notebookcheck: soc shootout x86 vs. arm

[monstercameron]

www.notebookcheck.net/SoC-Shootout-x86-vs-ARM.99496.0.html

Unfortunately they didnt have any results for the a4-1200-a6-1450 socs so they used a relatively OP a4-5000. I still don't trust geekbench comparing x-uarch and x-os...

 

[Exophase]

Shame they didn't use the recently released Geekbench 3 instead.

 

[Madpacket]

Looks like ARM is advancing much faster than X86.

 

[Arachnotronic]

Shame they didn't use the recently released Geekbench 3 instead.

Yeah, I'm excited to see the results for all of our favorite mobile SoCs on Geekbench

 

[Nothingness]

Looks like ARM is advancing much faster than X86.

Given where ARM starts from and where x86 stands, that's not surprising

 

[Nothingness]

The WebXPRT score seems to favor Intel (IIRC F Piednoel once said it was a great benchmark...). Does anyone know what's in there?

 

[Exophase]

The WebXPRT score seems to favor Intel (IIRC F Piednoel once said it was a great benchmark...). Does anyone know what's in there?

Yeah and I bet he thinks Geekbench is an awful benchmark I'd love to see him explain his criteria for this.

The WebXPRT page says this:

"WebXPRT 2013 is a browser-based multi-platform benchmark that uses representative Web-user scenarios to compare the performance of almost any Web-enabled device. It contains four HTML5- and JavaScript-based workloads: Photo Effects, Face Detect, Stocks Dashboard, and Offline Notes. Because the workloads are all HMTL5 based, they run on a wide variety of devices and operating systems. In addition to scores, WebXPRT 2013 gives a report on the HTML5 capabilities of a device."

Anand got some more info about what these tests actually do:

http://www.anandtech.com/show/6528/samsung-ativ-tab-review-qualcomms-first-windows-rt-tablet/2

So the compute part is Javascript based, but it's also heavily dependent on the performance of the browser's HTML5 APIs. Three out of the four tests are described as using Canvas 2D to access data.

 

[Nothingness]

There's MobileXPRT too: http://www.principledtechnologies.com/benchmarkxprt/mobilexprt/2013/results

K900 does surprisingly well.

Principled technologies seems to have a bias...

 

[FwFred]

Seems like a more meaningful mobile usage-driven benchmark compared to Geekbench. Say what you want about JavaScript, it has a real impact on all our phone usages.

At least Geekbench 3 (vs 2) seems to have more useful tests, and separates out the single thread score so that silly 8 core phones serve less marketing points.

 

[Nothingness]

Exactly: JS is more meaningful and in all other benchmarks Z2760 looks worse than it does on XPRT.

 

[FwFred]

There's MobileXPRT too: http://www.principledtechnologies.com/benchmarkxprt/mobilexprt/2013/results

K900 does surprisingly well.

Principled technologies seems to have a bias...

Looking at their site, it seems they provide souce code if you sign up. Has anyone tried this?

It seems to be more open than other mobile benchmarks.

 

[aigomorla]

ROFL i love how the atom just sits in the middle of every benchmark.

its like it doesnt care to win however its commited on not being last.

Also i use windows 7 on my atom setup as a torrent box.
Its not bad with a SSD, cpu loads are quite high all the time, however it still feels responsive, as long as i dont have too many programs running in the background.

Going from those benchmarks tho, your telling me my brothers galaxy S4 is faster then my torrentbox? :O
There has to be a way to load up windows 7 via VM or something on it... if the benchmark results are accurate.