Radeon RX 480 "Polaris" Launched at $199

[~>Dg<~]

If he knew he would say so.

you cant say NDA

 

[FoxEye]

yes thats how it works, you are a retailer not quality control agency, you go by what the manufacturer tells you, and you do basic tests, not extensive ones, and even if you do those it wouldn't be you it would be someone working with you let's say 8pack for exemple.
we are talking about major companies not an unkown startup.

I may be a massive hypocrite, but I'm going to say it anyhow...

...you have no idea what you're talking about (in this instance).

you cant say NDA

First rule of NDA Club...

 

[ICDP]

unless there was some major performance degredation, the TX480 @1266mhz is 5.8Tflops theoretical while the 390 is 5.1Tflops at 1ghz. so i don't see why people are saying 390 performance all of a sudden. Along with architecture enhancements it should be faster than a 390x and going on to nano.

Oh I totally agree, the specs indicate it should be 390X - Nano type performance. And if it is at this level of performance for ~£230 I will agree it is a great GPU on price/perf. My issue us that many leaks and benchmarks are putting it at 390 performance, which if true puts RX 480 into the category of meh.

 

[LightSpeed]

Oh I totally agree, the specs indicate it should be 390X - Nano type performance. And if it is at this level of performance for ~£230 I will agree it is a great GPU on price/perf. My issue us that many leaks and benchmarks are putting it at 390 performance, which if true puts RX 480 into the category of meh.

It would be extremely surprising if it is slower than the 390x given the specs and what we know of the improvements to the arch. Which leaks point to slower than 390x other than the steam VR thing? I thought the earlier 3dmark scores were faster than the 390x?

 

[ICDP]

If you are referring to my scores yes that is in a MSI 980Ti Gaming at stock clocks, Sir.

Thanks for confirming.

 

[Gregster]

I was speculating and he told me categorically I was wrong, which is fine, why being the man who should know the performance of from it visiting vendors about the card tell someone that categorically and then go back later and say he didn't actually test it. to say he didn't actually test it is not to say he doesn't know, being who he is of course he knows.

Whether or not he inadvertently breached some sort of NDA is another question all together.

it also fits with the only indication AMD themselves have put out, that Steam VR score which is lower, a good chunk lower than the 970.

Its clear all the other rumours are fabricated, isn't that what a lot of people want me to acknowledge?

For goodness sake Humbug, give it a rest with this garbage. The card isn't for you, so move on and stop spouting rubbish please.

 

[adam98150]

There's some info on the AMD subreddit if anyone's interested.

I have asked several retailers in my country, all say stock arrives around June 20th. So it should be a large inventory. But, they also say they will only carry the 8GB because the 4GB was only being supplied to them fractionally cheaper.

So today local Sapphire merchant in Egypt ( Venus Egypt) just stated that they have the stock ready which seems like a great thing in global stocks since we always get our tech gadgets like a month late , the state included that the 4 GB model is gonna be late and they've only received only the 8 GB model which will retail for 275$ msrp for the current stock.

RX 480 XFX reference for $350 CAD pre order up at NCIX.

Sapphire Italy is saying that the 8GB (4?) version could be coming at about 250 Euro. Also, the cheese grater design is legit, but it's just a prototype.
EDIT: They also said that it has crazy OC capabilities

 

[ICDP]

It would be extremely surprising if it is slower than the 390x given the specs and what we know of the improvements to the arch. Which leaks point to slower than 390x other than the steam VR thing? I thought the earlier 3dmark scores were faster than the 390x?

I will link to my own post replying to Humbug on this very subject.

https://forums.overclockers.co.uk/showpost.php?p=29612654&postcount=2152

As can be seen there are certainly some rumours showing 480 to be roughly 390 type performance. Hence my concerns, even though those concerns seem illogical given the known RX 480 specs and architectural improvements.

Basically if AMD release RX 480 and it doe snot beat 390X (due to similar TFLOPS and architectural improvements) then 480 will be a regression IMHO.

 

[meldarthx]

why do people keep thinking that 390x/980 - nano performance is bad when it is moving down to the mainstream segment? baffles me.

Ding Ding Ding.......People keep ignoring what I've been saying

Ref 480 will come in at 390X/980 at most likely boost speed in DX 11 - DX 12 we're looking more along the lines of between Nano and Fury speeds.....and it will only get faster as drivers improve.

Now OCing 480 to 1400/1500 speeds should get it to FuryX area in DX 12; in DX 11 it very well could end up passing Nano - as other's have pointed out is that as fast as 980TI ? Nope still a little slower than that.

Now where the aftermarket cards fall in with the 8 and 6 pin plugs to allow a lot higher voltage and speeds......that's the big unknown could those hit 1800+? maybe or higher under water.....that is where possibly it could rival 980ti stock and 1070......but its a big but.....its really unknown.

We do know 470 in a few benches we've gotten matches 970 in DX 11...so there is no way 480 is only 970.....its 980+ speeds.....how much faster after that we're going to have to way

470 at 970 speeds for its prices at 150 or less....is a damn good deal....and 8 gig ref 480s will come in around 210-220.......and 4 gig should be around 175....

We know there is stock......that's been confirmed....7 days and a few hours left until we know......

 

[Scramz]

Thanks for confirming.

Should boost to 1279 out of the box but it boosts beyond that to 1353Mhz

 

[ICDP]

Should boost to 1279 out of the box but it boosts beyond that to 1353Mhz

Yep, pretty normal for 980Ti (all versions) to boost well beyond the published boost clocks. I can't see a 480 getting anywhere near reference 980Ti let alone AIB custom OC ones.

 

[XeNoNF50]

What a day it's been for the thread. Can't wait for another 7 more.

 

[humbug]

I will link to my own post replying to Humbug on this very subject.

https://forums.overclockers.co.uk/showpost.php?p=29612654&postcount=2152

As can be seen there are certainly some rumours showing 480 to be roughly 390 type performance. Hence my concerns, even though those concerns seem illogical given the known RX 480 specs and architectural improvements.

Basically if AMD release RX 480 and it doe snot beat 390X (due to similar TFLOPS and architectural improvements) then 480 will be a regression IMHO.

I should have known really, one of the reasons I own a 970 (Aside from it being a great card) is because after two years of owning a 290 (great card) AMD has nothing for me to upgrade to at the same money I paid for it, other than a 290 under another name.

It is starting to look to me like they are doing it again, which I find incredibly frustrating and bizzar.
Its certainly not healthy for them.

For goodness sake Humbug, give it a rest with this garbage. The card isn't for you, so move on and stop spouting rubbish please.

i'm just trying to have a conversation here. so please stop chasing me around with posts like that, thanks

 

[Scramz]

Yep, pretty normal for 980Ti (all versions) to boost well beyond the published boost clocks. I can't see a 480 getting anywhere near reference 980Ti let alone AIB custom OC ones.

No chance. Shame the stream earlier was not recorded so most of you guys who missed it could watch it back.

Someone mentioned a few pages back that people was on here 'screaming' on this forum about it only being 970 performance. The only reason it was mentioned was because the stream chat box was firing up with people comparing the both scores, it was simply translated from Chinese into this forum. It was a real live stream, viewing the entire setup, real benchmark (albeit most in here want to ignore it), no news article to question. Un-editted live stream. Hopefully on the 29th we see that it does perform better than we saw and then everyone is a winner!

Just don't expect anywhere near Ti level performance for $200, at stock clocks I still think it will perform at about 390 non X level and I don't think it will be an overclockers dream. Hope I am wrong.

 

[harney]

Here we go again....

She is fuelled up and ready to go

all aboard ...

Just not sure the engine can take another 7 days of this will see

 

[akarypid]

I will link to my own post replying to Humbug on this very subject.

https://forums.overclockers.co.uk/showpost.php?p=29612654&postcount=2152

As can be seen there are certainly some rumours showing 480 to be roughly 390 type performance. Hence my concerns, even though those concerns seem illogical given the known RX 480 specs and architectural improvements.

Basically if AMD release RX 480 and it doe snot beat 390X (due to similar TFLOPS and architectural improvements) then 480 will be a regression IMHO.

Rumours are rumours. Did you hear the one about it trading blows with the 980ti?

It's going to be 390X performance at reference clocks. This is almost certain judging by the released specs and you're right in that those concerns are illogical.

The real question is: how well does it OC?

 

[SimBy]

I will link to my own post replying to Humbug on this very subject.

https://forums.overclockers.co.uk/showpost.php?p=29612654&postcount=2152

As can be seen there are certainly some rumours showing 480 to be roughly 390 type performance. Hence my concerns, even though those concerns seem illogical given the known RX 480 specs and architectural improvements.

Basically if AMD release RX 480 and it doe snot beat 390X (due to similar TFLOPS and architectural improvements) then 480 will be a regression IMHO.

Slight performance regression is not unheard of when you drastically improve efficiency. RX480 is likely 2.2x perf/W compared to Hawaii. That's obviously more than 1.7x FinFET gives you. So there was definitely some major work on architecture.

For comparison, Pascal perf/W increase over Maxwell is much lower.

 

[Scougar]

Best in mind they are releasing this card significantly cheaper than when the 290 and 390 came out. Hardly fair to compare prices based on old models that will dry up soon.

Vega is coming, a wait still, but coming.

 

[humbug]

Rumours are rumours. Did you hear the one about it trading blows with the 980ti?

It's going to be 390X performance at reference clocks. This is almost certain judging by the released specs and you're right in that those concerns are illogical.

The real question is: how well does it OC?

Which IMO isn't really good enough, its just replacing one card in a price bracket with another.

With the new processing node we all expected to get more for less, especially after 4 years of 28nm.

Sure the price of those 28nm cards has come down a bit in those years but they are still expensive, the mid range card is still £300 while the same card 2 years earlier was what £400? I don't know i'm just guessing but that was very much overpriced.

To launch the same level card for about the same money moving from 28nm to 14nm its a bit of a kick in the teeth frankly.

 

[FoxEye]

Slight performance regression is not unheard of when you drastically improve efficiency. RX480 is likely 2.2x perf/W compared to Hawaii. That's obviously more than 1.7x FinFET gives you. So there was definitely some major work on architecture.

For comparison, Pascal perf/W increase over Maxwell is much lower.

It's not the improvement in efficiency that's the problem, more efficiency is always better than less.

It's the fact that all of that extra efficiency has been used to lower power, rather than give more perf. All of it.

So now we've just got a really, really low power 390 instead of a ... 390.