Do you have a source for Skylake-U 64MB eDRAM power consumption? Can't simply extrapolate from different products.
Sure you can. It's the same eDRAM, just 64MB. It's probably the 2nd generation part with 1/4th standby power which makes it usable on low power 15W parts so it can have decent battery life. Active power though? Nothing changed. Well maybe a
little.
128MB eDRAM = 3W TDP
Now DRAM is different from a CPU because capacity doesn't necessarily determine how much of the cells are active. Whatever is being
used is active. If we take Intel at face value that 32MB is enough, both 128MB and 64MB would use same power. TDP isn't max power, but pretty much max power that it can sustain for a long period.
Even at 1.5W, which is literally half of 3W, that's still less favorable to a 15W part than a 3W part is to a 47W one. And the SoC is likely TDP constrained far more than 47W one is. There's no logical reason to expect that we'd get more from eDRAM on a 15W part than a 47W part.
Also 60% faster than the top HD Graphics 520 score is nothing to cough at given that these are the very first submissions so far.
Iris Pro's eDRAM can get 60% too. But
average is only 40%. Based on the 48 EUs being constrained by TDP + eDRAM taking up more TDP relative to the SoC if anything it'll be less, not more. Now I am not saying that's how it'll be, but only products with extensive testing will tell the truth, not single, synthetic, mobile-focused bench.
We all know those parts were targeted at Apple, PC OEMs were better served by regular Haswell/Broadwell + dGPUs.
Yes, and now you can add one more vendor, which is Microsoft.
And power consumption is much more critical here. If Intel can deliver anywhere close to Skylake-U GT2 + Geforce GT940M gaming performance with Skylake-U GT3e then they will have a very interesting product in their hands.
Please get back to me if Intel has changed this graph significantly. Broadwell is ~20% faster than Haswell and Skylake is ~20% faster than Broadwell. Geforce had 2-3x lead.
http://images.anandtech.com/doci/7834/NVIDIA-GeForce-800M-Slides (26).jpg
I know its an Nvidia marketing slide but something that's spot-on. Iris Pro 5200 parts were uncompetitive in perf/watt, perf/$ and Iris Pro 6200 changed nada. Benchmarks of Iris Pro 6200 showed that Geforce GTX 740 performed equal/better while using less power, while GTX 750 GDDR5 shows far better performance with lower power in one game and 8W higher in another.
http://pclab.pl/art64766-8.html
In some games, the Iris Pro 6200 falls behind GTX 740. The GTX 750 GDDR5 is sometimes 2x the performance.