Well, what we have here:
2010 - Clarkdale - max 12 EUs IGP, 900 MHz max clock. 43.2 GFLOPS peak
2011 - SandyBridge, max 12 EU IGP, 1350 MHz max clock, 129.6 GFLOPS peak
2012 - IvyBridge, max 16 EUs, max 1150 MHz clock, 294 GFLOPS
2013 - Haswell, max 40 EUs, max 1300 MHz clock, 832 GFLOPS
2014 - nothing happend, take same numbers
2015 - Broadwell+Skylake, max 48 EUs, max 1150 MHz clock, 883 GFLOPS
In case of Intel GPUs that`s right, by factor 20 instead of 16, but only because of starting pit was very deep. NVIDIA gets ony factor 6 at same time, AMD - 4... Unlikely next Intel solution will double GFLOPS, so the factor will be close to AMD/NVIDIA values.
Combine the figures from both the CPU and GPU and you will probably find the sheer number of transistors on the same die has at least doubled from year to year.
I would like to thank you for the feedback provided regarding Moore's Law and Intel.
Intel is always developing product to provided a much better experience regarding performance.
In transistor numbers there are no Moore`s law: from 2010 to 2015 in enthusiasts CPU line count jumps from 1170 Mil (Gulftown) to just 2600 Mil (Haswell-E).
Thank you for your feedback regarding Moore's law and Intel CPUs.