Not everyone with big/high resolution displays (or multiple of them) is using them for gaming.
I paid $799 for my HDMI 2.0 4k TV that I'm using as a monitor. They don't have to be expensive. I ended up with a 960 because it's the cheapest, lowest power draw card that would drive it correctly. I may upgrade eventually, but for now it is just about the perfect card for me.
Viper GTS
Fair enough. If you read my comments on the 960 on the forum, it is strictly related to gaming (OP's topic). I even went out of my way to mention that in case you have a very specific use such as the one you described, a case can be made for a 960. It feels unfortunate that there are no HDMI 2.0 choices at $50-70 levels so you didn't have to spend nearly $200 for only media. Hopefully with next gen AMD/NV refresh cards, all future cards will have HDMI 2.0.
If the gap between 2015 vs 2016 cards of the same price bracket really is going to be 70% on average, then it sounds like everyone should just ignore both the Maxwell's AND AMD 200/300 series and wait another year. TBH, I'm curious as to how you "know" to within 10% exactly how fast these future Pascal cards will be as I haven't seen any GTX 960 vs GTX 1060 or 300 vs 400 series benchmarks. (That's probably why people are "ignoring the key point" when most of us can't see two years into the future, and anything else is just random guessing...)
Nah, you missed this point too because you are not thinking about it. If a next generation x60 card in 2 years from now delivers even less than 60-70% faster, then it makes buying cards like R9 290/970 look even better since it's cheaper than buying 2 cards with a price of a 960 (one now and another in 2 years from now). That's the point. You get a 960 today for 165 pounds and some mythical 2017 card for 165 pounds that's likely at best will end up 60-70% faster, which is more or less where 970 sits today.
Where am I getting this data? Histrionically speaking we get a 40-70% bump in performance on new nodes + new architecture at similar price levels. Maybe AMD/NV will deliver better than that but looking at GTX460->560->660->760->960, the trends show:
460 (
July 10, 2010) / 560 (May 17, 2011) -> new generation+new node GTX660 (
Sept 13, 2012) = 87%/58%, respectively.
That means it took more than 2.5 years for NV to increase performance by 87% going from 460 to 660 and 1.5 years to increase it by 58% going from a 560 to a 660.
660 (
Sept 13, 2012) ->960 (
Jan 22, 2015) -> new architecture and nearly 2.5 years to get 44%.
Source:
http://www.computerbase.de/2015-03/geforce-gtx-460-560-660-760-960-vergleich/2/
That's where I am forecasting using reasonable historical data that chances are at best a card replacing 960 in 2 years will end up 60-70% faster.
You've repeated that several times but in many areas those cards are not even remotely in the same price bracket. Where I live, 2GB GTX960's in the are typically £155-£175 ($229-$259), R9 290's typically being £227-£250 ($335-$369), and R9 290X's typically being £275-£330 ($406-$488), and GTX970's £260-£300 ($385-$443). Those £72-£175 ($106-$259) price differences are not even remotely the same class of card.
But your salary is in pounds, correct? In other words in a similar job, chances are if I am making $50K USD, you are making 50K pounds. Looking at price/performance:
MSI Gaming 960 2GB =
165 pounds
960 4GB =
198 pounds
MSI Gaming R9 290 =
228 pounds
In both cases 960 loses. Since 960 2GB is a serious compromise for 1080P gaming, it loses even worse.
QUOTE=BSim500;37330633]Nor are 290w vs 120w TDP / power consumption or 438mm2 vs 228mm2 die sizes. [/quote]
What do die sizes have to do with price/performance and VRAM limits? Also, why are you comparing TDPs of AMD and NV when we know AMD and NV measure TDP differently? I realize that an R9 290 uses way more power but by you simply comparing the cards' TDPs exaggerates the differences. Further, you ignore the total system power usage.
It's not 120W of power vs. 290W of power as you are trying to make it. It's more like a 313W system vs. a 417W system but the 2nd system is 50% faster in gaming, thus having superior perf/watt.

:sneaky:
I've repeated this for many years now that using a card's TDP rating by comparing NV and AMD can be grossly misleading. By ignoring the perf/watt efficiency of the
entire system, you are also grossly misrepresenting the perf/watt metric the gamer enjoys. The perf/watt of each individual card is fine to compare for engineers at NV and AMD but it's not really what the gamer needs when building an entire rig since you can't just play games with a videocard alone.....
but the GTX 960 comes with Witcher 3 (listed at £30 at most places) which wipes out that plus draws a lot less power, chucks out less heat, needs only 1x 6-pin connector, full 4K/60fps H265 hardware decoding, 0db fans when idle / light gaming, etc, (which for many make up the premium in itself even without the game).
MSI 970 Gaming has most of those perks and it costs
284 pounds. That's only a 43% premium over the 960 4GB for 70%+ more performance. Sure a 165 pound 960 2GB prives similar price/performance but boy that 2GB of VRAM is basically worthless for so many modern games today unless you love playing at near PS4 level textures.
I agree with what you're saying on principle, but by extension you might as well recommend i7 / 32GB RAM / 1TB SSD components for a "mid-range" build to everyone regardless of their budget
Depends on a specific case - i.e., how long does a gamer intend to keep his parts, does he/she resell parts when upgrading, what the purpose of a computer is. If someone's budget is 600 pounds for a gaming rig and that build includes a GTX960, I will absolutely recommend getting a beefier PSU and getting an R9 290/970 because for 700 pounds or even 750 pounds allows one to get there which means 50-70% more gaming performance for just 17-25% increase in price. That makes a new gaming rig with a 960 a bad deal for gaming today. That's why 960 needs to come down in price or NV needs to fill this space with a 960Ti. Are you telling me when someone is spending 600 pounds on a new gaming rig, they can't be flexible with their budget to get 50-70% more performance for another 100 pounds considering they'll keep their GPU for 2-3 years? Sounds like a consumer that hasn't thought his PC building through. Budgets should be flexible to account for these scenarios. Sometimes spending more upfront saves more money in the long-run. That's all I am saying. I guess if you can't afford to stretch your budget by small amounts, then you make a point.