Originally Posted by Nickyvida
No amount of sugar coating will gloss over the fact that that wasnt a sound buisness decision. It should at least match a 1080Ti or at least slightly under it. Its been late to the party for more than a year and has been in development for longer.
It was a sound business decision given the resources, the risk and return on investment.
AMD poured all it's resources into CPU development over the last 3 years and this has shown in AMD very slow release schedule after the 7970/7870/7770 burst. AMD has outright said it focused everything into CPUs. The reason is AMD future had much more riding on ryzen. GPU's are a drop in the bucket, atleast at AMD's current level. Thus releasing another GPU design on top of what they did release likely would have lead to another delay with ryzen. A big mistake for AMD given the disastrous CPU division and lack of sales from it. Ryzen needed to come out ASAP for AMD to have a future.
This meant GPU's got very small budget and hence why AMD had to release only a couple GPU's it had no money for more as far as cash flow. This meant AMD had to direct its limited resources and had to pick some of the marketplace. The thing was, AMD was pissing off the OEM market,laptop market and mainstream market for basically only releasing rebrands. Cape verde was being rebranded a third time. No way they could do it a forth time. Also Tonga was a big let down and needed to replaced asap. It was too big of a die for the performance, particularly compared to the similarly sized gtx 970/980.
So what AMD did was released products that could get as much marketshare as possible while addressing the OEM and laptop market. In addition, AMD likely thought its original fury and fury X would still be sellable since it was only a year old.
In hindsight given Polaris performance per watt, die size and clock limitations, a high end polaris design would have been a failure because of bottlenecks and the shear die size of the chip would make it so that even a double sized polaris would have a tough time vs the 1080. A double sized polaris would likely clock at 1100-1150mhz and given the bottleneck of AMD's ROPs and the gtx 1080 would likely be tied.
Given the 1080 power consumption and die size, this would have been an engineering disaster for AMD where Nvidia could price their product in a manner to make sure AMD lost money or just didn't sell much cards.
Not only that, where AMD decided to attack was Nvidia's gx106 range. This has been Nvidia's weakest range because of the odd compromise between shaders and memory bus.
Thus given the resources, R and D and return on investment, low end and mainstream made more sense.Edited by tajoh111 - 4/10/17 at 11:51am