FUD: AMD Tonga GPU comes as Radeon R9 285

[MentalIlness]

http://fudzilla.com/home/item/35382-amd-tonga-gpu-comes-as-radeon-r9-285

In late August or early September

A fresh report suggest that we might see a first AMD Tonga GPU as soon as late August or early September and the first graphics card with the new GPU should be the Radeon R9 285.

The newest report coming from SweClockers confirms that AMD has finally decided on the name of the upcoming Tonga GPU based graphics card and it should launch as the Radeon R9 285. The report notes that the GPU behind the Radeon R9 285 will be the Tonga PRO, which also suggests that we might see a Tonga XT GPU at a later date, most likely as base for the Radeon R9 285X.

In case you missed the earlier report, Tonga GPU is pretty much the same GPU as the venerable Tahiti since it packs the same amount of stream processors, but will feature a narrower memory interface so the Radeon R9 285 should pack 1792 Stream Processors and 2GB of GDDR5 memory paired up with a 256-bit memory interface. Of course, AMD AIB partners will most likely launch R9 285 graphics card with 4GB of memory as well.

According to previous reports, Tonga GPU based graphics cards should eventually completely replace the Tahiti-based ones.
Our sources confirm that the name of upcoming Tonga PRO based graphics cards is pretty much carved in stone, although the boxes have not been printed yet. Graphics cards based on the Tonga GPU are being manufactured as we speak.

Unfortunately, the report from Sweclocker or our sources can not confirm the precise date of the launch or the final price but we will surely keep an eye out for more information.

 

[Techhog]

2GB standard is pretty bad. Also, this might not outperform 280X, which will make the naming confusing.

 

[AtenRa]

hahahaha same thing on both of them. 285 will not be faster than 280 in high resolutions :thumbsdown:

we really need 20-16nm asap

 

[SirPauly]

That's not necessarily true based on how efficient the arch may be

 

[tollingalong]

I thought the rumors said Tonga will be the mobile flagship for AMD. Either way as far as I can see this is a stab at Maxwell (ie power efficiency) and not raw performance.

 

[Blitzvogel]

hahahaha same thing on both of them. 285 will not be faster than 280 in high resolutions :thumbsdown:

we really need 20-16nm asap

It'll probably be just fine for 1080p, of which I'm in the market for a new card for, and Tonga looks like it could be a good value, much like Bonaire.

 

[SlickR12345]

Fake. All these articles are click bait. All reliable sources get the correct info about a week in advance and they only post it usually a day before the certain GPU/s are to be released, so they can claim mix up in time zones on their NDA agreement.

Occasionally there are some real leaks weeks in advance that have the correct info, but this is quite rare.

 

[f1sherman]

Fake. All these articles are click bait. All reliable sources get the correct info about a week in advance and they only post it usually a day before the certain GPU/s are to be released, so they can claim mix up in time zones on their NDA agreement.

Occasionally there are some real leaks weeks in advance that have the correct info, but this is quite rare.

Like during Kepler release. There has been an excellent, almost 1:1 Kepler leak, quite old too, but it got lost among all the the fake ones, and pretty much everyone called it BS

 

[Rickyyy369]

It wouldn't be a bad Tahiti replacement if it came with 4GB of vram standard. But the same amount of SP's with less vram.. and its positioned above the 280x. Flawless logic. Maybe if they can clock it substantially higher it might work.

 

[SPBHM]

2GB is so 2011

a 2GB card as a replacement for the 270x is OK I guess, but for the 280/280x is a strange choice.

 

[Cloudfire777]

I wonder how close to R9 290X the R9 285 will come.

 

[Techhog]

I wonder how close to R9 290X the R9 285 will come.

It probably won't even beat the R9 280X with those specs.

 

[ShintaiDK]

The ugly face of being stuck with 28nm keeps spreading :/

Remember some extra cache can do wonders as seen with Maxwell. So dont rule out yet that this cant be faster, cheaper and more efficient than 280. And 2GB isnt that bad. I still use a 2GB card problem free.

 

[Techhog]

AMD is in a very tight competitive position with NVIDIA. This forces them to be a bit more secretive before major launches from both companies happen. During the conference call that everyone is referring to, the AMD executives, including Lisa Su, hesitate and answer coyly. When they say 2014 is 28nm, 2015 is 20nm, and 2016 is FINFET, that could mean so many things. For instance, perhaps they are referring to average shipping product for the year. In this case, AMD is 28nm for 2014 as 90% of their shipped products will be on 28nm. They could also be referring to the products lifetime. Products that are meant to last through 2015 (including Tonga and Carizzo) will ship on 20nm. See how easy it is to be *honest* with what manufacturing process is being used? If I were an AMD executive, I would have said the exact same thing.... 28nm for 2014, 20nm for 2015.

They have to be honest at conference calls. Being misleading could get them in trouble.

If no 20nm parts come out this year, then there should be a new addition to your ignore list.

Too late.

 

[Leadbox]

They have to be honest at conference calls. Being misleading could get them in trouble.



Too late.

Timetables and schedules can change.

 

[Techhog]

Timetables and schedules can change.

I suppose. I still doubt it, though. At best, there might be a December paper launch.

 

[metalliax]

They have to be honest at conference calls. Being misleading could get them in trouble.

They were not misleading in any way, even if they release products on 20nm this year. They simply stated 28nm for 2014 and 20nm for 2015. What this means is up for interpretation. If they truly meant no 20nm products at all in 2014, they would have said we will start shipping 20nm in 2015 instead.

 

[tviceman]

AMD is in a very tight competitive position with NVIDIA. This forces them to be a bit more secretive before major launches from both companies happen. During the conference call that everyone is referring to, the AMD executives, including Lisa Su, hesitate and answer coyly. When they say 2014 is 28nm, 2015 is 20nm, and 2016 is FINFET, that could mean so many things. For instance, perhaps they are referring to average shipping product for the year. In this case, AMD is 28nm for 2014 as 90% of their shipped products will be on 28nm. They could also be referring to the products lifetime. Products that are meant to last through 2015 (including Tonga and Carizzo) will ship on 20nm. See how easy it is to be *honest* with what manufacturing process is being used? If I were an AMD executive, I would have said the exact same thing.... 28nm for 2014, 20nm for 2015.

There is so much to refute in your post that it is best to not tackle it. Common sense has left this thread.

 

[f1sherman]

When they say 2014 is 28nm, 2015 is 20nm, and 2016 is FINFET, that could mean so many things. For instance, perhaps they are referring to average shipping product for the year.

yeah we see it all the time
rolleyes
what's a little share holder meeting/quarterly conference call for a little averaging across the year response to a specific question

average shipping product for the year!!!

 

[bakalu]

I think 285 will be weaker than 280x because bandwidth 256bit ram

 

[tollingalong]

I dunno man something about his card just feels off to me. I'm wondering if AMD is launching this to just stab at Maxwell rather than bring out something worth buying. Looking at the specs of Maxwell it conserves power but it also has nicer caching than prev gen. I don't know if AMD is doing the same. Their caching isn't great on their CPU side so I dunno...

Either way time will tell.

 

[sze5003]

Why are they going to 2gb? I figure 3 as standard from now on or even more. Are these the 20nm cards that are suppose to be out soon?

 

[Keysplayr]

I dunno man something about his card just feels off to me. I'm wondering if AMD is launching this to just stab at Maxwell rather than bring out something worth buying. Looking at the specs of Maxwell it conserves power but it also has nicer caching than prev gen. I don't know if AMD is doing the same. Their caching isn't great on their CPU side so I dunno...

Either way time will tell.

AMD will likely do the same. Or at lease their "own" version of what Nvidia is doing. Usually how it goes. Or the other way around.

 

[ocre]

How could this be just a stab at maxell. Or amds own version of maxwell. It doesnt work like that. Gpus take yrs to develop. Nvidia has had maxwell coming for yrs now. It evolved out of kepler. There is no way amd seen the 750ti and had their version of maxwell coming to launch this quickly. It is completely impossible.

At best this is amds version of kepler. Even this would be tight but maybe they pulled it off. Amd might have tore into keplet and apply some of the stuff they liked into their own design.....but then i still dont know.

It could just be another chip pulled in from 20nm. Their mid ranged

 

[monstercameron]

We already saw gcn 1.1 with a lower power draw vs earlier gcn 1.0. It only makes sense for AMD to continue to produce faster more efficient parts, especially since they have been doing this for years now.