How much is the 1080 Ti?
If a 480 ties a 1060 in game x for y frames per second .. then who cares how many smurfs be running around pulling levers and pressing buttons inthere? The metric "overall performance" is rather vague.
What history has shown that AMD even with all around better card and better prices, they still won't get more then 40% of market, let alone 50%+.
I did look at the 1060 but was slightly pricer, 2GB less ram, no dual card support and other factors made my decision.
Not really. Contrary to popular belief, ATi had better cards with 9xxx series, X800 series and x1800 series. Didn't win them redacted. As I said, I worked as student at retail (6-7 years ago). Literally no one recomended ATi. Not because they knew something, its because it was Nvidia and Nvidia+Intel was the way if you were gamer. AMD needs better marketing and better perf per watt. 5 years ago they needed better marketing, now they need both.It has to be cumulative. For AMD to really sock it to NVidia, they will have to win consecutively for multiple cycles, like NVidia had done. Winning one cycle every now and then won't do it. In the mind of the consumer, AMD needs to be associated with high performing, high quality products, and not just bargain hunter cheap hardware.
Yes, but this hasn't been happening. The perf/watt gap between AMD and NVidia is now so large, that AMD has not had the performance crown in a very long time. Their products are not only slower, but they also use significantly more energy. It would be one thing if their products were significantly faster than NVidia's, yet used more power. But this is not the case.
It needs more ALUs, but what people forgot is that 1060 cores are clocked at much higher clocks. A clock that sometimes goes much higher (up to 2050mhz) in boost mode. Then ALU advantage is not really that big at all, when other runs at 50%+ higher clock.For the 480 to tie the 1060 requires nearly double the ALUs. The consequences however are felt more keenly on more powerful GPUs. Fury X has 4096 ALUs, but was so inefficient that it had difficulties using them effectively. This lead not only to less performance than the competition, it also forced AMD to use a AiO cooler to keep Fury X's wattage down. The GTX 980 Ti on the other hand ran perfectly fine on air cooling, and had tons of overclocking headroom.
This is how perf/watt can influence overall performance, and people's perception. Which is why AMD cannot afford to let NVidia keep owning them on perf/watt.
I don't know, I can't find any for sale.
Hmm, nope. Cypress was a TeraScale VLIW5 uarch, so divide SP count by 5 to get a number of cluster "units" - Cypress = 320 units/"cores"
Today RX480 8GB is equal to GTX1060 6GB in DX-11 and faster in DX-12/Vulkan, RX480 8GB cost less and have 20-30W higher power.
If you under-volt the RX480 and play latest DX-12/Vulkan games, RX 480 has the same or better perf/watt that GTX1060 6GB.
Edit: Fury Nano at 175W TDP is equal/faster than GTX 980Ti at 250W TDP in latest DX-12/Vulkan games (2016). Nobody talks about it though
This lead not only to less performance than the competition, it also forced AMD to use a AiO cooler to keep Fury X's wattage down.
So, you think it is equally valid to assume they will price it under the 1080 as that they will price it above the 1080?
Actually you can. Transistor switching speed increases as temeperature decreases, which reduces switching losses. Leakage current also decreases with temperature. Sometimes people perform measurements that verify this. That's why the Fury and Fury X have the same TDP, even though the latter is faster.You do not keep the wattage (e.g. active power) down by cooling, the only thing you keep down is the temperature...
That's why the Fury and Fury X have the same TDP, even though the latter is faster.
Actually you can. Transistor switching speed increases as temeperature decreases, which reduces switching losses. Leakage current also decreases with temperature.
Therefore i was specifically referring to active power and not leakage.
And at 28nm leakage is a minor contributor such that the statement, that AMD used water-cooling in order to keep the power down still is not valid.
Prices can't change between now and when the 1080 Ti releases?
Reducing the power usage is likely not the primary purpose of the AIO cooler, but it is a side benefit. Both the GPU itself and the VRMs will become more efficient at lower temperatures, though you're likely only talking a couple 10s of watts
That's why the Fury and Fury X have the same TDP, even though the latter is faster.
So, charging $600 (well, it was actually more like $700) at release for 1080 is cool, as is charging something like $500-550 (whatever the actual--read: "FE"-price is for consumer) for what will be the superior 1080ti, about a year later? That's not really what we've seen before, is it?
I'll take your work on that as I don't have a FuryX to test with at different temperatures, but I can personally attest that a 290 will draw more than 20W less at full load under water than running with the stock cooler. Give it a try sometime, I think you'll be surprised by how much of a difference it makes.It is an extremely small benefit, certainly not even close to "couple of 10s of Watts".
I'll take your work on that as I don't have a FuryX to test with at different temperatures, but I can personally attest that a 290 will draw more than 20W less at full load under water than running with the stock cooler. Give it a try sometime, I think you'll be surprised by how much of a difference it makes.
but I can personally attest that a 290 will draw more than 20W less at full load under water than running with the stock cooler
Not off the top of my head, sorry. It is very dramatic though, I might test it this weekend as I need to do some monkeying with a system anyway. I have 10 H220's sitting in boxes that really should get put to use or sold.That looks unusual high. Do you have more information, what is the temperature and voltage in both cases?
That's what happened to Tahiti, Hawaii, Fiji and even Polaris to an extent. At launch, all of these architectures were underperforming..
No, overall performance is just an expression or result of perf/watt. Perf/watt is what the engineers go after when designing these GPUs, because it's what determines the actual performance characteristics. Overall performance is what we consumers see in the benchmarks..