I found this slide illustrating what TDP can be used for Broadwell-Y Core M depending on device thickness and size:
I don't know if it has been posted in this thread already, but I think it's interesting regardless given the discussions about what to expect from this chip.
To put this into perspective, the Lenovo Yoga 3 Pro has a 13" display and is 12.8 mm thick. So in theory according to Intel's illustration it should be able to run the Broadwell SoC at 6 W TDP or so (assuming the notebook is well designed).
I talked about it and even the lenovo yoga pro's 3 thickness
That said the 12.8 mm is deceiving for that is two parts. Processor and Battery in the bottom, screen in the top. The bottom is little bigger than half but I have not see anybody measure it let alone use some nice calipers or something.
That said the performance is disappointing
for it does have a fan, there are limits you can do with such a thin device with cooling a desktop type performance. No fan and I would not complain that much.
Here is some comparison of computers in different form factors.
Cinebench 11.5 multithread comparison
2.48 Intel Broadwell Reference Platform at IDF
2.08 Yoga Pro 3 - mobilegeeks (german review)
35 / 37 w laptop chips. Picked i3s due to no turbo and pentiums due to no HT
2.60 Haswell i3 4000m (2.4 ghz) 37w tdp -notebook check
2.45 Ivy i3-3120M (2.5 ghz) 35 w tdp -notebook check
2.29 Sandy i3-2370M (2.4 ghz) 35 w tdp -notebook check
1.92 Ivy Pentium 2020M (2.5 ghz) 35 w tdp -notebook check
1.83 Sandy Pentium B980 (2.4 ghz) 35w tdp -notebook check
Big Tablets, Surface Pro 3 numbers
2.82 i5 4300u (1.9 base 2.6 ghz two cores, 2.9 one core) 15w tdp -anandtech
1.63 i3 4020y (1.5 ghz) 11.5w tdp -anandtech
Sunspider is a much more relevant benchmark, for both mobile and desktop.
Sunspider is garbage. I agree javascript performance is a relevant benchmark but Sunspider is too short of a task and is optimized heavily by the browser companies to look good. Javascript may have been useful in 2010 or 2011 but it is not useful anymore.
Sunspider is between .1 to 1 second for most nice devices now including high end phones and tablets. Besides optimizing for the test, that is so short every device will be able to sustain its max turbo and you can do turbo tricks such as running a chip at 4 ghz temporarily.
Different Javascript browsers are better. Mozilla Kraken is less than 2 seconds with the surface pro 3, 4.3 for Nvidia Shield Tablet, 4.6 for Iphone 6, 5.9 for Intel Atom Z3740, 6.6 for S801 at 2.5 ghz, 11.6 for Nexus 7 2013. I do not know how long google octane 2.0 takes to run for google purposefully tries tricks to see continuous use and thus higher is better with their benchmark. Personally I rather have a test that takes 5 mins to measure tablets and phones for sometimes you use sustained use of that length but anything over 5 mins is too synthetic.