[WCCF] AMD Radeon R9 390X Pictured

[RussianSensation]

so did you emmigrate to CA from RU and now you just post on ATF all day?

hehe, I do spend a lot of time here lately. I did immigrate from RU to Canada at 13, but I haven't lived in Canada in the last 5 years. Last 6 months I have a lot of downtime since the project I was on has ended; so I have been able to post a lot more often. Speaking of that, I am going to get really busy for the next 2 weeks. I have a deadline on June 17th. You jinxed me. Then I am off to Toronto around the 20th to visit some friends as I have been working in the US/UAE/Asia and haven't seen any of them since August 2010. I'll try to post around the 16th just to see the reactions on the forums in case Fiji XT beats Titan X. :biggrin:

About the leaked slide showing 4096, could it be real, but not representative of the flagship card?

I have no idea. If we use the 3 leaks I posted, I would say 4096 SP is the top as my best guess.

It's possible. Savage die?

Oh, that's what they should have called Fury. Radeon Savage.

https://www.youtube.com/watch?v=HgZEPPuG4Rc

 

[sm625]

Do we even know if Fury is TSMC or GlobalFoundries?

 

[sze5003]

hehe, I do spend a lot of time here lately. I did immigrate from RU to Canada at 13, but I haven't lived in Canada in the last 5 years. Last 6 months I have a lot of downtime since the project I was on has ended; so I have been able to post a lot more often. Speaking of that, I am going to get really busy for the next 2 weeks. I have a deadline on June 17th. You jinxed me. Then I am off to Toronto around the 20th to visit some friends as I have been working in the US/UAE/Asia and haven't seen any of them since August 2010. I'll try to post around the 16th just to see the reactions on the forums in case Fiji XT beats Titan X.



I have no idea. If we use the 3 leaks I posted, I would say 4096 SP is the top as my best guess.



Oh, that's what they should have called Fury. Radeon Savage.

https://www.youtube.com/watch?v=HgZEPPuG4Rc

You used to post a lot more few years ago if I remember correctly. Helped me out picking my card when I was upgrading. I've also cut down on posts since our office no longer uses cubicles.

 

[AnandThenMan]

Do we even know if Fury is TSMC or GlobalFoundries?

I don't recall any concrete info on this.

 

[poofyhairguy]

Speaking of that, I am going to get really busy for the next 2 weeks. I have a deadline on June 17th. You jinxed me.

Seems like AMD jinxed you lol.

 

[Vesku]

At least AMD has now confirmed they are doing the R300 top end reveal at their E3 conference June 16th:

http://www.amd.com/en-us/products/g...y?hootPostID=5702f524841825e46a77b1626bd8f980

"New Era of PC gaming" so they are apparently pretty confident in it.

 

[DearLord]

At least AMD has now confirmed they are doing the R300 top end reveal at their E3 conference June 16th:

http://www.amd.com/en-us/products/g...y?hootPostID=5702f524841825e46a77b1626bd8f980

"New Era of PC gaming" so they are apparently pretty confident in it.

Maybe it will be a single gpu card that can comfortably do 4k with high/ultra settings.

 

[Silverforce11]

With their Timesquare "Revolution" and now this "New Era".. it better destroy Titan X on performance or it's over-hyped! That kind of language is dialing up the hype-meter to 9999!

 

[chimaxi83]

Leave it to AMD to over hype and under deliver... I'll keep my expectations in check, but hoping for a beast finally.

 

[Azix]

technically, "new era" is a given. It has the next gen in VRAM.

 

[AnandThenMan]

Leave it to AMD to over hype and under deliver... I'll keep my expectations in check, but hoping for a beast finally.

Why do people keep saying this, what has been delivered? Care to explain?

 

[JDG1980]

Why do people keep saying this, what has been delivered? Care to explain?

I think this perception is largely the fault of John Fruehe and the failed Bulldozer launch. Prior to Bulldozer's release, Fruehe hyped Bulldozer to the skies on various forums, often getting people banned for disagreeing with him. When Bulldozer ended up being a turd, AMD got a reputation for over-hyping and dishonesty among many technical forum-goers.

 

[AnandThenMan]

That's cherry picking. Fermi was hyped to the moon for months it was going to be nothing short of a revolution a super computer on a chip. A fake non working product was shown on stage for crying out loud. Plus Bulldozer is not even a GPU, and right now AMD has done precious little to hype anything in fact before today we had basically nothing from them.

The hype surrounding stacked memory is happening all by itself AMD doesn't have to do anything since it is something very different than what we've seen before on a graphics card.

 

[MrTeal]

That's cherry picking. Fermi was hyped to the moon for months it was going to be nothing short of a revolution a super computer on a chip. A fake non working product was shown on stage for crying out loud. Plus Bulldozer is not even a GPU, and right now AMD has done precious little to hype anything in fact before today we had basically nothing from them.

The hype surrounding stacked memory is happening all by itself AMD doesn't have to do anything since it is something very different than what we've seen before on a graphics card.

I wouldn't say the product didn't work. Reliable sources within nVidia reported that Jen-Hsun Huang was able to reuse several parts of that demo when he built a deck the following year.

 

[Silverforce11]

That's cherry picking. Fermi was hyped to the moon for months it was going to be nothing short of a revolution a super computer on a chip. A fake non working product was shown on stage for crying out loud. Plus Bulldozer is not even a GPU, and right now AMD has done precious little to hype anything in fact before today we had basically nothing from them.

The hype surrounding stacked memory is happening all by itself AMD doesn't have to do anything since it is something very different than what we've seen before on a graphics card.

The taint of Faildozer is still strong.

At least Fermi was faster when it was released and yes, its compute abilities were tremendous and cemented NV's dominance in HPC.

For AMD to hype this up ("World's fastest GPU!" "Revolution!" "New ERA!"), they better deliver Titan X killing performance or its a dud.

 

[AnandThenMan]

For AMD to hype this up ("World's fastest GPU!" "Revolution!" "New ERA!"), they better deliver Titan X killing performance or its a dud.

Sounds like you're already expecting it to fail, because let's be honest it is not going to be a "Titan X killer" unless it is 40% faster and that is not going happen.

 

[Silverforce11]

I'll be happy with 20% faster, cheaper at R290X power usage levels. Easy to please!

 

[JDG1980]

That's cherry picking. Fermi was hyped to the moon for months it was going to be nothing short of a revolution a super computer on a chip. A fake non working product was shown on stage for crying out loud. Plus Bulldozer is not even a GPU, and right now AMD has done precious little to hype anything in fact before today we had basically nothing from them.

The hype surrounding stacked memory is happening all by itself AMD doesn't have to do anything since it is something very different than what we've seen before on a graphics card.

Agreed, Deckscrewgate was pretty embarrassing for Nvidia. And the 6-month delay for the Fermi launch was also a clear sign that something had gone very wrong. Delays usually are; that's one reason why the Fiji launch has a lot of people nervous. Even though AMD never specified an official release date prior to the E3 announcement, there is a perception that Fiji is "unofficially" late to the party.

By the way, here's the conclusion to AnandTech's review of GTX 480 (Fermi) at the time. We can see that power consumption isn't an issue that was just recently made up to embarrass AMD; it's been part of the conversation about video cards for a long time. And prior to Maxwell, AMD had the upper hand as often as not. Here's a card that beat AMD's contemporary top tier SKU by over 10%, yet reviewers were reluctant to recommend it for the same reason people are reluctant to recommend Hawaii today: it was too hot, loud, and power-hungry.

The GTX 480 Fermi could consume up to 320 watts under a stress load. That's pretty bad, but not as bad as I thought I remembered. Gaming loads peaked at 257W, which was about the rated TDP. So Fermi was about on par with Hawaii, more or less, in terms of power consumption. The GTX 580 was a respin (GF100 -> GF110) on the same process, and it dropped gaming power consumption to 226W peak. I would be disappointed if the Hawaii respin didn't provide at least that much power saving. It should be able to do even better, since the 480 was a cut-down part (480 shaders) while the 580 was fully enabled (512 shaders). A more apples-to-apples comparison would be the GTX 480 compared to the GTX 570 (which had the same 480 shaders on the same GF1x0 series part); the 570 only used 190W on gaming loads. If AMD could do that well with a respin, then Grenada wouldn't be too far behind GM204 in performance per watt...

 

[JDG1980]

Sounds like you're already expecting it to fail, because let's be honest it is not going to be a "Titan X killer" unless it is 40% faster and that is not going happen.

I would consider it to be a "Titan X killer" if the Fury X at stock settings has a comfortable lead over heavily factory OC'd Titan X cards (like the EVGA hybrid). If Fury X beats Titan X, stock to stock, by under 10%, that's a small enough lead that Fury X may lose to the EVGA hybrid Titan X, and Nvidia could release a fully-enabled, hot-clocked GM200 SKU (GTX 990?) to beat it at stock. But if the lead is, say, 20%, then that's not going to happen.

 

[DownTheSky]

Titan is already fully enabled.

 

[iiiankiii]

I'll be happy with 20% faster, cheaper at R290X power usage levels. Easy to please!

If they can manage that, it would be a winner. Even if Fiji is a poor overclocker, at 20% faster than the Titan X, that means an overclocked Titan X would just match it. Plus, a 20% overclocked Titan X consumes more power than the r9 290x. Much, much more.

 

[3DVagabond]

Your day job is safe. On the WC is this the first time a board is specifically designed for it? I mean the size of the card could not exist without a water cooled system right?

I think it's more the HBM that allows for the smaller PCB. Nothing stops the cooler from being bigger than the PCB.

 

[3DVagabond]

Naa but at =>600m2 and probably even denser and less area for mem controller what is the chance of 4096 shaders?
I would say its less than 5%
Performance we dont know because of primarily freq target and perf/w target but we are as close as confirmation we can be before facts hits us this is more than 4096. I will eat my hat if its less than 5k

Didn't we see a GPU over a year ago that showed 4096 SP but also 4x ARM cores?

 

[flopper]

For AMD to hype this up ("World's fastest GPU!" "Revolution!" "New ERA!"), they better deliver Titan X killing performance or its a dud.

You assume its in regard to whatever Nvidia has but gaming with VR is a new era of gaming. I assume Dx12 and win 10 will change how their new line up works as Dx11 and whatever windows before is going away.

so dx12, win 10, VR a new era of gaming right there.

I dont expect Miracles from Fury as even if they are equal to Titanx or even faster it be in the same ballpark with fps anyhow. its really a premium for a few more fps.

 

[LTC8K6]

Titan is already fully enabled.

I'm sure NV has a plan if Fury is much faster than the 980Ti.