tviceman
Diamond Member
In your example, 780Ti vs 980Ti, there is again a huge clock speed deficit on the 780Ti.
~25-30% clock speed gains on the 980Ti and extra vram which has an effect in many games from 2015 onwards. When you work it out, each Maxwell CC is effectively x1.2 to x1.3 Kepler CC only at the same clocks. Hence the 20-30% IPC.
If you want to claim Pascal has 30-50% IPC, with its ~20% clock speed advantage, GP104 will be 50-70% faster than Titan X.
Now, compare the last node shrink and uarch change, 580 -> 680. The 680 was ~25-30% faster. But the 580 chip itself was compute heavy and so it suffered perf/mm2 and perf/w, thus the 680 being a gaming focused chip, has even a better handicap.
This time, Titan X is already beast mode for gaming with gimped compute. It's already a lean-mean chip.
There's zero chance of GP104 being 50-70% faster. That's more for GP100. Because GM200 is already a gaming focused chip, I would say this time around, the delta will be potentially less than the 680 vs 580 comparison.
GP104 full ~Titan X + 20% is a good result on a small chip with the power savings.
However, Maxwell added back in compute performance vs. GK104 (and the less Kepler chips) and demonstrated better perf/transistor improvements on the same node than DP-stripped Kepler GPU's did on a new node vs. Fermi on an old node.
So trying to gather comparisons are getting to the point of moot. I think the best measurement will be to look at transistors and trying guess if there will be any improvement in perf/transitor. If GP104 has 50% more transistors than GM204 (5.2b x 1.5 = 7.8b, very plausible), even if perf/transitor stays the same, a 20% clock speed improvement will result in a 70-75% performance improvement over GM204, which ends up being 30-35% faster than Titan X at 4k.
Moving to finfets with a new / updated architecture throws comparisons out of the window. This isn't a typical shrink. There is so much up in the air to leave video card enthusiasts elated or disappointed, depending on how it all unravels.