AMD Radeon Fury X 3DMark performance

[Notional]

Misinformation? Thats rich coming from you. The kid who hides behind "I'm in the industry and I have insider knowledge but I can't share it with you so you should take my word". You're part of a whole group on this forum with a very specific agenda and get very, very upset when others don't agree. 

 

You get awfully upset when people question you or don't agree with what you say. Chill. 

 

Just stop. You are one of the biggest fanboys in here. In Danish we have a saying: "Thief thinks everyone steals". Just because you see fanboyism everywhere, doesn't mean there is; instead is more like projection really.

 

Fury/Fiji has a revolutionary type of vram, a new type of package/die, which results in new form factors possible for the highest end cards. Everyone, who loves tech, should be excited about this. You would be, if it was NVidia branded.

 

Exactly my point. The 290X keeps in lock-step and now even beats the 780TI at plenty of things now that Nvidia's stopped optimizing drivers for Kepler as much. The 290X has fewer SPs and less bandwidth.

 

I think the problem there lies in 780ti's measly 3GB of vram, which we know is a limitation in new games, even in 1080p. But it also looks like NVidia has just abandoned their 700 series.

[Briggsy]

interesting, especially considering 3dmark firestrike isn't a favorable test for AMD cards (to date). We need to see some Valley results to get a feel for HBM's potential.

[Techhog]

The only thing we know till now is that the GTX980Ti and Titan X are good cards.

What we dont know yet, is what AMD is going to set against it.

plain simple

Would you shut up already? Using a smiling face at the end of this kind of post just makes you seem like a know-it-all asshat. It's also annoying how defensive you get with this stuff.

[MEC-777]

I'm neither impressed, nor disappointed with these results. Don't know how legit they are, so I'm gonna withhold judgement until game day. 

 

The lack of VRAM made this card DOA for me to begin with... Now I'm just deciding between the 980ti or Titan X for my next build... If CaseLabs ever builds and ships my case to me...

 

You know there most likely will be 8GB versions.

 

why are you saying that when 4gb of hbm on the fury x is matching the titan x with 12gb of gddr5?

8gb hbm for next gen is going to absolutely blow nvidia's equivalent out of the water at that rate, especially if the vram is a limitation (which it might be, atm)

 

 

yes, that's right
AMD needed HBM to reach the Titan X in performance; Titan X that uses good ol' GDDR5

not even taking into account that Titan X has 3 times (!!!) more VRAM available to it

 

LOL. Comments like these just make me facepalm. 

[patrickjp93]

Just stop. You are one of the biggest fanboys in here. In Danish we have a saying: "Thief thinks everyone steals". Just because you see fanboyism everywhere, doesn't mean there is; instead is more like projection really.

 

Fury/Fiji has a revolutionary type of vram, a new type of package/die, which results in new form factors possible for the highest end cards. Everyone, who loves tech, should be excited about this. You would be, if it was NVidia branded.

 

 

I think the problem there lies in 780ti's measly 3GB of vram, which we know is a limitation in new games, even in 1080p. But it also looks like NVidia has just abandoned their 700 series.

Games should not be using more than 2GB for 1080p, flat out. It's just bad programming and design. Loading every texture in the game into the frame buffer and not using even 20% of them at any time is just using up memory for no good reason.

[Sintezza]

Would you shut up already? Using a smiling face at the end of this kind of post just makes you seem like a know-it-all asshat. It's also annoying how defensive you get with this stuff.

 

No sorry, im not going to shut up for anyone.

 

/end of that discussion.

[Techhog]

I'm neither impressed, nor disappointed with these results. Don't know how legit they are, so I'm gonna withhold judgement until game day.

You know there most likely will be 8GB versions.

LOL. Comments like these just make me facepalm.

4GB is the limit for HBM1. Sorry. We saw the die, and there's no room for more stacks.

[zMeul]

lower power, less heat, smaller form factors and the ability to not have uncooled vram chips on the back of the card (Titan X looking at you)

yeah .... GTX 980Ti looking right back at you 

[Techhog]

No sorry, im not going to shut up for anyone.

/end of that discussion.

We also know that it's limited to 4GB. We saw the die, and there's no room for more stacks. AMD said that the only way to add more memory is to add more stacks. Thus, 4GB is confirmed. If you say otherwise... Well, I hope you never serve in a jury.

[patrickjp93]

 

 

 

LOL. Comments like these just make me facepalm. 

Because more RAM != more bandwidth, or the claim AMD had to reach for HBM to match Nvidia's performance? Both are idiotic.

[patrickjp93]

4GB is the limit for HBM1. Sorry. We saw the die, and there's no room for more stacks.

There's the dual-interposer method, and Hynix could potentially make a respin to double the density of the 2.5D dies.

[RammerBee]

Looks like the 390x will just be a 290x on steroids.. They better give it a low price.

Yes and No..the 8gb gddr5 version is basically what you say....but the 4gb hbm version....*cough*

[Briggsy]

I want to see it in action all ready! The unofficial results are solid for what it is and I can see why AMD is charging so much for it, plus with HBM is cutting edge so there that as well.

 

exactly. people forget that new types of Ram always start off a bit faster than the previous type (i.e. DDR3 vs DDR4), but take a few generations to really shine. If the price is high, its a premium we always pay to get in on new (bleeding/cutting edge) tech.

[MEC-777]

4GB is the limit for HBM1. Sorry. We saw the die, and there's no room for more stacks.

 

Well, if that's the case I'm curious to see how it handles 4k gaming. I wouldn't be surprised if it's demonstrated to not be a limitation. Because it has such high bandwidth, that, along with data compression optimizations, could [somewhat] negate the need for more Vram capacity.

 

AMD must know it can handle high resolution gaming just fine with only 4GB, or they would have made 8GB available (at least as an option) with HBM1.

 

Time will tell... 

[Altecice]

Just a reminder that because of GPU Boost 2.0 those Nvidia "stock clocks" are probably not stock.

Waiting for real world benchmarks. Still impressive though.

 

I noticed this as well. They are testing the 980Ti at 1000 clock when it boosts to 1250+ by default... Yet they have that down as the 980Ti overclocked? 

Not to mention you can get ref 980Ti's to boost up to 1450 with little effort. 

 

So yeah synthetics are nice and all but I want to see real world.

[zMeul]

You do see the inside 1000 point difference on a benchmark running at 1440p and 4K between a card with 12GB and a card with 4GB, right?

no, I don't! because this is synthetic benchmark that doesn't even touch the VRAM needs of 1440p and UHD resolutions - then we can talk

[Techhog]

There's the dual-interposer method, and Hynix could potentially make a respin to double the density of the 2.5D dies.

There's no evidence whatsoever that dual interposers can be used, and what Hynx can do in the future only matters in the future.

[RammerBee]

color me disappointed: all that bragging from AMD, and in the end HBM didn't do that much for it

That 390x is not the Hbm version the hbm version has 4gb... So wait

[patrickjp93]

There's no evidence whatsoever that dual interposers can be used, and what Hynx can do in the future only matters in the future.

Actually AMD already demonstrated it, but it cuts the bandwidth in half.

[zMeul]

There's no actual difference between the GK 110 in a 780TI and the one in a Tesla K40

double precision?!!?!?

[patrickjp93]

That 390x is not the Hbm version the hbm version has 4gb... So wait

We have the Fury X performance numbers, the card with HBM, sitting in those slides. Did you miss that?

[Techhog]

Well, if that's the case I'm curious to see how it handles 4k gaming. I wouldn't be surprised if it's demonstrated to not be a limitation. Because it has such high bandwidth, that, along with data compression optimizations, could [somewhat] negate the need for more Vram capacity.

AMD most know it can handle high resolution gaming just fine with only 4GB, or they would have made 8GB available (at least as an option) with HBM1.

Time will tell...

It wasn't AMD's choice. It was a limit of the technology and of physics.

[patrickjp93]

double precision?!!?!?

That's only a microcode change to disable it in the 780TI. It's not an architectural difference or die difference.

[Techhog]

Actually AMD already demonstrated it, but it cuts the bandwidth in half.

Well, that explains why they won't do it.

[othertomperson]

Well, if that's the case I'm curious to see how it handles 4k gaming. I wouldn't be surprised if it's demonstrated to not be a limitation. Because it has such high bandwidth, that, along with data compression optimizations, could [somewhat] negate the need for more Vram capacity.

 

AMD most know it can handle high resolution gaming just fine with only 4GB, or they would have made 8GB available (at least as an option) with HBM1.

 

Time will tell... 

 

Has anyone ever been bottlenecked by a 290X's vram bandwidth? Is that a thing that has literally ever happened? Because if not, then is halving the amount of vram of the 390X, but having it a little bit faster going to actually make any difference except remove 4K surround as an option on their top-tier product (something a few Titan Xs might actually be pretty sweet for if you can justify the price)