Any Forza benchmarks? I'm trying to DL it now, but it's telling me i need to upgrade my Windows
I guess I don't have the latest version of Windows 10, how lame.
I have to back Windows 10 and DX12.
- After July 29th,
Windows 10 will cost $119.
Forza 6 requires DX10 and calls for 970/290X to hit 60 fps at 1080p. Don't we want true next gen games to look good and require decent hardware? What else is the point of new more powerful GPUs such as Pascal?
Sooner or later Windows 10 and DX12 will become the future of PC gaming, just like all previous Windows and DX versions were superseded by next gen Windows and DX versions.
Six DX12 titles (that are basically DX11 based + some DX12 features) is not many and a 980 Ti can play all of them just fine. We'd be lucky to be at a dozen titles by years end and a 980 Ti will still play them all.
Ya, but some of these DX12 titles are highly anticipated.
Total War: Warhammer
Deus Ex Mankind Divided
ARK Survival Evolved, Arma 3 and Star Citizen are rumoured to get DX12 support
There will be more games.
20% increase in performance over 980Ti, VRAM increase from 6->8GB, lower power usage, possibly DP1.3/HDMI 2.0b with support for HDR displays, updated NV video codecs, guaranteed NV driver support.
No doubt DX12 is the future but potentially improved DX12 performance isn't a major selling point to buy a card today given 99% of games are DX11 and performance there is looking to be very similar. Once DX12 really starts to matter GP104 will be a year old and we will want Big Pascal or Vega. And once DX12 games come out that a 980 Ti will struggle to run, we will be well into 2018 and getting ready for Volta.
OK, but your strategy of buying a $650 flagship NV (AMD) card such as a May 2015 980Ti and holding onto to it until Volta in 2018 is not optimal. It's simple mathematics:
2015 Buy a $650 Big Maxwell, sell it in May 2016 for $500.
2016 Buy a June $550 Mid-range Pascal, sell it for $400 2-3 weeks before Big Pascal drops.
2017 Buy a $650 Big Pascal, sell it for $500 in June 2018 before Mid-range Volta drops.
Total cost of ownership: -$650+$500-$550+$400-$650+$500 =
-$450
vs. your alternative:
2015 Buy a $650 Big Maxwell, sell it June 2018 for what $200?
Total cost of ownership:
-$450
Buying and holding onto to flagship AMD/NV cards makes no sense if you have access to information (i.e., roughly when new cards launch), if NV/AMD maintain their bifurcating a generation strategy (i.e, there is no indication they are stopping this strategy), if you have access to a 2nd hand market (most people in North America do).
Put it this way, if it costs slightly more to upgrade high-end cards, this is what happens:
Year 1: 2015-2016 = 980Ti = top-of-the-line
Year 2: 2016-2017 = 1080 = let's say 15-30% faster
Year 3: 2017-2018 = 1080Ti = 50-70% faster than 980Ti
In Years 2 and 3 you get a faster card. That means the TCO is roughly similar or barely more using the strategy I outlined but you have a very fast system in 2 out of 3 years.
RS, you make good points about retaining some cash by rolling over. I just don't like the idea of swapping out cards so damn often personally. I feel like I JUST installed these cards, lol. I literally feel like they are still brand new.
I could sell them for a loss, buy 1080's, then in 6 to 12 months sell THOSE for yet another loss and buy 1080ti's, but that's just too much activity as well as money and hassle for me to be honest. When I had my 670's, I had enough PC money ready to go that I could have built an entire new rig from the ground up with 980's and Haswell-E, but none of that stuff was anything like worth the money to me. I built a gym instead and got 980ti's and water cooling later, lol.
You left something out of the response: resale value of $800 GTX670 SLI when you got 980Tis and level of performance of 670 SLI in years 2 and 3 before you got 980Ti in 2015.
Look at the example I provided above, then do the math. Look at the level of performance too against the Base 2015 year. In 2017 when Big Pascal drops, how much will you be able to sell each 980Ti for? $300? In 2018, $200? Right now if you sell the 980Ti for $475-500, you roll over $475+ of capital into a new card that will lose very little until competition shows up.
If we set aside mid-range vs. large die for a second, it simply makes the most financial sense to roll-over the resale into the next gen as soon as possible because your upgrade cost ends up $100-150 per year per card and every year you have top-of-the-line. Well do the math using your own upgrading timelines and you'll see that my method of reselling flagship cards more frequently is better.
Btw, is it safe to say the more interesting thing is how well GP104 overclocks?
Because 25% ahead of a reference 980Ti is what many of us expected.
I am surprised it's not way faster if the Boost is 1860mhz.
2560 CC x1860mhz vs. 980 (2048 CC x
1265mhz) = 84% faster than GTX980
Theoretically should end up 47% faster than 980Ti at 4K:
980 @ 247 x 1.84 / 980Ti @ 310 = 46.6%
Something isn't adding up. Either the 2560 CC is too high, 1860mhz Boost is not stock, 1080 is bottlenecked elsewhere (for example still has 64 ROPs) or Pascal has lower IPC than Maxwell.