Wait, so I guess this means Fiji won't be called the 390/390X? Will AMD skip to the 400 series for their next lineup? Or will they go with a new naming convention?
The 285 outperforms the 280x on 1080p results but falls behind when it comes to higher resolutions due to the lack of vram. If they bump it up to 3 or 4GB it should be a decent performance improvement.
"Outperforms" is damning with faint praise in this case though. ~+10% performance and ~-10% power use up against a 3 year old architecture (HD7970) is not much to write home about. Especially since it's arguable that most of the power savings came from simply lopping off part of the memory bus, and failed to improve video decode power use. More realistically, how many OEMs are actually going to stick a ~250w card in their desktop products? This just seems like a card in search of a market.
Agreed - I was expecting this rebadge round to be really bad, but this still shocked me. They have an underperforming, uncompetitive expensive GPU, held back by low clock speeds (using high voltage nevertheless) being sold in a cut down configuration. Yet they do not even bother to improve it in any aspect but still give it a number upgrade.. *shaking head*
He's only saying that the retail 380 isn't the same thing as the OEM 380. Wreckage is partly justified in his comment in that this article is discussing the OEM products, but in a wider sense, the retail 380 is more likely to be a rebrand of the vanilla R9 290, and as such a completely different animal to the OEM 380.
I will also point out that there wasn't an OEM Tonga card before now (as far as I'm aware).
Now, the 370... pour your scorn on that one, it's time Pitcairn died a death.
Have to go 400 series I think at this point right? I mean you can't go from Tonga (circa 2011 280X level performance) at your 380 SKU to something that is supposed to be really fast at your 390 SKU, can you? Would be like Danny DeVito standing next to Arnold in Twins.
That means you are looking at the possibility some of these ASICs span across 5 different product stacks lol.
That's what I'm thinking, it's what usually happens when mobile or desktop dGPU's launch out of sync with the the next gen lineup, to appease the OEM's. So the only way to address this problem in naming convention is to skip a generation or change it entirely. I would be really surprised now if Fiji launches as the 390/390X, that wouldn't make any sense.
sad to see AMD rebranding so much as well, although these are OEM parts. I would hate to see a 380 consumer part that is just a r9 285 although that might happen :(.
960m/950m were technically launched inbetween generations as Maxwell Mark1, so they really fit fine either way. The difference is their performance still makes sense where they are slotted and Nvidia did produce new chips at the top of the stack with the 970m/980m based on Maxwell.
Yep exactly, usually you get rebrands as a "refresh" line-up during the same-gen, but in AMD's case they already did that 2x (8000 series and R9 200 series) and this will be the 3rd time and as we pointed out, they leave no room for Hawaii AND Fiji in that stack.
The worst possible scenario for AMD and their fanboys is that Fiji is just refined Hawaii. Ouch. The less likely scenario is that Fiji is a completely new ASIC that stands head and shoulders above the 380. More likely but even more awkward is like we said, they go to 400 series on the desktop, but that almost makes their OEM line-up look irrelevant right away.
I imagine some bright spark decided to have two 300 series, with the OEM ones labelled OEM to at least attempt to avoid confusion.
If both series were going to be closely aligned, at the very least it'd spell the end of Tahiti. I guess we find out in a few weeks. Prepared for disappointment.
Since when does Oland lack a video decoder? The chip always had one (though I guess it could be disabled to save on licensing costs or something), though of course it lacks thing like 4k decode. (AMD does have a chip without video decode, that would be the 64bit / 4 Rop / 5CU Hainan/Sun chip but this one is unsuitable for desktop as it lacks display outputs too.)
Sorry but you are 100% wrong about this. You can easily see this for instance in the open-source radeon graphics driver, and it is supported nowadays (http://www.phoronix.com/scan.php?page=news_item&am... (That aside, like I said, it may be possible it's not active in all chips but I don't think that's the case either). It does, however, not have video _encode_ but that's a fate it shares with all gcn 1.0 chips.
FWIW I think AMD is a bit confusing on their own site, as they only list "video codec engine" - this is however the encode block (vce). They do not list UVD at all nowadays, I guess take it for granted... Nevertheless, these chips support UVD, the R7 250/240 were sometimes recommended for HTPC builds...
Well, that's not very impressive. Sounds like all they're really doing is replacing the 290x with a new card and dropping out a few unnecessary versions. I'm sure they'll add some numbers for the retail cards but this is not starting out well.
Before anyone gets in a tizzy, these are OEM-branded cards sold strictly to the likes of HP or Dell, the big PC brands that get all the corporate orders, not even boutique PC vendors like CyberPower, etc.
Both AMD and nVidia routinely sell last-generation parts into the big-business OEM market and allow for them to be branded as OEM parts under the current-gen naming scheme. Its a shitty practice, IMO, but one both vendors are equally guitly of, and which has no bearing on what their retail-bound SKUs will be. This doesn't mean retail R9 380 and lesser cards will be rebrands (though, I do think we'll see Tonga at some level, and perhaps a tweaked Hawaii).
You should be careful if you buy a PC from HP or Dell, even as an individual, that you don't get one of these cards while expecting the retail SKUs, because it sometimes happens, but you don't have to worry about getting one of these last-gen parts from retail outlets once those launch.
Personally, I suspect a (possibly-tweaked) Hawaii in the retail R9 380/X slot, and a full-fat Tonga in the R9 370/X slot, and maybe Bonaire at the very low levels (R5) -- there's hole where Pitcairn used to sit with ~1024 shaders, but I don't think they'll run it again with GCN 1.0. What they need is something like a half-Tonga with 1280 or 1024 shaders and a 128-bit or 192-bit GDDR5 bus, to slot into the R7 360/X spot.
I am trying to find where either AMD or NVIDIA has done what you are implying. I haven't pored through the parts lists thoroughly, but what I found was that for all the NVIDIA parts I checked (desktop parts in the 600, 700, 900 series) the OEMs parts seems to be of the same generation as the retail parts in the same series. AMD did have 7000 series OEM parts that were from the previous generation as retail 7000 series parts (perhaps they did it later than that too, but I didn't look too thoroughly), but they were still differentiated by different model numbers! So not only didn't I see NVIDIA do what you claim at all, but for AMD to be in line with what they have done in the past it seems that the retail parts should be named something like "R9 385", "R9 375", etc. if these new retail cards will be some new generation, as you seem to be claiming. If you have any specific examples of what you are talking about, please include them.
Different model numbers meaning some arcane string, generally irrelevant to the consumer and mostly hidden, that's tacked on to the regular model number -- as when Lenovo sent out S-10 netbooks to reviewers and then sold S-10 netbooks to consumers with glossy screens to save a few pennies in manufacturing cost?
Rebadging should be illegal.
Selling GPUs with different specs but with the same main naming number (e.g. two different "8400 GS" gpus) should be illegal.
Taking "OEM" or an arcane, usually hidden part number on does not qualify as adequately clear product naming practice.
Stop making excuses for what is clearly meant to be legal fraud.
and matte for the reviewed S-10s. Matte netbooks were rare and reviews specifically cited the matte screen as being a reason to choose that product. Selling glossy models with the same name but with one of three or four arcane numbers tacked on and mostly hidden away is bait and switch. It does not require undue effort to sell an S-10M and an S-10G. Or, better yet, an "S-10 Glossy".
Where am I making excuses? My post had nothing to do with the ethics of rebranding, it only was to refute ravyne's claim that what AMD seems to be doing with the 300 series so far was done before by both NVIDIA and AMD. I think AMD is really pushing the envelope of rebranding here.
That being said I disagree with your assertion that it's fraud and that it should be made illegal. You're going off the deep end there. To begin with, how on earth can you complain about a company updating a product line with a new chip (as in the 8400 GS)? If Ford makes a Mustang GT they can never increase the horsepower and use the same name? Ridiculous. We don't need the government to step in and tell companies how they can and cannot brand their products. They aren't making any false claims. They may be trying to take advantage of people's poor assumptions but so what? Next thing you'll want to make it illegal to paint your house before you put it on the market to sell it. People don't buy a newly painted house for more money because they really are in the market for a newly painted house, but rather because they have the mistaken impression that the house is somehow in better condition simply because it looks shinier.
You are going on the assumption that the retail cards will be 300 series as well. They will probably NOT be, they will probably be 400 series. They wont make 2 cards with the same model number, one being retail and one being OEM, with different specs. When they said 'they do this all the time' they were referring to rebadging in general, and having OEM only 'generations' (nVidia 100, 300 series, AMD 8000 series, etc)
So when will these 400 series retail cards based on a new architecture come out? If you mean in 2016 then you are probably right. Whatever comes out this year is likely to be some sort of GCN rebadge or minor architectural change. So if AMD comes out with a new brand of retail cards in the next few months, it is irrelevant to the point of my post whether they are called 300 series or 400 series. But don't say that *I* am assuming anything, I am only replying to what the OP said: "This doesn't mean retail R9 380 and lesser cards will be rebrands (though, I do think we'll see Tonga at some level, and perhaps a tweaked Hawaii)."
So the only new part will be R9 390/390X depending on if AMD/ATI released 1 or 2 cards. Or maybe 390X will be the new chip and 390 just a rebadged 290X. After all, what to do with 290X if 380 is 285?
BTW, I am very very confused with AMD's nomenclature
I'm guessing then the 390 would be a 290 on a new, more optimized PCB with 6-8 GB HBM. Overall improving power efficiency over its predecessor, in performance the X variant would probably fall between the 980 and the Titan X.
For the power saving that HBM can bring, even 4gb is quite acceptable as long as the 390/390X duo can effectively find the ideal performance/dollar sweet spot to topple the 980/980Ti. Although since the ability of DX12 to use unmirrored VRAM in multi GPU setups doesn't help anything pre-DX12, which is basically everything that is actually available right now, I sincerely hope they can find a way to push past this 4gb limit for their flagship models.
This is a disaster for AMD. Brand new cards that won't support Freesync, and believing the 285 rebranded straight to a 380 is in a competitive position is absolute lunacy.
There are rumours that the Pirate Islands chips will be the 400 series, consisting of Tobago (GCN1.1), Trinidad (GCN1.0), Grenada (GCN1.1) and Fiji (GCN1.2) GPUs. It's unclear what GCN1.0 parts would come across but I'm at least expecting the top end Tahiti as the 470X.
I hope to not see Pitcairn. Curacao has been rumoured but current implementations are too small/cut down.
For me it seems that AMD is not even trying to be competitive anymore.
GM206 is pretty much on par with this cut-down Tonga in performance with 100w less TDP or something like that and the age old Maxwell Gen 1 beats comfortably all of the other cards.
The street prices are mostly just a reflection on how "good" these chips / cards are and manufacturers set the price depending on their competitiveness. (goodness being measured by the perception of the card by potential customers). Nvidia is expensive, because they can get away with it while AMD offerings are so weak.
I guess this is down to lot of facts likeTSMC problems hurting AMD more than Nvidia. NV has plenty of resources to spend so they were able GM200 series on 28nm while AMD probably had to concentrate their more limited rnd-effortss to get products ready for when the "big thing" comes with next process upgrade.
Actually it will be the third, Remember the 8860 OEM and R7 265 and R7 265X OEM and R9 270 and R9 270X are called Curacao, And the 8870 OEM and 7870 is called Pitcarin, But they are the same exact chip. So this will be allot of releases of the same GPU. AMD is becoming like a psychotic merry go round!
So, I've been saying for a while I don't think AMD have two different Fiji Silicons in development. I also don't think anything but 390/390x will come out with HBM, so the options are:
380 is a rebrand of 290 (looks more likely now for the retail part, with OEM being a 285 rebrand) Fiji has both a HBM interface and a GDDR5 memory controller, wasting die space on both the 390 and the 380, and, as we're at the limits of 28nm because people's obsession with mobile devices means there is no sub 28nm high power process, means Fiji will be gimped.
This is incredible disappointing. I was hoping for a 380/380X that was a cut down of the 390/390X but still the new architecture. Regardless, get a move on it AMD. You were supposed to show us the 390/390X in March. And even last year.
Why the hell are people all in a tizzy about this? Rebranding for OEMs has been going on for forever now. The OEMs demand it. They want higher numbers for their new products whether there's anything new or not. Nvidia is no better.
And this has SHIT ALL to do with the retail parts.
Yep like i said earlier Nvidia rebranded their mobile line like crazy and ever have a single card with a drr3 and gddr5 config that will probably not be presented to customers. The ddr3 is probably because an OEM wanted to be able to make a cheaper part.
I think the reason for concern is...people were expecting new chips to alleviate the fear these were all going to be rebrands as those leaked INFs indicated months ago. Now that its confirmed these are rebrands, the bigger concern is that their mid to low-end Tonga is designated so high as the x80 part. That leaves no room for a Hawaii based rebrand AND their new Fiji chip under the current product stack and naming conventions unless AMD plans to confuse the market even more by using mismatched numbers for their OEM and desktop line-up....
Only when they don't have new products from that generation. For the most part, OEM series and desktop discrete will feature chips from the same family, the reason for alarm is because we are expecting desktop to be launched imminently, so you can imagine, the kinds of problems it would cause on the OEM market if AMD tried to distance itself from this 300 series OEM stack by launching a 400 series in a few weeks. Instantly obsoleted.
Disappointment overall, old parts, new stock number, hope for R390. Meanings is seen in new Super computer Summit coming in few years and Nvidia part. maybe buy Nvidia. perhaps computers are at last shot of wad.
Since these are OEM cards their still could be hope! But I do not see the point in this OEM series. It is the exact same as the last series! How the hell could their possibly be any performance gain? I understand re-branding if you make a last gen card a lower next gen model, But when you do it more than once! And you come out with one new chip that is just pointless! But hopefully AMD can hang in! Maybe this will be a good thing, the chip in the R9 290 is still a strong chip and good enough to run ultra settings at 1080p and 1440p, So if people can buy this chip as a $180 price point midrange GPU this will be great! But I have a bad felling they will put the 380 stamp on it and sale it at a price over $250! But if AMD is really re-branding again, Then they need to offer a strong GPU like a re-branded R9 290 at a price point of $180. Besides! Nobody is going to buy any 300 series GPU's knowing they are next gen priced but old gen chips!
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
74 Comments
Back to Article
dragonsqrrl - Wednesday, May 6, 2015 - link
Wait, so I guess this means Fiji won't be called the 390/390X? Will AMD skip to the 400 series for their next lineup? Or will they go with a new naming convention?Wreckage - Wednesday, May 6, 2015 - link
So 380 = 285??? That has to be the worst rebadge ever.CaptainDoug - Wednesday, May 6, 2015 - link
The 285 outperforms the 280x on 1080p results but falls behind when it comes to higher resolutions due to the lack of vram. If they bump it up to 3 or 4GB it should be a decent performance improvement.takeship - Wednesday, May 6, 2015 - link
"Outperforms" is damning with faint praise in this case though. ~+10% performance and ~-10% power use up against a 3 year old architecture (HD7970) is not much to write home about. Especially since it's arguable that most of the power savings came from simply lopping off part of the memory bus, and failed to improve video decode power use. More realistically, how many OEMs are actually going to stick a ~250w card in their desktop products? This just seems like a card in search of a market.ET - Thursday, May 7, 2015 - link
You must have missed the 370 = 265 above. If there's a worst rebadge award, surely it must go to it.MrSpadge - Thursday, May 7, 2015 - link
Agreed - I was expecting this rebadge round to be really bad, but this still shocked me. They have an underperforming, uncompetitive expensive GPU, held back by low clock speeds (using high voltage nevertheless) being sold in a cut down configuration. Yet they do not even bother to improve it in any aspect but still give it a number upgrade.. *shaking head*Creig - Thursday, May 7, 2015 - link
Wrong. The R9 380 OEM is a variant of the R9 285. We don't know the specs of the RETAIL R9 380 yet.chizow - Thursday, May 7, 2015 - link
So much defense of all these rebrands Creig, is this good for the consumer, in your opinion? :)silverblue - Friday, May 15, 2015 - link
He's only saying that the retail 380 isn't the same thing as the OEM 380. Wreckage is partly justified in his comment in that this article is discussing the OEM products, but in a wider sense, the retail 380 is more likely to be a rebrand of the vanilla R9 290, and as such a completely different animal to the OEM 380.I will also point out that there wasn't an OEM Tonga card before now (as far as I'm aware).
Now, the 370... pour your scorn on that one, it's time Pitcairn died a death.
WinterCharm - Thursday, May 14, 2015 - link
It still doesn't match the price/performance I get from an nvidia card :/chizow - Wednesday, May 6, 2015 - link
Have to go 400 series I think at this point right? I mean you can't go from Tonga (circa 2011 280X level performance) at your 380 SKU to something that is supposed to be really fast at your 390 SKU, can you? Would be like Danny DeVito standing next to Arnold in Twins.That means you are looking at the possibility some of these ASICs span across 5 different product stacks lol.
7000, 8000 (OEM), 200, 300 (OEM), 400?
dragonsqrrl - Thursday, May 7, 2015 - link
That's what I'm thinking, it's what usually happens when mobile or desktop dGPU's launch out of sync with the the next gen lineup, to appease the OEM's. So the only way to address this problem in naming convention is to skip a generation or change it entirely. I would be really surprised now if Fiji launches as the 390/390X, that wouldn't make any sense.Crunchy005 - Thursday, May 7, 2015 - link
Ya, rebranding sucks. Nvidia did it with their 960m/950m cards as well.http://www.anandtech.com/show/9077/nvidia-launches...
sad to see AMD rebranding so much as well, although these are OEM parts. I would hate to see a 380 consumer part that is just a r9 285 although that might happen :(.
chizow - Thursday, May 7, 2015 - link
960m/950m were technically launched inbetween generations as Maxwell Mark1, so they really fit fine either way. The difference is their performance still makes sense where they are slotted and Nvidia did produce new chips at the top of the stack with the 970m/980m based on Maxwell.chizow - Thursday, May 7, 2015 - link
Yep exactly, usually you get rebrands as a "refresh" line-up during the same-gen, but in AMD's case they already did that 2x (8000 series and R9 200 series) and this will be the 3rd time and as we pointed out, they leave no room for Hawaii AND Fiji in that stack.The worst possible scenario for AMD and their fanboys is that Fiji is just refined Hawaii. Ouch. The less likely scenario is that Fiji is a completely new ASIC that stands head and shoulders above the 380. More likely but even more awkward is like we said, they go to 400 series on the desktop, but that almost makes their OEM line-up look irrelevant right away.
silverblue - Friday, May 15, 2015 - link
I imagine some bright spark decided to have two 300 series, with the OEM ones labelled OEM to at least attempt to avoid confusion.If both series were going to be closely aligned, at the very least it'd spell the end of Tahiti. I guess we find out in a few weeks. Prepared for disappointment.
Zefeh - Wednesday, May 6, 2015 - link
For those wondering - Fiji is the 390X and 390. Fiji is not the 380 or 380X.chizow - Wednesday, May 6, 2015 - link
Yeah we can see that, from this slide 380 = Tonga.MrSpadge - Thursday, May 7, 2015 - link
This leaves no naming space for the Hawaii rebadge, which is very strange.ImSpartacus - Thursday, May 7, 2015 - link
I suppose a 380x could be a Hawaii part, though I expect a fully enabled tonga somewhere (or maybe it's not as big as we thought) . It's weird.Alexvrb - Thursday, May 7, 2015 - link
These are OEM rebadges, just like the 8000 series. It should be reported as such: "These are OEM rebadges. You can ignore all of them."chizow - Saturday, May 9, 2015 - link
And what if the desktop line-up follows suit? We can ignore all of them too? No, not a fanboy at all, defend/deflect at all costs!silverblue - Friday, May 15, 2015 - link
What if?Nobody knows yet. Patience, grasshopper.
chizow - Wednesday, May 6, 2015 - link
News keeps getting worst by the minute. Those 2xTonga rumors might be real?ImSpartacus - Thursday, May 7, 2015 - link
Lord help us.Alexvrb - Wednesday, May 6, 2015 - link
Ugh OEM naming always mucks things up.chizow - Saturday, May 9, 2015 - link
Is the problem with the naming, more than the fact AMD chose to Rebadgeon their entire OEM line-up?mczak - Wednesday, May 6, 2015 - link
Since when does Oland lack a video decoder? The chip always had one (though I guess it could be disabled to save on licensing costs or something), though of course it lacks thing like 4k decode.(AMD does have a chip without video decode, that would be the 64bit / 4 Rop / 5CU Hainan/Sun chip but this one is unsuitable for desktop as it lacks display outputs too.)
Ryan Smith - Wednesday, May 6, 2015 - link
Oland has never had a video decoder. It's one of the things AMD left out to make it a lean chip.mczak - Thursday, May 7, 2015 - link
Sorry but you are 100% wrong about this. You can easily see this for instance in the open-source radeon graphics driver, and it is supported nowadays (http://www.phoronix.com/scan.php?page=news_item&am...(That aside, like I said, it may be possible it's not active in all chips but I don't think that's the case either).
It does, however, not have video _encode_ but that's a fate it shares with all gcn 1.0 chips.
mczak - Thursday, May 7, 2015 - link
FWIW I think AMD is a bit confusing on their own site, as they only list "video codec engine" - this is however the encode block (vce). They do not list UVD at all nowadays, I guess take it for granted... Nevertheless, these chips support UVD, the R7 250/240 were sometimes recommended for HTPC builds...Clauzii - Thursday, May 7, 2015 - link
I can recommend the Sapphire R7 Ultimate any day for that purpose :)Flunk - Wednesday, May 6, 2015 - link
Well, that's not very impressive. Sounds like all they're really doing is replacing the 290x with a new card and dropping out a few unnecessary versions. I'm sure they'll add some numbers for the retail cards but this is not starting out well.ravyne - Wednesday, May 6, 2015 - link
Before anyone gets in a tizzy, these are OEM-branded cards sold strictly to the likes of HP or Dell, the big PC brands that get all the corporate orders, not even boutique PC vendors like CyberPower, etc.Both AMD and nVidia routinely sell last-generation parts into the big-business OEM market and allow for them to be branded as OEM parts under the current-gen naming scheme. Its a shitty practice, IMO, but one both vendors are equally guitly of, and which has no bearing on what their retail-bound SKUs will be. This doesn't mean retail R9 380 and lesser cards will be rebrands (though, I do think we'll see Tonga at some level, and perhaps a tweaked Hawaii).
You should be careful if you buy a PC from HP or Dell, even as an individual, that you don't get one of these cards while expecting the retail SKUs, because it sometimes happens, but you don't have to worry about getting one of these last-gen parts from retail outlets once those launch.
ravyne - Wednesday, May 6, 2015 - link
Personally, I suspect a (possibly-tweaked) Hawaii in the retail R9 380/X slot, and a full-fat Tonga in the R9 370/X slot, and maybe Bonaire at the very low levels (R5) -- there's hole where Pitcairn used to sit with ~1024 shaders, but I don't think they'll run it again with GCN 1.0. What they need is something like a half-Tonga with 1280 or 1024 shaders and a 128-bit or 192-bit GDDR5 bus, to slot into the R7 360/X spot.Yojimbo - Wednesday, May 6, 2015 - link
I am trying to find where either AMD or NVIDIA has done what you are implying. I haven't pored through the parts lists thoroughly, but what I found was that for all the NVIDIA parts I checked (desktop parts in the 600, 700, 900 series) the OEMs parts seems to be of the same generation as the retail parts in the same series. AMD did have 7000 series OEM parts that were from the previous generation as retail 7000 series parts (perhaps they did it later than that too, but I didn't look too thoroughly), but they were still differentiated by different model numbers! So not only didn't I see NVIDIA do what you claim at all, but for AMD to be in line with what they have done in the past it seems that the retail parts should be named something like "R9 385", "R9 375", etc. if these new retail cards will be some new generation, as you seem to be claiming. If you have any specific examples of what you are talking about, please include them.Oxford Guy - Thursday, May 7, 2015 - link
Different model numbers meaning some arcane string, generally irrelevant to the consumer and mostly hidden, that's tacked on to the regular model number -- as when Lenovo sent out S-10 netbooks to reviewers and then sold S-10 netbooks to consumers with glossy screens to save a few pennies in manufacturing cost?Rebadging should be illegal.
Selling GPUs with different specs but with the same main naming number (e.g. two different "8400 GS" gpus) should be illegal.
Taking "OEM" or an arcane, usually hidden part number on does not qualify as adequately clear product naming practice.
Stop making excuses for what is clearly meant to be legal fraud.
Oxford Guy - Thursday, May 7, 2015 - link
tacking, not takingand matte for the reviewed S-10s. Matte netbooks were rare and reviews specifically cited the matte screen as being a reason to choose that product. Selling glossy models with the same name but with one of three or four arcane numbers tacked on and mostly hidden away is bait and switch. It does not require undue effort to sell an S-10M and an S-10G. Or, better yet, an "S-10 Glossy".
Yojimbo - Thursday, May 7, 2015 - link
Where am I making excuses? My post had nothing to do with the ethics of rebranding, it only was to refute ravyne's claim that what AMD seems to be doing with the 300 series so far was done before by both NVIDIA and AMD. I think AMD is really pushing the envelope of rebranding here.That being said I disagree with your assertion that it's fraud and that it should be made illegal. You're going off the deep end there. To begin with, how on earth can you complain about a company updating a product line with a new chip (as in the 8400 GS)? If Ford makes a Mustang GT they can never increase the horsepower and use the same name? Ridiculous. We don't need the government to step in and tell companies how they can and cannot brand their products. They aren't making any false claims. They may be trying to take advantage of people's poor assumptions but so what? Next thing you'll want to make it illegal to paint your house before you put it on the market to sell it. People don't buy a newly painted house for more money because they really are in the market for a newly painted house, but rather because they have the mistaken impression that the house is somehow in better condition simply because it looks shinier.
extide - Thursday, May 7, 2015 - link
You are going on the assumption that the retail cards will be 300 series as well. They will probably NOT be, they will probably be 400 series. They wont make 2 cards with the same model number, one being retail and one being OEM, with different specs. When they said 'they do this all the time' they were referring to rebadging in general, and having OEM only 'generations' (nVidia 100, 300 series, AMD 8000 series, etc)Yojimbo - Thursday, May 7, 2015 - link
So when will these 400 series retail cards based on a new architecture come out? If you mean in 2016 then you are probably right. Whatever comes out this year is likely to be some sort of GCN rebadge or minor architectural change. So if AMD comes out with a new brand of retail cards in the next few months, it is irrelevant to the point of my post whether they are called 300 series or 400 series. But don't say that *I* am assuming anything, I am only replying to what the OP said: "This doesn't mean retail R9 380 and lesser cards will be rebrands (though, I do think we'll see Tonga at some level, and perhaps a tweaked Hawaii)."Oxford Guy - Thursday, May 7, 2015 - link
Your attempt at a rebuttal concerning the 8400 GS fails utterly.Yojimbo - Thursday, May 7, 2015 - link
Your attempt to reply in any relevant or meaningful way to any point I made fails utterly.Oxford Guy - Friday, May 8, 2015 - link
The truth is what I posted.Yojimbo - Thursday, May 7, 2015 - link
Although I don't think you were really making such an attempt.Peichen - Thursday, May 7, 2015 - link
So the only new part will be R9 390/390X depending on if AMD/ATI released 1 or 2 cards. Or maybe 390X will be the new chip and 390 just a rebadged 290X. After all, what to do with 290X if 380 is 285?BTW, I am very very confused with AMD's nomenclature
D. Lister - Thursday, May 7, 2015 - link
I'm guessing then the 390 would be a 290 on a new, more optimized PCB with 6-8 GB HBM. Overall improving power efficiency over its predecessor, in performance the X variant would probably fall between the 980 and the Titan X.ImSpartacus - Thursday, May 7, 2015 - link
We're only getting 4gb of hbm per card in these until designs.The "special sauce" that allows AMD to exceed that has been rumored to simply be a multi gpu setup.
xthetenth - Thursday, May 7, 2015 - link
It's perfectly possible and makes a ton more sense to just have multiple stacks on an interposer like every sane rumor indicates.D. Lister - Thursday, May 7, 2015 - link
For the power saving that HBM can bring, even 4gb is quite acceptable as long as the 390/390X duo can effectively find the ideal performance/dollar sweet spot to topple the 980/980Ti. Although since the ability of DX12 to use unmirrored VRAM in multi GPU setups doesn't help anything pre-DX12, which is basically everything that is actually available right now, I sincerely hope they can find a way to push past this 4gb limit for their flagship models.piroroadkill - Thursday, May 7, 2015 - link
This is a disaster for AMD. Brand new cards that won't support Freesync, and believing the 285 rebranded straight to a 380 is in a competitive position is absolute lunacy.Oxford Guy - Thursday, May 7, 2015 - link
The people who buy low-end OEM stuff probably are not in the target market.ET - Thursday, May 7, 2015 - link
I'm very disappointed by GCN 1.0 being carried to yet another generation.ImSpartacus - Thursday, May 7, 2015 - link
Yeah, pitcairn is a champ, but it's not THAT good.silverblue - Friday, May 15, 2015 - link
There are rumours that the Pirate Islands chips will be the 400 series, consisting of Tobago (GCN1.1), Trinidad (GCN1.0), Grenada (GCN1.1) and Fiji (GCN1.2) GPUs. It's unclear what GCN1.0 parts would come across but I'm at least expecting the top end Tahiti as the 470X.I hope to not see Pitcairn. Curacao has been rumoured but current implementations are too small/cut down.
zepi - Thursday, May 7, 2015 - link
For me it seems that AMD is not even trying to be competitive anymore.GM206 is pretty much on par with this cut-down Tonga in performance with 100w less TDP or something like that and the age old Maxwell Gen 1 beats comfortably all of the other cards.
The street prices are mostly just a reflection on how "good" these chips / cards are and manufacturers set the price depending on their competitiveness. (goodness being measured by the perception of the card by potential customers). Nvidia is expensive, because they can get away with it while AMD offerings are so weak.
I guess this is down to lot of facts likeTSMC problems hurting AMD more than Nvidia. NV has plenty of resources to spend so they were able GM200 series on 28nm while AMD probably had to concentrate their more limited rnd-effortss to get products ready for when the "big thing" comes with next process upgrade.
Torashin - Thursday, May 7, 2015 - link
I really hope you're wrong about them rebadging Pitcairn again... that would be truly disgraceful.silverblue - Friday, May 15, 2015 - link
This would only be the second retail series with a Pitcairn part, but agreed, it needs to go.P39Airacobra - Monday, June 1, 2015 - link
Actually it will be the third, Remember the 8860 OEM and R7 265 and R7 265X OEM and R9 270 and R9 270X are called Curacao, And the 8870 OEM and 7870 is called Pitcarin, But they are the same exact chip. So this will be allot of releases of the same GPU. AMD is becoming like a psychotic merry go round!rtho782 - Thursday, May 7, 2015 - link
So, I've been saying for a while I don't think AMD have two different Fiji Silicons in development. I also don't think anything but 390/390x will come out with HBM, so the options are:380 is a rebrand of 290 (looks more likely now for the retail part, with OEM being a 285 rebrand)
Fiji has both a HBM interface and a GDDR5 memory controller, wasting die space on both the 390 and the 380, and, as we're at the limits of 28nm because people's obsession with mobile devices means there is no sub 28nm high power process, means Fiji will be gimped.
greenfyah - Thursday, May 7, 2015 - link
Why people are already hating these cards without even give them a try?Alexey291 - Thursday, May 7, 2015 - link
Because they are literally the same cards we've seen for (in case of pitcairn) for almost half a decade now.greenfyah - Thursday, May 7, 2015 - link
Is possible that the 390x could be more powerfull than the 980 or even more than the Titanx?Oxford Guy - Thursday, May 7, 2015 - link
Pitcairn rebadges are rather irrelevant to enthusiasts. People want to see a competitor for 970 SLI and the 980.Shadowmaster625 - Thursday, May 7, 2015 - link
Why does Oland lack a video decoder?FITCamaro - Thursday, May 7, 2015 - link
This is incredible disappointing. I was hoping for a 380/380X that was a cut down of the 390/390X but still the new architecture. Regardless, get a move on it AMD. You were supposed to show us the 390/390X in March. And even last year.You're only hurting yourselves.
kyuu - Thursday, May 7, 2015 - link
Why the hell are people all in a tizzy about this? Rebranding for OEMs has been going on for forever now. The OEMs demand it. They want higher numbers for their new products whether there's anything new or not. Nvidia is no better.And this has SHIT ALL to do with the retail parts.
Crunchy005 - Thursday, May 7, 2015 - link
Yep like i said earlier Nvidia rebranded their mobile line like crazy and ever have a single card with a drr3 and gddr5 config that will probably not be presented to customers. The ddr3 is probably because an OEM wanted to be able to make a cheaper part.http://www.anandtech.com/show/9077/nvidia-launches...
Oxford Guy - Thursday, May 7, 2015 - link
Human sacrifice, witch burning, lip plates, foot binding, head hunting...Don't get yourself into a tizzy.
chizow - Thursday, May 7, 2015 - link
I think the reason for concern is...people were expecting new chips to alleviate the fear these were all going to be rebrands as those leaked INFs indicated months ago. Now that its confirmed these are rebrands, the bigger concern is that their mid to low-end Tonga is designated so high as the x80 part. That leaves no room for a Hawaii based rebrand AND their new Fiji chip under the current product stack and naming conventions unless AMD plans to confuse the market even more by using mismatched numbers for their OEM and desktop line-up....kyuu - Thursday, May 7, 2015 - link
I could be wrong but aren't OEM and retail desktop lineups already mismatched in terms of specs in previous generations as well?chizow - Thursday, May 7, 2015 - link
Only when they don't have new products from that generation. For the most part, OEM series and desktop discrete will feature chips from the same family, the reason for alarm is because we are expecting desktop to be launched imminently, so you can imagine, the kinds of problems it would cause on the OEM market if AMD tried to distance itself from this 300 series OEM stack by launching a 400 series in a few weeks. Instantly obsoleted.thomasxstewart - Sunday, May 10, 2015 - link
Disappointment overall, old parts, new stock number, hope for R390. Meanings is seen in new Super computer Summit coming in few years and Nvidia part. maybe buy Nvidia. perhaps computers are at last shot of wad.drashek 4k, 4k,4k....
P39Airacobra - Monday, June 1, 2015 - link
Since these are OEM cards their still could be hope! But I do not see the point in this OEM series. It is the exact same as the last series! How the hell could their possibly be any performance gain? I understand re-branding if you make a last gen card a lower next gen model, But when you do it more than once! And you come out with one new chip that is just pointless! But hopefully AMD can hang in! Maybe this will be a good thing, the chip in the R9 290 is still a strong chip and good enough to run ultra settings at 1080p and 1440p, So if people can buy this chip as a $180 price point midrange GPU this will be great! But I have a bad felling they will put the 380 stamp on it and sale it at a price over $250! But if AMD is really re-branding again, Then they need to offer a strong GPU like a re-branded R9 290 at a price point of $180. Besides! Nobody is going to buy any 300 series GPU's knowing they are next gen priced but old gen chips!