Support NeoGAF

[Sectorseven]

That Hawaii PR trip seems awfully expensive for a rebrand.

 

[DieH@rd]

No 20nm = No new arch for both nVidia and AMD.

No, this is brand new architecture for AMD. Nobody wans to go down to 20nm because TMSC and GLOFLO cant make high-powered products at that range. If they use same building materials, 20nm will use same or more power than 28nm.

That Hawaii PR trip seems awfully expensive for a rebrand.

2 highest cards wont be re-brands.

 

[chaosblade]

No, this is brand new architecture for AMD. Nobody wans to go down to 20nm because TMSC and GLOFLO cant make high-powered products at that range. If they use same building materials, 20nm will use same or more power than 28nm.



2 highest cards wont be re-brands.

I'd hope none of them would be (except maybe the low end stuff). Since I'm probably upgrading now I was thinking about the 280/X, but if it's just a 7970GHz with a new sticker I might have to rethink everything.

 

[Irobot82]

This is typical of tick tock methods anyhow right?

New GCN, then refresh?

Kepler, then refresh.

 

[RSTEIN]

770 is about 5% better than 680 w.r.t pixel/texel rate. Memory bandwidth is almost 20% higher. Plus 7xx gets various newfangled Nvidia tech. But, yes, basically it's just a bit of a bump over 680. A great value though, and it comes in a 4gb variant.

 

[Irobot82]

770 is about 5% better than 680 w.r.t pixel/texel rate. Memory bandwidth is almost 20% higher. Plus 7xx gets various newfangled Nvidia tech. But, yes, basically it's just a bit of a bump over 680. A great value though, and it comes in a 4gb variant.

If you were to clock the 680 at the exact same rates as the 770, does that still hold true?

 

[ethomaz]

This is typical of tick tock methods anyhow right?

New GCN, then refresh?

Kepler, then refresh.

Intel do that.

nVidia and AMD for GPUs always shows a new arch with die shrink... like Maxwell in 20nm next year.

I can't be wrong but that's like the GPU market works.

 

[chaosblade]

This is typical of tick tock methods anyhow right?

New GCN, then refresh?

Kepler, then refresh.

7000 series was the refresh, 6000 series introduced GCN. The rumor is that R9 290/X is new (whether that's GCN2 or just a bigger GCN die than the 7970, I don't know), while the other cards in the series are rebrands.

I don't really mind rebrands, especially on the lower end, but it's not cool when we're talking about cards that are already approaching 2 years old.

 

[Seanspeed]

I don't understand why you're laughing.

Is it not considered 'new' unless its totally new architecture?

EDIT: Ok jesus artist, you're one hell of an AMD fanboy, it seems. Will have to ignore you in the future.

 

[Irobot82]

7000 series was the refresh, 6000 series introduced GCN. The rumor is that R9 290/X is new (whether that's GCN2 or just a bigger GCN die than the 7970, I don't know), while the other cards in the series are rebrands.

I don't really mind rebrands, especially on the lower end, but it's not cool when we're talking about cards that are already approaching 2 years old.

6000 series were VLIW5 not GCN.

Is bolded confirmed? I'm not sure about that... I'm expecting a new Arch in 20nm next year... the new chip will be based in GCN I guess.

I was under the impression GCN 2.0 is a refresh, not a "new" architecture. Kind of like Bulldozer to Piledriver, to Steamroller, etc (they're all the same general architecture)

 

[ethomaz]

No, this is brand new architecture for AMD. Nobody wans to go down to 20nm because TMSC and GLOFLO cant make high-powered products at that range. If they use same building materials, 20nm will use same or more power than 28nm.

Is bolded confirmed? I'm not sure about that... I'm expecting a new Arch in 20nm next year... the new chip will be based in GCN I guess.

 

[kharma45]

Intel do that.

nVidia and AMD for GPUs always shows a new arch with die shrink... like Maxwell in 20nm next year.

I can't be wrong but that's like the GPU market works.

Last rumour on Maxwell was it won't be 20nm.

 

[artist]

7000 series was the refresh, 6000 series introduced GCN. The rumor is that R9 290/X is new (whether that's GCN2 or just a bigger GCN die than the 7970, I don't know), while the other cards in the series are rebrands.

I don't really mind rebrands, especially on the lower end, but it's not cool when we're talking about cards that are already approaching 2 years old.

GCN is 7000.

I don't understand why you're laughing.

Is it not considered 'new' unless its totally new architecture?

Because you find the possibility of a rebranded 7970 (as 280) disappointing yet when Nvidia does the exact same thing, it is "new". I was laughing at that double standard.

 

[chaosblade]

6000 series were VLIW5 not GCN.

Yeah, you're right. I guess these cards are just so damn old I'm getting them mixed up.

 

[Seanspeed]

Because you find the possibility of a rebranded 7970 (as 280) disappointing yet when Nvidia does the exact samething, it is "new". I was laughing at that double standard.

Just have different ideas of what 'rebranded' means, is all. No double standard.

 

[ethomaz]

Last rumour on Maxwell was it won't be 20nm.

Well I can't see this happening... I guess nVidia will continue to rebrand until 20nm gets ready.

Like I said I can be wrong too

 

[artist]

Just have different ideas of what 'rebranded' means, is all. No double standard.

So what constitutes "new" and how you would classify a rebrand. Lay it out ..

 

[Irobot82]

Can we assume then that a 280/280x can crossfire with a 7950 7970?

I have a 7950 and I'd be interested in that depending on cost. A 7950 right now is around $200. If the price isn't there then I'll wait for the R9-300 series to upgrade.

 

[RSTEIN]

If you were to clock the 680 at the exact same rates as the 770, does that still hold true?

I haven't seen a comparison. I'll try to find one. The 680 would need to add 40 MHz to its clock and 250 MHz to memory. ROPs, SPUs, and TAUs are the same for both so I'm guessing the performance gap would be almost completely erased by if the 680 could be OCed to match the 770.

 

[TouchMyBox]

6000 series were VLIW5 not GCN.

You're all wrong!

5000 was VLIW5

6000 was VLIW4

7000 was GCN

 

[Naminator]

Ok jesus artist, you're one hell of an AMD fanboy, it seems. Will have to ignore you in the future.

Took you long enough

 

[artist]

EDIT: Ok jesus artist, you're one hell of an AMD fanboy, it seems. Will have to ignore you in the future.

Your double standard is clear for everyone to see.

Took you long enough

If you think so.

 

[ethomaz]

You're all wrong!

5000 was VLIW5

6000 was VLIW4

7000 was GCN

This

 

[Laughing_Man__]

So what does this all have to do with them mentioning consoles?

 

[Seanspeed]

So what constitutes "new" and how you would classify a rebrand. Lay it out ..

Same exact product with a different label.

Now you're going to argue that is what it is, turning this pointless semantic discussion from 'rebrand' to what 'the same' is. And blah blah. I'm not interested, though. You seem to be pushing an agenda everywhere I see, which prevents you from seeing what I'm saying.

 

[chaosblade]

So what constitutes "new" and how you would classify a rebrand. Lay it out ..

Same die = rebrand, even if clocks or whatnot change.

770 = GK104 = rebrand
280X =? Tahiti (XT2) = rebrand

You're all wrong!

5000 was VLIW5

6000 was VLIW4

7000 was GCN

IIRC 6000 series used both depending on the subseries. But we've already seen how bad my memory is on this stuff.

 

[Irobot82]

So what does this all have to do with them mentioning consoles?

I'm guessing they have some kind of a program that makes it easy to translate code between consoles and PC's. It's all x86 architecture.... soo......

 

[kharma45]

Well I can't see this happening... I guess nVidia will continue to rebrand until 20nm gets ready.

Like I said I can be wrong too

Rumours point to Maxwell in Q2 next year and by the looks of it 20nm won't be ready in time.

Quoting from this

To summarize, Nvidia is probably going to make at least some 28nm Maxwell chips because they don’t know when they could launch 20nm Maxwell in large numbers, they don’t want to launch their new architecture over half a year after AMD launched theirs, and they would be caught in a position of having no speed bump against a massive 20%+ speed bump.  Maxwell also ensures that Nvidia won’t fall behind in data management, which is a large part of GCN 2.0.  Nvidia is launching 28nm to remain

http://videocardz.com/45417/why-maxwell-will-probably-launch-on-28nm-process

I cannot see Maxwell launching with anything other than 28nm.

 

[TouchMyBox]

IIRC 6000 series used both depending on the subseries. But we've already seen how bad my memory is on this stuff.

I'm just taking in terms of flagship chips,

Lower-end rebrands always muddy things.

 

[artist]

So what does this all have to do with them mentioning consoles?

http://neogaf.com/forum/showpost.php?p=83646209&postcount=274

Same exact product with a different label.

Now you're going to argue that is what it is, turning this pointless semantic discussion from 'rebrand' to what 'the same' is. And blah blah. I'm not interested, though. You seem to be pushing an agenda everywhere I see, which prevents you from seeing what I'm saying.

Do you know that the 280X is the same exact product as Tahiti? No. So yeah.

I'm judging Nvidia and AMD with the same ruler, unlike you.

Same die = rebrand, even if clocks or whatnot change.

770 = GK104 = rebrand
280X =? Tahiti (XT2) = rebrand

Woah dude, you are an AMD fanboy!!!1

 

[Irobot82]

Is Maxwell the chip where they want to do stacked memory or is it where they start putting ARM chips on it?

 

[kharma45]

Is Maxwell the chip where they want to do stacked memory or is it where they start putting ARM chips on it?

ARM ones. Volta is the stacked memory.

 

[Seanspeed]

Do you know that the 280X is the same exact product as Tahiti? No. So yeah.

I'm judging Nvidia and AMD with the same ruler, unlike you.

You're seriously dense, man.

 

[ethomaz]

Is Maxwell the chip where they want to do stacked memory or is it where they start putting ARM chips on it?

ARM I guess... Volta is staked memory.

edit -

 

[Irobot82]

ARM ones. Volta is the stacked memory.

That sounds awesome, I mean, it's almost like the GPU can itself be a full computer.

 

[velociraptor]

Isn't the 760 or 770 a 680?

The GTX 770 is an ever so slightly better GTX 680. But for all intents and purposes, they have practically the same benchmarks.

 

[JJD]

Awwww yissssss!

Cannot wait for AMD and MS reveal the dedicated Xbone Volcanic Island GPU!

5 teraflops!

Misterxmedia, parasite and insider redeemed!

Spoiler

LOL!

 

[TheGreyHulk]

770 isn't a new chip at all. It's the 680 with a better bios and faster ram. It's not a "re-worked" chip or a refresh. It's a rebadge + overclock (thanks to the better titan cooler) and faster ram.

these new AMD chips aren't the same GCN architecture. it's their next architecture.

 

[DieH@rd]

Is bolded confirmed? I'm not sure about that... I'm expecting a new Arch in 20nm next year... the new chip will be based in GCN I guess.

Yes, new architecture for 290/290X is GCN 2.0.

Intel always creates new arch on the same process node, and then shrinks that in the following year.

 

[artist]

You're seriously dense, man.

No double standards you mean. I guess all these people are dense too;

The GTX 770 is an ever so slightly better GTX 680. But for all intents and purposes, they have practically the same benchmarks.

770 isn't a new chip at all. It's the 680 with a better bios and faster ram. It's not a "re-worked" chip or a refresh. It's a rebadge + overclock (thanks to the better titan cooler) and faster ram.

these new AMD chips aren't the same GCN architecture. it's their next architecture.

 

[ethomaz]

Intel always creates new arch on the same process node, and then shrinks that in the following year.

Yeah... Intel uses this tick tock stuff :-D

 

[Irobot82]

Yeah... Intel uses this tick tock stuff :-D

I though AMD and nVidia both followed this similiar stuff too. I guess not.

Can anyone explain what unified virtual memory means?

 

[solarus]

Will these new GPUs make me regret the £200 MSI gtx 670 power edition OC I got a few weeks back? Most likely. Why didn't I wait.

 

[phosphor112]

First AMD "Steam Machine" to feature an APU. That's my guess.

 

[kharma45]

Will these new GPUs make me regret the £200 MSI gtx 670 power edition OC I got a few weeks back? Most likely. Why didn't I wait.

No, the 670 at that price is a steal.

 

[TheGreyHulk]

Will these new GPUs make me regret the £200 MSI gtx 670 power edition OC I got a few weeks back? Most likely. Why didn't I wait.

nope. the enthusiast GPUs are mainly geared at people gaming at 1440p and beyond. 670 is still fine for 1080p games out there unless you really need the highest performance possible.

 

[Seanspeed]

No double standards you mean. I guess all these people are dense too;

Dude, it was just a difference of what was meant by the word 'rebrand'. Its really not difficult.

 

[TheGreyHulk]

AMD vs Nvidia continues.

I'm so glad AMD is in it for the long haul for enthusiast GPU's and didn't bow out like the did for enthusiast CPU's.

Competition is good. It brings the best out of both companies and helps with prices.

 

[RSTEIN]

Competition is good. It brings the best out of both companies and helps with prices.

Yes. What's also interesting is that AMD has aligned themselves with the consoles this gen whereas Nvidia is throwing it's weight behind steambox. Even more battle lines to be drawn!

 

[shadowsdarknes]

I laughed. But this made me wonder, what happens to his blog once it's revealed that it's nothing? Does he convince himself that it's time activated? That when the Xbox really needs it it'll light their darkest hour like the Matrix of Leadership?

He has tried to buy himself a year claiming that the Xbox One won't run away from the PS4 in terms of performance until at least a year after launch because Microsoft wasn't ready to give away the awesome sauce to developers so soon.