[VC] AMD Radeon HD 9000 Series Launching in October?

[bill_d]

I'm thinking (the biggest they would do):
More shaders- 44CUs or double the SIMDs and TFUs per CU and go with ~24CUs(3072ALUs and 192TFUs) [depends on how much a diff the stronger FE makes]
Better frontend- the 4ACEs and 3geometry rumor thats been floating around sounds like a decent increase.
Slightly more ROPs- I don't see them going with 48ROPs unless they move to 512bit w/ 6ghz GDDR5(unlikely) or starve/make the ROPs more bandwidth efficient.
If they use 7Gbps GDDR5, that would offer ~27% more bandwidth allowing them to do the 40ROPs, a 25% increase.

Core clocks would be under 1ghz, probably around 900-950mhz(one would hope).

That should fit in <500mm2(imo 440-480mm2) and I would assume stay around 250-265w TDP.

That should roughly match Titan.

as a cfx user with a 1600p screen if i wanted to match titan i would have bought a titan but titan will not do 1600p with a single card reliably

i want a card that is very close to a 7990 so i will wait for 20nm on both camps

cfx and sli was fine with cards that were sub 450 bucks but at 650 to 1000 no, i'll keep my 7970 cfx and 680 sli till they can do it with a single card

[tajoh111]

as a cfx user with a 1600p screen if i wanted to match titan i would have bought a titan but titan will not do 1600p with a single card reliably

i want a card that is very close to a 7990 so i will wait for 20nm on both camps

cfx and sli was fine with cards that were sub 450 bucks but at 650 to 1000 no, i'll keep my 7970 cfx and 680 sli till they can do it with a single card

A 28nm maxwell at around gk110 size should do that.

gk110 with gk6000 specs(fully enabled), 100 more mhz and ddr5 at gtx 770 clocks should be within 10 percent of the gtx 690 or 7990. Heck a respin, with better memory on current gk110, should get this. Particularly if they are willing to raise the TDP and if the respin increases yields.

Adding the gains with maxwell should enable Nvidia to get even more than that.

[bill_d]

A 28nm maxwell at around gk110 size should do that.

gk110 with gk6000 specs(fully enabled), 100 more mhz and ddr5 at gtx 770 clocks should be within 10 percent of the gtx 690 or 7990. Heck a respin, with better memory on current gk110, should get this. Particularly if they are willing to raise the TDP and if the respin increases yields.

Adding the gains with maxwell should enable Nvidia to get even more than that.

if maxwell is 28nm no thanks i like in phoenx if i whan that much heat i'll go outside

and a "gk110 with gk6000 specs(fully enabled)" hey i guess the 1500 buck price point is open pass

[zalbard]

A 28nm maxwell at around gk110 size should do that

Maxwell won't be built on 28nm...

[boxleitnerb]

The initial Maxwell might. Depends on how ready 20nm will be in Q1 2014, how good yields are and how accurately Nvidia predicted this state at least 18 months when the decision what node to use was made.

[tajoh111]

The initial Maxwell might. Depends on how ready 20nm will be in Q1 2014, how good yields are and how accurately Nvidia predicted this state at least 18 months when the decision what node to use was made.

If AMD is willing to build a near 500mm2 chip, then 20nm is further off than we think and I could see some maxwell chips being 28nm.

It also might occur due to TSMC pricing on 20nm. 28nm was a huge jump in cost over 40nm. 20nm might be also a big jump which makes big chips too expensive to manufacture initially, similar to gk110.

[PedantOne]

The initial Maxwell might. Depends on how ready 20nm will be in Q1 2014, how good yields are and how accurately Nvidia predicted this state at least 18 months when the decision what node to use was made.

maxwell will be 28nm!!! Trust me First samples are done