Processing power

[v-600]

I've not really thought about processing power (recently everything has had more than enough for most of what I do) since I had an atom based aspire one 5 years ago.

The new macbook and the new surface have me interested again. From what I've read the Core M is good enough most of the time for most people, but it struggles with long periods of intensive action. Right now I have a surface pro 3 streaming to a chromecast doing on the fly transcoding from an .avi file. Watching the task manager the cpu is floating around 10-12% usage at about 0.8-1.2GHz (depending on the scene) with the odd spike up to max every 15 seconds or so. This is something my old Tegra3 Nexus 7 can just about cope with in short bursts so we're around the kind of real world level that starts to become an issue.

Is a Broadwell Core M running at 1.1GHz going to have comperable performance to a Broadwell Core i5 running at 1.1GHz? If not, what causes the differences? How will an Atom x7 compare?

 

[ShintaiDK]

The 1.1Ghz is a baseclock.

http://ark.intel.com/products/family/83613/Intel-Core-M-Processors#@Mobile

For Core M a lot depends on the PC its integrated to. Rather than the chip itself.

TDP limits set by the OEM, cooling made by the OEM and so on.

Transcoding can be done via quicksync.

 

[v-600]

I understand that its just a base clock. I was wondering more whether the Core M would perform the same tasks at a similar clock speed rather than the maximum performance possible from each. I assume the different models in the table are just binned depending on their quality rather than being completely different models.

 

[ShintaiDK]

Depends on on the OEM implemenation.

An example is that a 5Y10 device that outperforms a 5Y71 device due to the OEM.
http://www.anandtech.com/show/9117/analyzing-intel-core-m-performance
http://www.anandtech.com/show/9136/the-2015-macbook-review/10

 

[v-600]

Yeah, I was starting to get that from a few more of the reviews I read. It would be interesting to see from a purely theoretical POV whether you could lock a CPU at a certain frequency when testing devices to see any differences, but in practice, you're right and it seems like per device varies more than per cpu.

 

[Yuriman]

So far as I can tell, at a given clock, Core M performs the same as a Broadwell i5. It's just going to downclock a lot sooner due to TDP restrictions.