Nvidia has claiming for a couple of years now that their good results are primarily because their mix has been shifting towards the high-end gaming products.That is quite surprising to see the performance segment as so big though.....
Nvidia has claiming for a couple of years now that their good results are primarily because their mix has been shifting towards the high-end gaming products.That is quite surprising to see the performance segment as so big though.....
Well, even if the shift is only a couple of percent per quarter, it will add up over the years.I read that but wasn't expected that much of shift
Nvidia has claiming for a couple of years now that their good results are primarily because their mix has been shifting towards the high-end gaming products.
Where did you see mainstream before April of P10?
Link it, I have not been able to FIND it. Prior to that, they talked about TAM, in January, but never talked about which Polaris chips.
The whole TAM thing is a joke anyway. Polaris does not do anything to extend VR install base. It is not the product you will buy when going after VR. In that case you are very likely looking for a graphics card in the 400$ and even way higher price range.That is simply one of those Roy Taylor bullshit messages. AMD should really fire this guy. Marketing wise he is really harming the company more then doing anything good.
Polaris is probably to make a very hard posh in notebooks. Without news of GP106 (yet) and with it sounding like Nvidia is doing a repeat of the GTX 980 for notebooks with the 1080/1070. Based on the best guess I can get from Steam Hardware Survey the sales of notebooks in that range is totaled is about the number of GTX 970 sold. AMD has no market presence here. While Pascal is ahead of Polaris in performance per watt so far AMD is close enough that their pricing could be very lucrative for the boutiques and especially the OEMs which run around with their $1000 950m/960m laptops.Per the table of contents links in the following, the graphics segments are defined in terms of the price and usage of the systems (display price included?) they go into:
https://jonpeddie.com/publications/pc_gaming_hardware_market_report
Enthusiast averages ~$2.2-2K for PC/notebook, minimum $1.8K.
Performance bleeds into the lower end of enthusiast or uses the last-gen enthusiast, averages are $1.5K PC, $1.3K notebook, minimum $1K.
Mainstream averages $775 PC, $579 notebook, maximum is $1K.
If one can see Polaris 10 going into a PC that manages to crack $1K, then it can at least get to the lower end of Performance.
Polaris does not do anything to extend VR install base.
I'd not really take that leap yet. While Gear VR does "something", it is somewhat far from what one could get from a full-fat VR experience, and the latter requires quite a bit of compute. If the argument is that the few things that Gear VR provides are good enough, then that's a bit of a different conversation, however I'd argue that the main issue is lack of killer AAA VR experiences now (I think only Elite Dangerous sort of qualifies?). Having said that, I'm not necessarily super rah-rah about VR, and find AR (e.g. HoloLens) more interesting, since it appears to have wider spanning relevance. Just my 2c.I hear Samsung Gear VR is selling and it certainly isn't powered by some 500W GPU configuration. I tried a demo unit at Best Buy awhile back and it has the VR bullet points covered. In other words, I think the supposed great demands of VR are a false assumption. Probably a perception fueled by the PC vendors who are targeting the big spender enthusiasts with a cunning "you can't have enough power" strategy.
Exactly my feeling as well.Having said that, I'm not necessarily super rah-rah about VR, and find AR (e.g. HoloLens) more interesting, since it appears to have wider spanning relevance. Just my 2c.
As far as I know, that one was a cut down GK104.True for Maxwell, with Kepler, they had GK106 at at least 62,5% of GK104 (I am under the impression the 660 OEM with 1152 ALUs was also GK106 based, but could be mistaken here).
According to WCCFTech, there's a Zauba shipping entry for what seems to be a 256-bit GP106. If true, then I think that GP106 won't use GDDR5X, and I have doubts about 8 Gbps GDDR5 as well.I think GP106 is in trouble. GTX 1080 is begging for more memory bandwidth. In terms of performance scaling. I think the post exploring differences has 1070 scaling nearly exactly to it's extra memory bandwidth + shader power. Where 1080 is scaling just over extra memory bandwidth.
Nvidia normal has cut 4" in half for 6". And GDDR5x is probably to expensive for $150-225 card. Hopefully Nvidia surprises with 192b or 256b bus.
% of GTX 1080
Speed Bandwidth bandwidth
5.4 Gbps 173 GB/s 54%
6.0 Gbps 192 GB/s 60%
7.0 Gbps 224 GB/s 70%
8.0 Gbps 256 GB/s 80%