AMD "Fiji" Dual-GPU Graphics Card Delayed

[DeathtoGnomes]

Ship the damn card to EK ( ) and have them make a water block for it. Then attach a radiator with a built-in pump, add some decent length tubing, and....

 

[rtwjunkie]

In other words, we could launch it now but we need something available to counter Nvidia's Pascal launch.

Actually that would be a bad business tactic. Now, when it was slated to release is when it would have gotten the best sales. It's like opening weekend for the new Star Wars movie. What movie studio in their right mind would release a new movie the same weekend? Same principle here.

 

[EarthDog]

I don't think it is a valid counter to Pascal... Sure, performance wise it will likely beat it, however, as people mentioned earlier, its a dual GPU card, and its going to be $1K+ I would imagine. Perhaps there is a Pascal Titan, but outside of that, You are looking at $700 for a single card.

 

[Vayra86]

I think the article makes the right link to the Cooler Master patent infringement on Asetek's AIO solution. This may very well turn out to be the true reason and in all honesty, it is the only reason that really makes sense.

Postponing this card to counter Pascal makes no sense, Nvidia has no plans for any dual GPU solution and there hasn't been one since the GTX 690. AMD is betting on a market segment that is incredibly limited and Nvidia doesn't even care about it any longer, because why would they when you can pair two single cards just as well. Pascal won't be any different.

The other underlying reason could be that there are many recent game titles that lack proper Crossfire support or just are inherently unsuited for AFR because of engine limitations (Anandtech pointed this out). Launching a top end card with a high price tag that won't run recent big games well, is not going to work at all.

The VR- reasoning behind it is complete nonsense in my opinion. There simply is no market for this product release, and when it does get released there are better options on the horizon. The best time, honestly, was still now.

 

[GhostRyder]

Postponing this card to counter Pascal makes no sense, Nvidia has no plans for any dual GPU solution and there hasn't been one since the GTX 690. AMD is betting on a market segment that is incredibly limited and Nvidia doesn't even care about it any longer, because why would they when you can pair two single cards just as well. Pascal won't be any different.

The other underlying reason could be that there are many recent game titles that lack proper Crossfire support or just are inherently unsuited for AFR because of engine limitations (Anandtech pointed this out). Launching a top end card with a high price tag that won't run recent big games well, is not going to work at all.

First, Titan-Z was after GTX 690.

Second, like what games are you speaking of? Because as I look through the list of major game releases and such they all seem to have a working CFX profile.

Actually that would be a bad business tactic. Now, when it was slated to release is when it would have gotten the best sales. It's like opening weekend for the new Star Wars movie. What movie studio in their right mind would release a new movie the same weekend? Same principle here.

Yea, I highly doubt (Least I hope) the reason is competing with Pascal. I stand on the belief its related to the cooler so they are trying to find an alternative (Since they are even being hit on Fury X's cooler).

 

[$ReaPeR$]

damn it AMD!

 

[Nokiron]

Unless Lisa Su is holding a cardboard card in her hands, they have a product ready and they should launch it.

That's not how it works. When Nvidia presented Pascal back in March 2014 Jen held a Pascal-part in his hands.

It's not done if its a very early prototype.

 

[lilhasselhoffer]

Actually that would be a bad business tactic. Now, when it was slated to release is when it would have gotten the best sales. It's like opening weekend for the new Star Wars movie. What movie studio in their right mind would release a new movie the same weekend? Same principle here.

Didn't that Alvin and The Chipmunks movie release the exact same day as the Star Wars movie?

The point of releasing a card in the same exact time frame would be if it catered to a different audience. If Fiji x2 actually does VR appreciably better than Pascal we'd have two very separate markets being tapped at once. As little is concrete (due to actual benchmarks), we're assuming that Pascal is equally suited to VR and therefore direct competition. Best case, that is not the case.

I'm hoping that the VR bluster is somehow justified. While I personally believe it's BS, that's the only way Fiji x2 isn't going to be in a losing battle from the word go. That, or maybe the crap Nvidia stirred with Samsung will push Pascal off a little bit. AMD is long overdue for some kind of win...

 

[Dieinafire]

In testing they found that 3 out of 10 were catching on fire due to heat issues. #Bigproblems:-(

 

[deemon]

Around same time frame as Pascal? O boy.

Pascal? Maybe... but this is about to launch at the same time Arctic Islands is supposed to launch... so is AMD trying to shoot itself into the leg here? It's like Intel launching Broadwell and Skylake at the same time... kinda pointless.

 

[deemon]

The point of releasing a card in the same exact time frame would be if it catered to a different audience. If Fiji x2 actually does VR appreciably better than Pascal we'd have two very separate markets being tapped at once. As little is concrete (due to actual benchmarks), we're assuming that Pascal is equally suited to VR and therefore direct competition. Best case, that is not the case.

I'm hoping that the VR bluster is somehow justified. While I personally believe it's BS, that's the only way Fiji x2 isn't going to be in a losing battle from the word go. That, or maybe the crap Nvidia stirred with Samsung will push Pascal off a little bit. AMD is long overdue for some kind of win...

Question is, did they mess up something with Arctic Islands for VR, so they now have to launch some old crap in the same time "for VR"?

 

[arbiter]

Pascal? Maybe... but this is about to launch at the same time Arctic Islands is supposed to launch... so is AMD trying to shoot itself into the leg here? It's like Intel launching Broadwell and Skylake at the same time... kinda pointless.

Um no, arctic islands is not gonna launch at around that time, that is least q3. Rumor has Nvidia Pascal in april but Arctic islands is more still months after that. They only made a prototype around 1 month ago, nvidia had prototype for pascal 6 months ago.

 

[lilhasselhoffer]

Question is, did they mess up something with Arctic Islands for VR, so they now have to launch some old crap in the same time "for VR"?

I don't believe you are asking the right question. Please, indulge me for a moment.

Let's say that AMD does release Arctic Islands at the same time as Fiji x2. Arctic Islands based GPUs are currently all slated to be middle to high end consumer cards, designed to pump out either 1080p or at best 4k resolution pictures. These cards do well in that arena, but the software and hardware for VR is more demanding. The calculations for spatial relations have to be done, and there are a ton of them. Thus, you've got GPU based calculations. GPU calculations which favor massive stream processor counts.

Now, let's assume that Arctic Islands is built off of a 16nm process. Fiji is built off of a 28nm process. Fiji x2 will therefore support more processors, despite the much better processors per area of Arctic Islands.

The reasonable person asks why not just strap two Arctic Islands GPUs together and get better processor count and less power draw. To that, I ask what platform would allow this. The mainstream offering from Intel doesn't offer enough PCI-e lanes. The enthusiast offering makes the already expensive VR hardware all the more so. To answer this, AMD offers the Crossfire on a stick, so that VR is more affordable.


So, here's the best case. AMD is theoretically waiting on VR. They release Fiji x2 as a one card VR machine. Fiji sells because it's the best option for a budget minded VR experiment. Worst case AMD is lying and Fiji x2 competes directly with Pascal and Arctic Islands due to the delay. I choose positivity here, because AMD really needs wins here. Whether that's stupidity or realistic is up to you to decide.


Edit:
Let me make this clear, I understand Arctic Islands is much later in the year. The reason I've indulged in all three releasing at once is absolute worst case scenario (for AMD/Nvidia). It won't happen as such, but helps to illustrate why the situation isn't particularly apocalyptic.

 

[DeathtoGnomes]

So, the world ends if all three are released at the same time? WOOHOO!

 

[lilhasselhoffer]

So, the world ends if all three are released at the same time? WOOHOO!

Read some of the crap posted above. It sounds like this delay is going to be the sole cause of AMD going belly-up. That particular situation is apocalyptic to the consumer computer industry. When you deal with doomsayers, the apocalypse is something that is frequently harped upon.

 

[HisDivineOrder]

So, here's the best case. AMD is theoretically waiting on VR. They release Fiji x2 as a one card VR machine. Fiji sells because it's the best option for a budget minded VR experiment. Worst case AMD is lying and Fiji x2 competes directly with Pascal and Arctic Islands due to the delay. I choose positivity here, because AMD really needs wins here. Whether that's stupidity or realistic is up to you to decide.


Edit:
Let me make this clear, I understand Arctic Islands is much later in the year. The reason I've indulged in all three releasing at once is absolute worst case scenario (for AMD/Nvidia). It won't happen as such, but helps to illustrate why the situation isn't particularly apocalyptic.

Choosing positivity because logic and a glance over of the track record of the subject at hand suggests that historically there's reason to give them the benefit of the doubt regarding a subject... well, that's great. Choosing it because you'd rather not see what'll happen otherwise is not logic. It's ignorance disguised as optimism.

The problem with your lean toward positivity is that you act as though AMD doesn't have a long, storied history of delays, missed deadlines, stretching out developed products across multiple years to fill in huge holes in their lines that they have because they further missed even more deadlines.

To my eye, stretching the new theoretical high end into 2016 to compete with Pascal sounds like AMD to a tee. Remember when they used the Jedi Mind Trick on the press to tell everyone they NEVER promised an update for the 7970 and 7950 at the beginning of a year even after releasing an updated version of GCN for a single low end product that was meant to be the precursor of a whole other generation of product? One that they'd had on their roadmaps for over a year before that moment?

What happened there? They delayed that release for almost a year into the R9 290 and 290X along with a refresh of existing products. And if you look at the Fury X2, that's what they're doing again.

And that's not even bringing up the debacle that was the Rage Fury MAXX, which is the entire Fury line's namesake that no one seems to recall for the horror story it wound up being.

Choose what is logical. Optimism isn't logical here. Sorry. This delay is to make them seem relevant when nVidia shows up with Pascal everywhere and all they've got are more price drops, more 4GB Fiji cards, and probably Never Settle Like Ever bundles.

 

[lilhasselhoffer]

Choosing positivity because logic and a glance over of the track record of the subject at hand suggests that historically there's reason to give them the benefit of the doubt regarding a subject... well, that's great. Choosing it because you'd rather not see what'll happen otherwise is not logic. It's ignorance disguised as optimism.

The problem with your lean toward positivity is that you act as though AMD doesn't have a long, storied history of delays, missed deadlines, stretching out developed products across multiple years to fill in huge holes in their lines that they have because they further missed even more deadlines.

To my eye, stretching the new theoretical high end into 2016 to compete with Pascal sounds like AMD to a tee. Remember when they used the Jedi Mind Trick on the press to tell everyone they NEVER promised an update for the 7970 and 7950 at the beginning of a year even after releasing an updated version of GCN for a single low end product that was meant to be the precursor of a whole other generation of product? One that they'd had on their roadmaps for over a year before that moment?

What happened there? They delayed that release for almost a year into the R9 290 and 290X along with a refresh of existing products. And if you look at the Fury X2, that's what they're doing again.

And that's not even bringing up the debacle that was the Rage Fury MAXX, which is the entire Fury line's namesake that no one seems to recall for the horror story it wound up being.

Choose what is logical. Optimism isn't logical here. Sorry. This delay is to make them seem relevant when nVidia shows up with Pascal everywhere and all they've got are more price drops, more 4GB Fiji cards, and probably Never Settle Like Ever bundles.

Before this becomes a pissing match to see who remembers more history, and how shaded you want history to be, can we end it?


If not, I'm going to have to start with the 5xxx series of cards, which competed with the 4xx series from Nvidia (despite being released about 6 months later). I'm also going to have to ask you how a single card can compete with a line of GPUs. That seems rather silly, but I'll let you dig out justifications for it.

I'd also like you to justify how comparing the one flagship double GPU card to something that the average consumer might buy is reasonable. I can understand AMD wanting to say they have the best card (numerically) on the market, but it's one card. A single card isn't business success (as proven by Fury and a bunch of other high performing cards from both AMD and Nvidia).


You seem to want to deny any possible good here, and favor worst case without any facts. That's all fine and dandy for thought experiments, but what does it get us? We don't profit from believing in conspiranoya. We don't benefit from holding off on GPU purchases (read: this was not a consumer tuned card at the word go). What does predicting doom and gloom get? I understand it may be out of character, but I'd like to see this in a god light. I won't be surprised if it's not, if you read above you'd know that, but assuming this is shenanigans before ever seeing anything isn't productive.

Besides, somebody needs to explore the positive side. I may not love AMD, but the do deserve the benefit of the doubt. It seems like most people here can't be bothered to give them that.

 

[Ja.KooLit]

so this dual cards has PLX chips on them? I wonder the difference PLX-PCI vs this type of PLX (performance wise)

 

[EarthDog]

Same difference, slightly lower latency, but nothing that is noticed outside of measuring it.

 

[deemon]

Read some of the crap posted above. It sounds like this delay is going to be the sole cause of AMD going belly-up. That particular situation is apocalyptic to the consumer computer industry. When you deal with doomsayers, the apocalypse is something that is frequently harped upon.

read around the web. arctic islands is to launch at the end of Q2 2016.
http://wccftech.com/amd-greenland-14nm-production-q2-2016/

 

[arbiter]

read around the web. arctic islands is to launch at the end of Q2 2016.
http://wccftech.com/amd-greenland-14nm-production-q2-2016/

it says "enter mass production in q2. They don't say if its start of q2 or mid, so at earliest q3 will be launch not end of q2. they won't launch program at end of a quarter. As article said that is rumor so still a lot that could happen between now and then.

 

[rtwjunkie]

read around the web. arctic islands is to launch at the end of Q2 2016.
http://wccftech.com/amd-greenland-14nm-production-q2-2016/

Just like this card was going to launch now?

 

[DeathtoGnomes]

Read some of the crap posted above. It sounds like this delay is going to be the sole cause of AMD going belly-up. That particular situation is apocalyptic to the consumer computer industry. When you deal with doomsayers, the apocalypse is something that is frequently harped upon.

AMD has always been on the brink of destruction. If you live paycheck to paycheck, AMD can be said to be living new card to new card. Its nothing new to them and their 8-ball.

 

[Vayra86]

I don't believe you are asking the right question. Please, indulge me for a moment.

Let's say that AMD does release Arctic Islands at the same time as Fiji x2. Arctic Islands based GPUs are currently all slated to be middle to high end consumer cards, designed to pump out either 1080p or at best 4k resolution pictures. These cards do well in that arena, but the software and hardware for VR is more demanding. The calculations for spatial relations have to be done, and there are a ton of them. Thus, you've got GPU based calculations. GPU calculations which favor massive stream processor counts.

Now, let's assume that Arctic Islands is built off of a 16nm process. Fiji is built off of a 28nm process. Fiji x2 will therefore support more processors, despite the much better processors per area of Arctic Islands.

The reasonable person asks why not just strap two Arctic Islands GPUs together and get better processor count and less power draw. To that, I ask what platform would allow this. The mainstream offering from Intel doesn't offer enough PCI-e lanes. The enthusiast offering makes the already expensive VR hardware all the more so. To answer this, AMD offers the Crossfire on a stick, so that VR is more affordable.


So, here's the best case. AMD is theoretically waiting on VR. They release Fiji x2 as a one card VR machine. Fiji sells because it's the best option for a budget minded VR experiment. Worst case AMD is lying and Fiji x2 competes directly with Pascal and Arctic Islands due to the delay. I choose positivity here, because AMD really needs wins here. Whether that's stupidity or realistic is up to you to decide.


Edit:
Let me make this clear, I understand Arctic Islands is much later in the year. The reason I've indulged in all three releasing at once is absolute worst case scenario (for AMD/Nvidia). It won't happen as such, but helps to illustrate why the situation isn't particularly apocalyptic.

Good story with legitimate considerations, but it seems a little far fetched that 2x Fiji will be bandwidth saturation on PCI 3.0 x8... (If you draw a line across generations, you can see PCI bandwidth has never ever been saturated, not even at PCI 3.0 x4 on an R9 295X2) Which kinda throws the whole story in the gutter.

My two bitcoins say AMD just doesn't see a market now and they're right, there is none. I won't be surprised if the card won't ever see the light of day, or it will be just as much of a non-seller as the 295x2. Let's face it, GTX 690 didn't sell for shit either and it's the reason Nvidia abandoned the segment.

 

[deemon]

Let's face it, GTX 690 didn't sell for shit either and it's the reason Nvidia abandoned the segment.

mmmm... no they didn't, they renamed their segment into "titan" .... to make it sound cooler and sell more maybe? but is still ultimately pointless.