3.0x PR headlines. Or the 2.0x which you tell investors when lying and unsubstantiated claims have legal ramifications.

 

[D.P.]

Not this yet again.

Fury-X 375 Watts board power. actual use 275 Watts
290 300 Watts board power actual use 275 Watts

RX 480 150 Watts Board power.

AMD stated performance per watt increase = 2.8x, so 390X @ 275 Watts ~ by 2.8 = 98 Watts.

Not this again.

AMD claim 275w for the FuryX, not 375w, it uses 250-300w in gaming on average depending on utilisation.

The 2.8x is only in VR. 2.0x is the official average.

 

[AllBodies]

The 2.8x is only in VR. 2.0x is the official average.

Source?

 

[karik666]

Im sure Amd never mentioned Vr when they said about 2.8x.

 

[D.P.]

Yeah, those figures are for the core alone. Memory system is another beast, fury nano gives a good example of power differences the memory system can make. Even at 1ghz the card uses less power with a bigger gpu core compared to the 390x. Etc etc gddr5 and larger bus uses more power with 16 packages compared to a faster 256bus with 8 chips.

I believe AMD claimed a 30w saving with HBM
So maybe there is 15w to save with a 256bit bus over the 390x. That would push my estimate down to about 125w.


The nano is not useful for Compariaons because it is a hand selected chip with the best power characteristics, and then downclocked

 

[AllBodies]

Still waiting on that source...

To be honest, it wouldn't be surprising if AMD managed 2.8x performance per watt in general (not just specific VR circumstances), because this would only put them pretty much identical with the GTX 1080 (which holds the perf/w crown right now).

Either they'll be behind a bit, ahead a bit, or identical to the 1080. Depends which 28nm AMD card you take as the start point, since they never said 2.8x more than "insert card here".

Unless you think Nvidia mix unicorn farts into their cards, there's no reason why AMD couldn't match them, or get mildly ahead. They leapfrog each other often, if you look back.

 

[D.P.]

Im sure Amd never mentioned Vr when they said about 2.8x.

You have to read the fine print, which is under NDA. But what they claim is improvements with Radeon technologies, which will include VR.

 

[D.P.]

Still waiting on that source...

To be honest, it wouldn't be surprising if AMD managed 2.8x performance per watt in general (not just specific VR circumstances), because this would only put them pretty much identical with the GTX 1080 (which holds the perf/w crown right now).

Either they'll be behind a bit, ahead a bit, or identical to the 1080. Depends which 28nm AMD card you take as the start point, since they never said 2.8x more than "insert card here".

They said 2.0x times the previous AMD mid-range in the most recent AMD financials

 

[AlamoX]

so a lot of retailers received their RX480 stock, reviewers apparently also received their samples to review, the next week we might see some definitive leaks about performance.