Wednesday, August 7th 2013

New AMD GPU Family Codenames "Volcanic Islands" and "Pirate Islands"

AMD's next generation GPU family that leverages upcoming silicon fab technologies to increase transistor counts, while maintaining or lowering thermal envelopes, is codenamed "Volcanic Islands," and we've known about that for some time now.

The centerpiece of "Volcanic Islands" family is "Hawaii," a high-end GPU that makes up top single- and dual-GPU SKUs; followed by "Maui" and "Tonga." Not much is known about these two. A dual-GPU product with two "Hawaii" chips is confusingly codenamed "New Zealand," which is already used to designate certain Radeon HD 7990 graphics cards. AMD is expected to debut its first "Volcanic Islands" GPUs in Q4-2013, when foundry partner TSMC's swanky new 20 nm node is expected to take flight.

When digging through change-logs of system information tool HWInfo, 3DCenter.org discovered what it hypothesizes to be a successor to "Volcanic Islands." Called "Pirate Islands," the GPU family contains chips codenamed after popular islands where sea pirates took shore leave; that's "Bermuda," "Fiji," and "Treasure Island."

AMD could also do away with the Radeon HD xxxx model number scheme, replacing it with something that looks like "Radeon R# xxxx." We have two theories on how something like that could be worked out. First, of course, is that "R#" could denote generation, "xxxx" the model number (eg: Radeon R9 1900), a simple replacement of the "HD" moniker; and second is that "R#" could denote market segment, and "xxxx" model number (eg: Radeon R9-170 for "Hawaii XT," R8-170 for "Maui XT," R7-170 for "Tonga XT," and R9-270 for "Bermuda XT," etc.)Source: 3DCenter.org
Add your own comment

33 Comments on New AMD GPU Family Codenames "Volcanic Islands" and "Pirate Islands"

#1
Jack1n
20nm? should be good.
Posted on Reply
#2
HumanSmoke
AMD is expected to debut its first "Volcanic Islands" GPUs in Q4-2013, when foundry partner TSMC's swanky new 20 nm node is expected to take flight
So when did TSMC's 20nm node get moved up for series production? TSMC's own symposium noted "Mass production for 20nm is planned for the first half of next year (H1 2014)"
by: btarunr
"R#" could denote market segment, and "xxxx" model number (eg: Radeon R9-170 for "Hawaii XT," R8-170 for "Maui XT," R7-170 for "Tonga XT," and R9-270 for "Bermuda XT," etc.)
R2-D2 presumably for a half-height card?
Posted on Reply
#4
cadaveca
My name is Dave
by: erocker
I'm missing the part where it says 20nm for first half of next year. Seriously, I can't find it.
Maybe here?:
"20nm planar HKMG [high-k metal gate] technology has already passed risk production with a very high yield and we are preparing for a very steep ramp in two GIGAFABs," Sun said.
Just under the first pic of dude's head.


Plus this:
20nm is production-ready now, with numerous EDA tools qualified and 185 design kits available. All foundation IP and some critical interface IP has passed qualification.
By the next head-pic down. :p

Qualification takes 3-6 months. That's assuming the designs for these GPUs are already pre-tested, but that date is pretty realisitic.

Once qualification testing is passed, designs go into mass production = 1H 2014. It is August, after all. Thing is, that article is form April, so really, we're looking at the end of September. Go figure, that's the date that's been passed around lately for AMD cards. I am pretty sure we also heard they shipped already to OEM partners, those units would be ES qualification units. That was just a short while ago, so if going by 6 months, 1H 2014 still makes reasonable sense. Add in that AMD might push qualification in 3 months, and September rings a bell again.


So what did we learn today, class?



I dunno.



:banghead:
Posted on Reply
#5
HumanSmoke
by: cadaveca
Once qualification testing is passed, designs go into mass production = 1H 2014. It is August, after all. Thing is, that article is form April, so really, we're looking at the end of September. Go figure, that's the date that's been passed around lately for AMD cards. I am pretty sure we also heard they shipped already to OEM partners, those units would be ES qualification units. That was just a short while ago, so if going by 6 months, 1H 2014 still makes reasonable sense. Add in that AMD might push qualification in 3 months, and September rings a bell again.
Sorry about that, I had the Cadence and SemiInd articles bookmarked for another forum thread. A more recent indicator would be TSMC's Q2 financials a couple of weeks ago:
20nm: Risk production has started and volume production starts Q1 2014.
Probably safe to assume that AMD (and everyone else using TSMC's 20nm) will have risk wafers going through the fabs- and I presume there will be what amounts to hand-building some cards for runtime/validation (Do AMD still have their Canadian workshop for hand-building cards?), but I'm picking that Apple might be at the head of the queue for actual production.
Posted on Reply
#6
Jorge
The smaller the node size the higher the transistor density and the more difficult to cool based on reduced surface area. There isn't much in life that's free... Reduced node size below ~32nm basically allows for reduced power consumption, which is always appreciated on highend graphics cards, which tend to consume a lot of power and produce a lot of heat.
Posted on Reply
#7
alwayssts
Yeah, 20nm designs probably taped out/started production in April, which multiple sources have reported they started ramping production from risk. Perhaps the first yields on those designs were better than expected...or perhaps good enough for AMD (later it was also reported it had reached a point Apple was also satisfied with the process).

While it made sense for amd to stick to 28nm for products pretty much any time before October, the farther we got away from first half '13 and closer to at least a six month break from that 20nm news, the more realistic the possibility became such a refresh may be cancelled (as it would need to have a shorter life). Perhaps they realized they were just not going to win in performance (which seemed obvious if you compare core designs even if it was theoretically close) and taking the value backseat (especially at the high-end without a huge redundancy/professional gpgpu route to make it feasible) was not as strong a financial option as eatting the process cost on 20nm and selling those products at a huge premium (ie what they did with Pitcairn when it launched at a ridiculous $350 for a 212mm chip with 5ghz value gddr5).

Another thing to note is 16nm risk is starting in q4 THIS YEAR (according to tsmc). That means the typical 1.425 increase we expect between AMD generations will probably get roughly a 1/3 increase over that sooner rather than later...if not a somewhat rapid turnover of slightly larger 20nm products with perhaps a bigger bus and/or high-speed gddr5 to a smaller chips with gddr6 (like r600->r670). I think that potentially explains why Pirate Islands is listed already.
Posted on Reply
#8
hardcore_gamer
by: erocker
I'm missing the part where it says 20nm for first half of next year. Seriously, I can't find it.
Under the subheading "Manufacturing and Capacity", first paragraph, final sentence:
"Mass production for 20nm is planned for the first half of next year."
Posted on Reply
#9
RCoon
Forum Gypsy
Nobody?

YAAAAAAAAAAAAARRRRRRRRRRRRRRRR!

I'm sorry.
Posted on Reply
#10
kn00tcn
i wanted to post in the news feedback thread, but it didnt warn me that there is a time limit on how old the last post is, so i lost my post there...

i wanted to say that TPU should post more rumor stories that are seen on videocardz, it's not like TPU avoids rumors, but a few stories were missed like the frame-pacing driver was dated for july31 according to the official amdradeon twitter
Posted on Reply
#11
Recus
Don’t get too excited about the 20nm rumors, this is not going to happen, including new Radeon series.
videocardz.com
.
Posted on Reply
#12
PLSG08
by: RCoon
Nobody?

YAAAAAAAAAAAAARRRRRRRRRRRRRRRR!

I'm sorry.
YAR YAR FIDLLY YII A PIRATES LIFE IS THE LIFE FER MEH

But to the point I hope the new naming system would be simpler
Posted on Reply
#13
hrrmph
Huh?

So lets see...

Hawaii is a large volcanic island (the 'Big Island') originally populated by Polynesians; and

New Zealand is two large volcanic islands (the South Island and the North Island) originally populated by Polynesians.

--------

So what's so confusing about a single GPU model being named Hawaii and a dual-GPU model being named New Zealand?

-
Posted on Reply
#14
nt300
This is what I am waiting for the high end next gen graphics,:D
Posted on Reply
#15
xvi
High yield? Sounds good. I just hope I don't have to plunder my bank account to get one.

seewhatIdidthere?
Posted on Reply
#16
theoneandonlymrk
by: nt300
This is what I am waiting for the high end next gen graphics,:D
+1 and christmass might well be enjoyable this year(at my age only my presents to me/ from me arent socks):D
Posted on Reply
#17
midnightoil
Press launch event is in September.

No way in hell this is 20nm. 28nm for sure.
Posted on Reply
#18
NC37
So are we finally getting new GPUs or just another gen of rehashes? ;)
Posted on Reply
#19
Xzibit
Everyone forget about Globalfoundries ?
11/30/2012 - Mike Noonen, executive vice president of worldwide marketing and sales
"The first multiproject wafer runs for customers testing Globalfoundries' 14-nm FinFET manufacturing process technology could start as soon as the first quarter of 2013"

"Noonen said Globalfoundries is already able to run the full 20-nm manufacturing process flow at its recently completed Fab 8 in Malta, New York. Noonen declined to say whether Fab 8 was already running commercial, high volume production of 20-nm circuits for customers"
Posted on Reply
#20
bencrutz
by: Xzibit
Everyone forget about Globalfoundries ?
switching to different process (gate last vs gate first) would cause a hell of a headache and definitely not cheap. i think amd would keep glofo for cpus only. it's a bad (and done) deal anyway.
Posted on Reply
#21
Casecutter
Odd didn’t notice this topic yesterday...

First an FYI here in the states; it appears AMD is clearing the pipes. Egg's had a bunch of AMD deals the past week... like a P-C 7850 1Gb for $110 –AR, even now TigerD has the MSI R7950-3GD5/OC BE for $180 –AR$30.

Giving what's materialized for end of September in Hawaii, some prject it as Technical/Press presentation for Shock & Awe, I'm not sure what will transpire anymore. I thought we'd see a 20Nm "Hawaii" as more or less "paper launch" in December with volumes Q1 2014. While I figured they’d surprise everyone with a with 28Nm cards as Pitcairn replacement (Maui) for October. The rumors are pretty clear the at this point "Hawaii" is coming on the 28Nm process.

I reckon this end of September hurricane will be an all-encompassing technical presentation of AMD graphic hardware, roadmaps, software, and how it all compliments their Gaming Evolved program. Figure they be floating stuff on Kaveri APU's, perhaps something on the game consoles, HSA, and OpenCL
Posted on Reply
#22
Casecutter
Consider these second generation GCN chips probably don't have the R&D or risk this time around, while something tells me AMD will try harvesting 3 "true derivatives" from Hawaii not XT/LE and a gelding. When Nvidia binned 5 derivatives from the GK104, I think AMD has been rightly enlighten!
Posted on Reply
#23
HumanSmoke
by: Casecutter
Consider these second generation GCN chips probably don't have the R&D or risk this time around.
Kind of depends on what you're expecting. If GCN 2.0 is an incremental tweak of GCN 1.0/1.1 then that would be true. Likewise if you're only expecting a modest increase in core count.

That of course would be at odds with some of the expectations of others (~25% increase in die size, core/TMU/ROP count, TDP etc. seems to a favourite configuration), or Charlie D's assertion that Hawaii will comfortably depose Titan's performance (and presumably a fully enabled GK110) with his hyperbolic "This one [Hawaii] is going to change the market mainly because there is no competition for it this year"
Posted on Reply
#24
radrok
I'm fairly sure you can delete your own posts, just click EDIT and then Delete :toast:


I just hope they can manage atleast 30% more performance than Titan with the new manufacturing node, if it's less then Nvidia will probably charge over the moon for Maxwell...
Posted on Reply
#25
HumanSmoke
by: radrok
I just hope they can manage atleast 30% more performance than Titan with the new manufacturing node, if it's less then Nvidia will probably charge over the moon for Maxwell...
If its 30% more than Titan then we're fucked. AMD will use the same pricing strategy as the initial HD 7970 launch ($549) and the HD 7990 ($1k) and price according to the competition...then we have to rely upon Nvidia lowering prices :laugh: or AMD suddenly deciding to not cash in on...a cash cow.

30% more performance than Titan means roughly 50% increase over the 7970. Even with a tweaked architecture, that's going to be a big-ass die considering the process node is still 28nm and heat/power are still going to limit core clock to what is presently in use.
Posted on Reply
Add your own comment