AMD’s Fastest Radeon , Fury X2, Launching In December – Two Fiji XT GPUs And A Terabyte Of Memory Bandwidth

[MEC-777]

I'm curious how it will perform and cope with 4GB of HBM at high resolutions. One thing we can say, for now at least, is that this dual Fiji card will have no direct competition for a while. Unless Nvidia is hiding a dual 980Ti up their sleeves that we don't know about...

[thekeemo]

Birthday on the 22nd *hint hint*

 

tags entire friend list in hopes they will pool resources

@TheCovertCamper @Sauron @Salty @MblaZe7run @Lolzious @Juanse @iamdarkyoshi @hypnodog @EarthboundHero @CoolaxGaming @CtW  @Comessy @BurgerBum  @AlexStriz @Sam Z Man @The Official Czex 

[Dabombinable]

doesnt matter really... the 290/390 core itself isnt powerful enough to fully utilize all that VRAM... it can push maybe 5-6GB tops...

hell, even a Titan X, with 12GB of VRAM wont actually have the GPU horsepower to give you reasonable framerates if a game NEEDS 12GB of VRAM (games can hog VRAM... but being so damanding that they NEED it is something different from hogging it for themselves)

WRONG

Powercolor released a air cooled R9 390x2. which is based upon 2x R9 390 cores. Unlike the 295x2 which is based on 2x 290X cores AND is watercooled

"GTX 650ti isn't powerful enough to use 2GB of vRAM", oh, look, the GPU is running at 89% because it ran out of vRAM.

[The Benjamins]

But wasn't the only difference between them the cooler, clock speed, and power delivery?

 

Didn't the nanos never reach 1000mhz without bumping up the voltage though? They usually only reached around 900mhz, so wouldn't the actual tdp still be higher?

 

Yes, but AMD also only used the MOST power efficient chips off the production line to be used in the nano. I feel they will do the same thing to the Fury X2. a good fiji chip might need 250W but a bad one (still in spec) may need 300W for the same task.

 

It is like how Lamborghini saves all best aventador engines for the SV model.

[Sinder5]

W-O-A-H I've been hearing a few things about this, this is the first picture I've seen of it.  I can't wait to see a LTT Review of this if it happens............  IF.

[Sakkura]

W-O-A-H I've been hearing a few things about this, this is the first picture I've seen of it.  I can't wait to see a LTT Review of this if it happens............  IF.

 

Pretty sure it will happen, Lisa Su was running around with a PCB in her hand months ago. That's where that pic is from.

[patrickjp93]

Pretty sure it will happen, Lisa Su was running around with a PCB in her hand months ago. That's where that pic is from.

It's totally fake, definitely held together by wood screws.

[OneShotMan]

I'm tempted, but dat 4GB per die doe. Arctic Islands is going to be a MUCH better die and will have MUCH better software-hardware integration. Oh, and HDMI 2.0. Don't forget about that. What about us folks, with cheap Korean 4K panels?

[don_svetlio]

I'm tempted, but dat 4GB per die doe. Arctic Islands is going to be a MUCH better die and will have MUCH better software-hardware integration. Oh, and HDMI 2.0. Don't forget about that. What about us folks, with cheap Korean 4K panels?

Fury X was designed for 4K - look at the memory usage graphs posted - when 980 Ti needs 5GB the Fury X needs around 3.8 @ 1440p

[Dabombinable]

Fury X was designed for 4K - look at the memory usage graphs posted - when 980 Ti needs 5GB the Fury X needs around 3.8 @ 1440p

Dat Celeron A/Pentium II cache effect.

[don_svetlio]

Dat Celeron A/Pentium II cache effect.

Still no stuttering

[Dabombinable]

Still no stuttering

Think about it. The Pentium II had twice the cache of the Celeron A, but since the Celeron A had faster on die cache, the performance clock for clock was the same.

[OneShotMan]

Fury X was designed for 4K - look at the memory usage graphs posted - when 980 Ti needs 5GB the Fury X needs around 3.8 @ 1440p

 

You're not wrong, but 8GB should be the bare minimum for a flagship GPU atm. And I would like those 8GB to HBM.

[don_svetlio]

You're not wrong, but 8GB should be the bare minimum for a flagship GPU atm. And I would like those 8GB to HBM.

$$$ please

[Dabombinable]

You're not wrong, but 8GB should be the bare minimum for a flagship GPU atm. And I would like those 8GB to HBM.

8GB isn't actually needed though....at least for 4K.

[OneShotMan]

8GB isn't actually needed though....at least for 4K.

 

You're not wrong about 4K with current games. But VR is a different story. And I just feel that 8GB is the sweet-spot for a 2-3 year upgrade cycle.

[Prysin]

It's totally fake, definitely held together by wood screws.

mhm and this shiny squares in the middle.... those are actually just mirrors that has been engraved with some text

 

the "PCB" is only cardboard with lacquor on it

[don_svetlio]

mhm and this shiny squares in the middle.... those are actually just mirrors that has been engraved with some text

 

the "PCB" is only cardboard with lacquor on it

[Dabombinable]

You mean "Trik'Stari"

[Fetzie]

Yay - it's TDP (and likely power draw) will be on par with a GTX 480

 

Except you have two chips on the card, and each of them performs something like three times as well as a 480. So that'd make its performance per watt 6 times better?

[ZestyJalapeno]

[SuperCookie78]

To be honest I thought this would have launched on Black Friday or maybe it will launche on Monday.

[EmeraldKiwi]

I'm wondering if this will be stock water cooled like the 295x2 was.

[SuperCookie78]

8GB isn't actually needed though....at least for 4K.

my 980. Goes over it 4gb. Frame buffer in war thunder everything maxed in 900p! 8gb is needed for games at 4K

[Guest Generallee]

Great, I was wanting to buy a heater for my house but seems like this card has me covered. #TeamRed

 

Dank maymay friend but you're a bit late