AMD: Pirate Islands (R* 3** series) Speculation/Rumor Thread

Actually R9 270 was already with 1x 6pin connector
That isn't really compatible with these news:
http://wccftech.com/xfx-radeon-r9-3...gddr5-vram-launches-april-2015/#ixzz3U4TxXZhL
I can't see how they'd lower Pitcairn's power consumption by at least 30% without somehow changing the chip. PCB optimization would only get you so far.
They could have just changed the formula for TBP (and is the mentioned TDP even same as TBP in this case?)
R9 270's TBP is 150W, but actual average consumption is around 110W and peak (gaming, not furmark) around 120W (TPU's Asus R9 270 review)
For example R9 285's TBP (190W) and R9 280X's TBP (250W) are clearly measured on different formulas already
 
I guess if they sell those Pitcairn cards in the $100-$120 range, they might sell well.
If they go for $150, I think most people will prefer to pay a bit more for a GTX 960, which performs a lot better.
 
Just noticed something - could they be using different chips in OEM and retail markets?
The drivers mention "R9 360" - the 260-series was R9 only for OEMs, on retail it was R7, which could indicate the new entries are for OEM models

very late edit:
Someone raised a good point on another forum
Assuming that 370 is indeed Pitcairn, then the line-up from there on up seems impossible

Option 1:
380(X) = Grenada (Hawaii rebrand)
390(X) = Fiji
Problems: They didn't create Tonga just to release 1 partly cut GPU from it (R9 285), huge performance difference from 370 to 380

Option 2:
380(X) = Tonga
390(X) = Fiji
Problems: Huge performace difference from 380 to 390

Option 3:
380(X) = Grenada (new chip)
390(X) = Fiji
Problems: They didn't create Tonga just to release 1 partly cut GPU from it (R9 285), huge performance difference from 370 to 380 or from 380 to 390

Of course there's always the possibility for 375/385 but I doubt they'd go there on launch of new line-up
 
Last edited:
370 -> Pitcairn-in-disguise
370X -> Tonga XT
380 -> Hawaii-Pro-in-disguise
380 -> Hawaii-XT-in-disguise
390 -> Fiji Pro
390 -> Fiji XT
 
370 -> Pitcairn-in-disguise
370X -> Tonga XT
380 -> Hawaii-Pro-in-disguise
380 -> Hawaii-XT-in-disguise
390 -> Fiji Pro
390 -> Fiji XT

Based on the XFX-leak, if we assume it was true, 370 consumes far too little to be Pitcairn XT, if it's Pitcairn Pro, we'd be looking at around 45%+ performance difference between 370 and 370X, surely that can't be right either? (I'm assuming here Tonga XT would be at least 5% faster than Tonga Pro is). Even if it was Pitcairn XT the performance difference would be around 25-30%+ depending on Pitcairn XT variant
 
Based on the XFX-leak, if we assume it was true, 370 consumes far too little to be Pitcairn XT, if it's Pitcairn Pro, we'd be looking at around 45%+ performance difference between 370 and 370X, surely that can't be right either? (I'm assuming here Tonga XT would be at least 5% faster than Tonga Pro is). Even if it was Pitcairn XT the performance difference would be around 25-30%+ depending on Pitcairn XT variant

The PCB length from this leak also suggests a much smaller card than Pitcairn http://videocardz.com/55051/xfx-radeon-r9-370-core-edition-leaks-out-coming-early-april

Pitcairn could be 8" or more in PCB length, while this shows a reference design of just over 6". Meaning a mostly new lineup with Freesync across the board is possible. Clock Tonga up to 1ghz and it will generally beat out the 960 reliably in every game, heck give it 4gb of ram and charge $250 for it, why not? No doubt that will be a huge bottleneck that a 960 can't overcome for some games.

Then take Hawaii's basic design and update it with the new GCN design and ISA, cut the bus size 256bit, clock the ram to 7ghz with framebuffer compression, and charge $399 for it. It'll be cheaper to make than original Hawaii thanks to the cut bus, and with that extra geometry engine and possibly a higher clock from better TDP for perf it'll outperform a GTX 970 reliably, especially on frame delta time as it won't have any slow memory pool to choke bandwidth.
 
The PCB length from this leak also suggests a much smaller card than Pitcairn http://videocardz.com/55051/xfx-radeon-r9-370-core-edition-leaks-out-coming-early-april

Pitcairn could be 8" or more in PCB length, while this shows a reference design of just over 6". Meaning a mostly new lineup with Freesync across the board is possible. Clock Tonga up to 1ghz and it will generally beat out the 960 reliably in every game, heck give it 4gb of ram and charge $250 for it, why not? No doubt that will be a huge bottleneck that a 960 can't overcome for some games.

Then take Hawaii's basic design and update it with the new GCN design and ISA, cut the bus size 256bit, clock the ram to 7ghz with framebuffer compression, and charge $399 for it. It'll be cheaper to make than original Hawaii thanks to the cut bus, and with that extra geometry engine and possibly a higher clock from better TDP for perf it'll outperform a GTX 970 reliably, especially on frame delta time as it won't have any slow memory pool to choke bandwidth.

You can't just add 7ghz memory to any memory controller and expect it to handle it. Typically the faster the memory the memory controller can handle, the larger it is. The bus size also affects it, but overall you wouldn't save that much room. You just would make the PCB a tad less complex but this would be offset with the increase in the price in ram.

If we look at hawaii's memory controller vs Tahitis, hawaii's was 20% smaller while being a 512bit bus vs the 384 bit of tahitis. However Hawaiis mc can really only handle 5ghz memory vs the 6ghz memory tahitis can.

http://www.anandtech.com/show/7457/the-radeon-r9-290x-review/3

Thus adding a 7ghz capable 256bit bus wouldn't save all that much die size. Hawaii's is already efficient and doesn't take up that much space. I wouldn't doubt if hawaiis memory controller was actually smaller than gm204's. Look at the area they take on on diagrams and they are probably pretty similar in size.
 
If we look at hawaii's memory controller vs Tahitis, hawaii's was 20% smaller while being a 512bit bus vs the 384 bit of tahitis. However Hawaiis mc can really only handle 5ghz memory vs the 6ghz memory tahitis can.
Although not disputing the point of your post, the 8GB 290X models seem to all clock their memory at 5.5Ghz and allow overclocking up to ~6Ghz, so better memory dies or more mature Hawaii dies have allowed more leeway.
e.g. http://www.techwarelabs.com/powercolor-pcs-r9-290x-8gb-review/4/
 
Although not disputing the point of your post, the 8GB 290X models seem to all clock their memory at 5.5Ghz and allow overclocking up to ~6Ghz, so better memory dies or more mature Hawaii dies have allowed more leeway.
e.g. http://www.techwarelabs.com/powercolor-pcs-r9-290x-8gb-review/4/

Indeed, and it's been a year and a half+, redesigning a memory controller to handle higher frequencies is hardly out of the question. Especially as it's doubtful they'll just rebrand the big and expensive Hawaii GPU and re-release it without doing anything else.
 
Addendum, bus doesn't just take up silicon, which Hawaii's is admittedly very good there, but quite a bit of TDP and heat as well. Both are problems with Hawaii. And shrinking physical size of a bus isn't going to make it more power efficient, and almost certainly makes it worse in terms of heat dissipation, concentrating more work into even less of an area.
 
Last edited:
Unbelievable that a 300 series GPU will (potentially) support neither TrueAudio or FreeSync. It's seems AMD simply doesn't give a damn about properly supporting those technologies (if true).

Worse still is that they are re-branding the >3 year old Pitcairn and leaving it in the same performance segment (x70) as it's original version!! A re-brand I can accept if it at least falls down a performance segment with each re-brand. So by now Pitcairn should be named the 360(x). I really can't believe they're still billing it as a performance mid range gaming GPU (if true).
 
It's still silly season, so I'm still going to wait before making any accusatory statements. However, I will also state that I'd already planned to leave the AMD graphics regime for my next equipment refresh. AMD was once the value leader for price vs performance, but in a larger sense, NVIDIA is providing more and more value at a price that is only marginally higher than AMD's performance-equivalent device. This is all opinion, of course.

I will probably drop some coin when the "980ti 6GB" starts seeing some discounts...
 
How feasible would be Pitcairn updated with GCN 1.2 (or even 1.3) blocks? Same for the rest of the supposed rebrands (except Tonga)
 
It wouldn't be Pitcairn and so it wouldn't be a rebrand. Changing the components would go through the process of implementing a new chip, so it would be a different derivative that happened to have a few unit counts that were the same.
 
Rebrands are old news, more importantly can AMD cut down the power consumption and be competitive at the same time?

Power efficiency is an oft-used negative against the large-die Hawaii chips, but I've been playing with powertune settings and Furmark recently as an experiment to fit a "hot and noisy" AMD card into an SFF with limited cooling.
Actually, I stand by an earlier post I made that says I think AMD pushed Hawaii silicon too far.
With both GPU-Z and Furmark able to report power consumptions, I can see a 100W reduction in power consumption on 290X cards for as little as 5% performance loss.
If you have a Hawaii card, I urge you to crank power limits down in the overdrive tab of CCC and see what the resulting clockspeed is under full load. Even in a worst-case scenario, I'm seeing a typical clockspeed of 850MHz with the slider all the way to the left at -50%
That means that Hawaii (the two samples I personally own, at least) can run at 850+MHz on only 145W (half the 290W TDP). As mentioned, that's a worst-case scenario using a power-virus like Furmark. Under real gaming situations (I was messing around with Alien Isolation on 1440p ultra settings) the clocks averaged about 925MHz yet my PC was inaudible; Fans that normally hum along at 55% were barely spinning at 30% during my gameplay.
As Nvidia has proved, you can make a 28nm chip run efficiently. I think the design of Hawaii holds up very well under vastly reduced power constraints - AMD just pushed it outside its comfort zone in order to get the most out of it.
In saying that, the "underpowered" 290X is around the same performance as my GTX970 and also the same cost - significantly higher than a GTX960 4GB. I don't know if die-harvested 290 cards deal with power limit caps like the cherry-picked 290X cards.

http://techreport.com/news/27996/4gb-gtx-960s-trickle-into-retail-channels?post=893388#893388
 
It wouldn't be Pitcairn and so it wouldn't be a rebrand. Changing the components would go through the process of implementing a new chip, so it would be a different derivative that happened to have a few unit counts that were the same.
Obviously, I meant how feasible is that they'd just update the GPU with newest tech and leave it as it is otherwise
 
Back
Top