I'm reading on Nordic Hardware that their sources tell them that we're unlikely to see any mass availability of the 380X until Q3 of 2015, even if the card will be officially released in late Q2(around Computex).
By that time, Nvidia will have had ample time to prepare a consumer version GM200 after releading a Quadro/Tesla GM200 GPU in late March at the GPU conference.
Considering that AMD is bleeding red and that 2015, which should have been a slam dunk for them considering their node advantage and HBM, now seems to be another lost opportunity. NHW's sources indicate that AMD is going for 28 for time reasons, even if it's technically possible to do 20 nm at this stage with decent yield. (Money may be another factor). We're going to see some kind of GPUs on 20 nm, as Lisa Su stated, but now it looks likely that it could just end up being their Carrizo APU's graphics part and/or enterprise GPUs.
In 2016, Nvidia will use HBM so that advantage goes away. If AMD stays on 28 nm, then their node advantage will probably be lost in 2016 too. And since most of this year will be lost until Q3, their revenues won't improve.
Anyway, I'm hoping that they'll survive - because monopoly isn't good for anyone, the least of which our wallets and the pace of technical innovation - but it increasingly looks like that they won't and/or will be consigned to niche stuff. Who can/could step in in the desktop GPU space?
By that time, Nvidia will have had ample time to prepare a consumer version GM200 after releading a Quadro/Tesla GM200 GPU in late March at the GPU conference.
Considering that AMD is bleeding red and that 2015, which should have been a slam dunk for them considering their node advantage and HBM, now seems to be another lost opportunity. NHW's sources indicate that AMD is going for 28 for time reasons, even if it's technically possible to do 20 nm at this stage with decent yield. (Money may be another factor). We're going to see some kind of GPUs on 20 nm, as Lisa Su stated, but now it looks likely that it could just end up being their Carrizo APU's graphics part and/or enterprise GPUs.
In 2016, Nvidia will use HBM so that advantage goes away. If AMD stays on 28 nm, then their node advantage will probably be lost in 2016 too. And since most of this year will be lost until Q3, their revenues won't improve.
Anyway, I'm hoping that they'll survive - because monopoly isn't good for anyone, the least of which our wallets and the pace of technical innovation - but it increasingly looks like that they won't and/or will be consigned to niche stuff. Who can/could step in in the desktop GPU space?
Last edited: