AMD: Volcanic Islands R1100/1200 (8***/9*** series) Speculation/ Rumour Thread
DuckThor Evil
Legend
I got 432mm2, also quick and sloppy 
, but imo it can't be far off 430mm2, even if we didn't have that "nearly 30%" comment hehe.
, but imo it can't be far off 430mm2, even if we didn't have that "nearly 30%" comment hehe.That would mean a package size of around 455mm2. By the same measure i get 360 for tahiti... Sounds a bit high for gk110 to be "nearly" 30% bigger.
(I have a hard time to see how to get down to 432? I'm not including the shadows of the memory chips.. I get those to 84x72pixels, and 112x146 for hawaii)
I would rather use the mounting hole distance as basis than those small ICs - If we assume they are the same, I get a chip that is 24% larger than tahiti (ie 452 if we use the official 365mm2 tahiti).
(I have a hard time to see how to get down to 432? I'm not including the shadows of the memory chips.. I get those to 84x72pixels, and 112x146 for hawaii)
I would rather use the mounting hole distance as basis than those small ICs - If we assume they are the same, I get a chip that is 24% larger than tahiti (ie 452 if we use the official 365mm2 tahiti).
That seems a bit premature.Remember there's about 1mm to account for in the packaging of a die.
Also, pleased to see this is 512-bit
I'm pretty sure that's not the (main) reason.
Tahiti's 384-bit interface was rather large due to PHY optimizations for high clocks. They probably figured that a 512-bit interface with less PHY optimization would fit into nearly the same space, while allowing them to install 33% more memory and increase bandwidth without the need for expensive and power-hungry 7 GHz memory.
Additionally, keep in mind this chip is aimed at the HPC/GPGPU-Space as well, where more memory and bandwidth are actually more important than in the gaming space.
Sure, that's why I write package size. Wouldn't affect the relative scaling to tahiti too much though.
More pictures of ES.
http://videocardz.com/45723/amd-radeon-r9-290x-hawaii-gallery
Tahiti sized PCB.
8+6pin
6+2 PWM
Looks like 2 DL-DVI + other outputs (yay for when I get my QX/DP2710)
16 memoy ICs all on front, none on back
Has some sort of mod to it, soldered wire.
http://videocardz.com/45723/amd-radeon-r9-290x-hawaii-gallery
Tahiti sized PCB.
8+6pin
6+2 PWM
Looks like 2 DL-DVI + other outputs (yay for when I get my QX/DP2710)
16 memoy ICs all on front, none on back
Has some sort of mod to it, soldered wire.
Last edited by a moderator:
UniversalTruth
Veteran
Why is Pitcairn's (and supposedly Hawaii's) memory controllers so much more efficient than Tahiti's? What's wrong with that Tahiti?
Oh, if that's just an engineering sample and not even the reference design. But even if it is the reference, then what's the obstacle to go for the customised AIBs' versions?
iMacmatician
Regular
Not much more efficient, the comment was only on the size it take on the die... I believe the main reason is the ECC memory controllers and other Pro stuffs are disabled.. Pitcain dont have them.
I still have some doubt about this 512bits but.. but well.. it look pretty confirmed now ( 16 IC; 32bits each on the crossbar )
It won't be. They just slapped an older reference cooler on the ES.
Possibly built conservatively(redundancy?) because of the relatively unknown process?
Other possible factors, the speed(decrease on ROI [PHY area vs speed]) or the crossbar is added into the PHY?
No obstacle, it's just that AMD would fail to convey the same premium feel that NVIDIA does with the GTX 770/780 and Titan. In my opinion, this is not how you build a brand.
Plus, reviewers usually test the reference design, and base their comments on temperatures and noise upon it.
I think it's pretty obvious that entire card is an engineering sample (considering the jumper pins, the quick-fix jumper wire and what appears to be wifi antenna connectors on the rear; probably some kind of signal probe connectors, I suppose), so drawing any far-reaching conclusions about the cooler on actual production units would seem rather premature at this point...
That being said, I sure hope the cayman-style boxy blower does not make a comeback, because that design just friggin sucked ass, TBH. It is impossible to get the card to work nicely in a tight crossfire setup with such a design. The fan howls, and the card runs hot as a volcano. Not good.
That being said, I sure hope the cayman-style boxy blower does not make a comeback, because that design just friggin sucked ass, TBH. It is impossible to get the card to work nicely in a tight crossfire setup with such a design. The fan howls, and the card runs hot as a volcano. Not good.
"benchmarks"
http://translate.google.com/transla...UTF-8&u=http://udteam.tistory.com/539&act=url
does seem kind of legit considering they have pics with the an es and the other cards.
It these numbers are true...it will be quite astonishing to say the least
http://translate.google.com/transla...UTF-8&u=http://udteam.tistory.com/539&act=url
does seem kind of legit considering they have pics with the an es and the other cards.
It these numbers are true...it will be quite astonishing to say the least
Does it? There's 16 memchips
edit:
Claimed benches: http://wccftech.com/amd-hawaii-r929...osed-fastest-gpu-planet-1020-mhz-clock-speed/
Last edited by a moderator:
hurleybird
Newcomer
And that looks like ES hardware to boot. Who knows how mature the drivers they are using are.
Bravo AMD.
performance looks great , it will murder my 6950 but i'm still not sure on it. I think I will wait for the 20nm drop. I don't see enough gotta have them games this year to buy it and I would think in spring or fall the latest we will see 20nm cards on both sides even if its just Hawaii shrunk