[Rumor (Various)] AMD R7/9 3xx / Fiji / Fury

[bryanW1995]

Maybe it is a lot and I'm considering sticking to 1080p and downsample instead. And my budget is that way because I don't want to purchase that many tvs.... I've purchase too many as of late and I don't want to invest a ton into a 4k TV right now when I know I'll upgrade in literally a handful of months. I'm just waiting for the 2015 Vizio p series lineup so a 2014 refurbished for 450 is a good stopgap.
Single card is all I need because I'm not on the latest games like you guys are. There's also a ton of performance to be gained waiting for a game to get updated and drivers to mature. I usually play a game 2-3 years after release to ensure I have every update/driver for that game.
I actually don't want to play games anywhere near day 1 because I want all of the options available to me. So 4k on a single fiji card is possible. For me anyway.

I just have to see the reviews but it looks like I may just downsample but my brother just told me he won't ship my TV for a couple of months so now leaning back towards the 50 inch Vizio p series refurbished to hold me over.

You wait 2-3 years to play a game just to ensure that you have a few % better experience? Dude, you might be ok with just waiting for skylake, I hear that they are going to have 7850* level of performance...



* = I totally made that part up.

That's the trouble with these adapters. It is labeled a dual link DVI output adapter, but only up to 1600p @ 60hz. It's only partly dual link.

No way would I buy a graphics card/monitor or tv combo with the intention of getting an adapter for it. It's enough of a pita to use your computer on a TV monitor already, even if you discount the potential for additional problems you get by using an adapter. Though, at least on a TV, you already have some lag...if you are stuck with a monitor with only dvi connections then may God Have Mercy On Your (gaming) Soul.

Has there been any information on when the nano will be out? Information on the price?

August I think and probably $499

August seems about the right time frame, but where did you get the $499 price?

 

[looncraz]

As great as this sounds I don't believe AMD had the budget to apply Tonga GCN 1.2 across the lineup, even if they wanted to.

Okay, I don't like spreading rumors, but I heard an interesting rumor from a close long-time friend - who has never led me wrong - that the rumors (yeah, rumors about rumors... this is getting deep into the piles of salt now) in his company, which works closely with AMD and nVidia, are that AMD is already working on designing a full line replacement, with HBM going much further down the product line.

That isn't the juicy part, though. Apparently, according to nVidia reps he talked to, AMD has some supply priority agreement with Hynix and they are, therefore, trying to capture as much of the supply as they can in an effort to prevent nVidia from getting Pascal to market in quantity :sneaky::ninja: And, as you might imagine, nVidia is livid

Of course, this is even more internal rumors... so...:hmm:

 

[bystander36]

1600p@ 60hz is the maximum spec for true dual link so the label is correct. those of us pushing dvi past that with overclocked Korean's are outside of spec and hence why even a low quality dvi cable can affect the ability to overclock them.

The problem with the label of Dual Link DVI-D, is a true Dual Link DVI-D can handle 120hz, these can't. I'm referring to 1080p 120hz.

 

[MrTeal]

I'm tired of this tonga/hawaii mentions. Think what a 3/4 or even 1/2 a Fiji would do in HTPC or laptops. 75W TDP with r9 285 or higher perf. I say bring it on!

... Considering how little we know about Fiji, why would you think that a 1/2 a Fiji wouldn't be essentially a full Tonga just with HBM? Most people are assuming that Fiji is GCN 1.2 as well, just with HBM.

HBM or no, a 75W GCN1.2 part isn't going to be faster than a 2k shader Tonga which would be a ~200W part.

 

[bryanW1995]

Has there been any information on when the nano will be out? Information on the price?

Okay, I don't like spreading rumors, but I heard an interesting rumor from a close long-time friend - who has never led me wrong - that the rumors (yeah, rumors about rumors... this is getting deep into the piles of salt now) in his company, which works closely with AMD and nVidia, are that AMD is already working on designing a full line replacement, with HBM going much further down the product line.

That isn't the juicy part, though. Apparently, according to nVidia reps he talked to, AMD has some supply priority agreement with Hynix and they are, therefore, trying to capture as much of the supply as they can in an effort to prevent nVidia from getting Pascal to market in quantity :sneaky::ninja: And, as you might imagine, nVidia is livid

Of course, this is even more internal rumors... so...:hmm:

OFC they're working on the replacement of the rest of the lineup, they need something to release next year when 14/16 gets, inevitably, "delayed". Nvidia already has their next flagship available today (highly oc'd 980ti) in case they need to bring it to market as a holdover in 2016. They could even steal a page from AMD and get ~ 5% more clock/clock perfromance out of it, bump clocks 20%, and call it the gtx 990.

And it's interesting to hear about AMD dirty tricks vs NV...I wonder if the red team forum police (rtfp) will now descend upon you with Righteous Anger and Fury-ous Vengeance now? They'll probably even try to say "it's different when our team does it" or some such.

Meh, hynix would be dumb to favor either camp, it's to their benefit to maintain cordial relations with both camps.

 

[3DVagabond]

That might be promising. If the press has samples of the r9 nano as well as fury x, maybe the nano is soon to be released. Maybe same day as fury x or at least before Fiji pro. Assuming they don't have fiji pro.

If it falls under the Fiji pro in price I can understand not publicizing the release date to let the 390x sell more.

I read Nano release is August. Fury (non X) in July.

 

[SolMiester]

snip

And it's interesting to hear about AMD dirty tricks vs NV...I wonder if the red team forum police (rtfp) will now descend upon you with Righteous Anger and Fury-ous Vengeance now? They'll probably even try to say "it's different when our team does it" or some such.

snip.

i LOL!...:biggrin:

 

[maddie]

The problem with the label of Dual Link DVI-D, is a true Dual Link DVI-D can handle 120hz, these can't. I'm referring to 1080p 120hz.

You do realize that 2560x1600 @60Hz is within about 1% to 1920x1080 @120Hz.

It is Dual Link DVI

 

[krumme]

OFC they're working on the replacement of the rest of the lineup, they need something to release next year when 14/16 gets, inevitably, "delayed". Nvidia already has their next flagship available today (highly oc'd 980ti) in case they need to bring it to market as a holdover in 2016. They could even steal a page from AMD and get ~ 5% more clock/clock perfromance out of it, bump clocks 20%, and call it the gtx 990.

And it's interesting to hear about AMD dirty tricks vs NV...I wonder if the red team forum police (rtfp) will now descend upon you with Righteous Anger and Fury-ous Vengeance now? They'll probably even try to say "it's different when our team does it" or some such.

Meh, hynix would be dumb to favor either camp, it's to their benefit to maintain cordial relations with both camps.

I dont know why you call that dirty tricks?
When you work with suppliers and develop something together you normally make sure your competitor dont get access to it.
I dont know what benefits amd gets here but surely they must get some.
I prefer nv developed tech like mantle (dx12) and hbm. Nv own custom cores was a fine attemp and i appriciate that very much taking risk instead of crapworks and the likes. Not worthy of their capacity and technology strength. They should be the ones developing hbm.

 

[3DVagabond]

Are 4k monitors so bad that you'd prefer to have 22-24ms lag on a tv??

In "PC mode" (4:4:4 color) it's more like 30ms+. Imagine if Gsync or Freesync had that type of latency. How big of a fail would that be considered? Why would anyone bother with such a substandard connection? A true gaming connection like Gsync/Freesync using DP is less than 3ms (+ screen response time), worse case.

 

[looncraz]

OFC they're working on the replacement of the rest of the lineup, they need something to release next year when 14/16 gets, inevitably, "delayed". Nvidia already has their next flagship available today (highly oc'd 980ti) in case they need to bring it to market as a holdover in 2016. They could even steal a page from AMD and get ~ 5% more clock/clock perfromance out of it, bump clocks 20%, and call it the gtx 990.

And it's interesting to hear about AMD dirty tricks vs NV...I wonder if the red team forum police (rtfp) will now descend upon you with Righteous Anger and Fury-ous Vengeance now? They'll probably even try to say "it's different when our team does it" or some such.

Meh, hynix would be dumb to favor either camp, it's to their benefit to maintain cordial relations with both camps.

My take on the whole thing is just that a couple of low-level nVidia guys think AMD might be trying something like that. Or, perhaps, they are simply confused as to how AMD beat them to HBM and are making uneducated assumptions. Typical water-cooler level stuff here, nothing to get riled about. I just like the idea... then again, AMD does have some exclusivity contracts with Hynix, but I've never heard of anything leaking over to HBM2 as well. That could be interesting (but, then, I don't think nVidia would jump on the bandwagon without some capacity assurances).

 

[bystander36]

You do realize that 2560x1600 @60Hz is within about 1% to 1920x1080 @120Hz.

It is Dual Link DVI

What does that have to do with it. It won't support 120hz, period. People thinking that it will won't get 120hz.

 

[CP5670]

I was going the other way for a long time, and had an old video card with only DVI and a 1920x1080 120hz monitor with only DP. I used one of those Atlona converter boxes that officially only supported 2560x1600 at 60hz but worked nicely with that monitor.

 

[JDG1980]

Okay, I don't like spreading rumors, but I heard an interesting rumor from a close long-time friend - who has never led me wrong - that the rumors (yeah, rumors about rumors... this is getting deep into the piles of salt now) in his company, which works closely with AMD and nVidia, are that AMD is already working on designing a full line replacement, with HBM going much further down the product line.

Given the lead time on GPUs, it would be very worrisome if AMD wasn't already well into development on Arctic Islands.

HBM saves power and enables smaller cards (and could be especially useful in laptops), so of course AMD will try to use it wherever it is economically feasible.

That isn't the juicy part, though. Apparently, according to nVidia reps he talked to, AMD has some supply priority agreement with Hynix and they are, therefore, trying to capture as much of the supply as they can in an effort to prevent nVidia from getting Pascal to market in quantity :sneaky::ninja: And, as you might imagine, nVidia is livid

Of course, this is even more internal rumors... so...:hmm:

Not sure how much credence to give a second-hand rumor, but this could explain a lot. It seems odd that AMD, strapped for cash as it is, would invest so much in R&D on HBM (which will be a public standard) without getting anything in return. If they agreed to be the guinea pigs for HBM when no one else wanted it, in exchange for supply priority on HBM2... well, that would be a payoff that might actually be worth it.

AMD needs to go to FinFET ASAP, and they need to beat Nvidia there. If they get a six-month lead, it could make a huge difference in terms of market share and profits.

(Note that AMD beat Nvidia to 28nm by 3 months.)

 

[Elixer]

Meh, hynix would be dumb to favor either camp, it's to their benefit to maintain cordial relations with both camps.

While hynix can sell to anyone, you do know that AMD was their partner in developing HBM right?
http://electroiq.com/blog/2013/12/amd-and-hynix-announce-joint-development-of-hbm-memory-stacks/

That pretty much means that AMD will be the first kid on the block with HBM and HBM 2.

 

[ocre]

Maybe it is a lot and I'm considering sticking to 1080p and downsample instead. And my budget is that way because I don't want to purchase that many tvs.... I've purchase too many as of late and I don't want to invest a ton into a 4k TV right now when I know I'll upgrade in literally a handful of months. I'm just waiting for the 2015 Vizio p series lineup so a 2014 refurbished for 450 is a good stopgap.
Single card is all I need because I'm not on the latest games like you guys are. There's also a ton of performance to be gained waiting for a game to get updated and drivers to mature. I usually play a game 2-3 years after release to ensure I have every update/driver for that game.
I actually don't want to play games anywhere near day 1 because I want all of the options available to me. So 4k on a single fiji card is possible. For me anyway.

I just have to see the reviews but it looks like I may just downsample but my brother just told me he won't ship my TV for a couple of months so now leaning back towards the 50 inch Vizio p series refurbished to hold me over.

I like DSR a lot. It is pretty neat. But just want to say, its not at all like real 4k.
The effect is cool but nothing like the real thing.

I don't even have 4k for my PC but a friend of mine has a 4k monitor. There is a big difference in down sampling and true 4k. I would even say massive.

I like higher frame rates, so I made the trade off that works for me. You get what you like, I ink everyone should. Just wanted to let you know, if you already didn't, true 4k is much different than DSR/VSR

 

[The Alias]

OFC they're working on the replacement of the rest of the lineup, they need something to release next year when 14/16 gets, inevitably, "delayed". Nvidia already has their next flagship available today (highly oc'd 980ti) in case they need to bring it to market as a holdover in 2016. They could even steal a page from AMD and get ~ 5% more clock/clock perfromance out of it, bump clocks 20%, and call it the gtx 990.

And it's interesting to hear about AMD dirty tricks vs NV...I wonder if the red team forum police (rtfp) will now descend upon you with Righteous Anger and Fury-ous Vengeance now? They'll probably even try to say "it's different when our team does it" or some such.

Meh, hynix would be dumb to favor either camp, it's to their benefit to maintain cordial relations with both camps.

Do note AMD had a hand in designing HBM so they would be in a position to do so. Also note Nvidia guys have been asking for AMD to do an Nvidia gameworks equal. Seems their gameworks is Nvidia's own hardware lol.

 

[3DVagabond]

Given the lead time on GPUs, it would be very worrisome if AMD wasn't already well into development on Arctic Islands.

HBM saves power and enables smaller cards (and could be especially useful in laptops), so of course AMD will try to use it wherever it is economically feasible.



Not sure how much credence to give a second-hand rumor, but this could explain a lot. It seems odd that AMD, strapped for cash as it is, would invest so much in R&D on HBM (which will be a public standard) without getting anything in return. If they agreed to be the guinea pigs for HBM when no one else wanted it, in exchange for supply priority on HBM2... well, that would be a payoff that might actually be worth it.

AMD needs to go to FinFET ASAP, and they need to beat Nvidia there. If they get a six-month lead, it could make a huge difference in terms of market share and profits.

(Note that AMD beat Nvidia to 28nm by 3 months.)

Where are you getting the assumption that AMD is some sort of guinea pig? They are co-developers.

 

[DeathReborn]

Isn't it all kind of moot with HBM being a JEDEC standard now? AMD may get a small amount of money per chip but I doubt it will be enough to pay any of AMD's bills. AMD may get HBM2 first or last, either way it's a certainty that AMD & NVIDIA will use it.

 

[3DVagabond]

Isn't it all kind of moot with HBM being a JEDEC standard now? AMD may get a small amount of money per chip but I doubt it will be enough to pay any of AMD's bills. AMD may get HBM2 first or last, either way it's a certainty that AMD & NVIDIA will use it.

Having the latest tech first can be very valuable.

 

[ocre]

Having the latest tech first can be very valuable.

Absolutely

 

[jpiniero]

August seems about the right time frame, but where did you get the $499 price?

$499 kind of fits into the rest of the lineup.

 

[xthetenth]

Isn't it all kind of moot with HBM being a JEDEC standard now? AMD may get a small amount of money per chip but I doubt it will be enough to pay any of AMD's bills. AMD may get HBM2 first or last, either way it's a certainty that AMD & NVIDIA will use it.

Ahh, but if they buy up all of it and use it, then others can't. For example, the guy who made Softsoap bought multiple years' production of the right pumps for the soap dispensers so the product couldn't be duplicated until it established itself.

 

[Arachnotronic]

Ahh, but if they buy up all of it and use it, then others can't. For example, the guy who made Softsoap bought multiple years' production of the right pumps for the soap dispensers so the product couldn't be duplicated until it established itself.

Do you really think AMD has that kind of supply chain clout?

 

[BryanC]

Do you really think AMD has that kind of supply chain clout?

With 25% of the discrete GPU market, this seems unlikely to me.