Hacker News new | past | comments | ask | show | jobs | submit login

The ONLY CPU benchmark that was not browser-based (and therefore running wildly different software) is the 3DMark Physics test, and A8 lost bad on that test.

Everything else was either single threaded (all the browser benchmarks), in which case dual vs. quad is irrelevant, or they were GPU bottlenecked in which case you aren't talking 1.4GHz dual core, and on the GPU front Apple spends a huge amount of transistors on that. They throw oodles of power at the GPU.

Also thanks to power management the vast, vast majority of those "2+GHz Quad-cores" were not running at 2+GHz, they were running at closer to 1.4-1.6 GHz. Apple is really good about settings its max frequency to something sustainable, whereas everyone else advertises burst frequencies.




Could you pls explore browser benchmarks being single threaded? I mean, when I use Chrome or Firefox on Android isn't the browsing experience implemented as multi-threaded (e.g. when I have the 8-core S5)?


There are multiple threads, but the benchmarks are specifically looking at JavaScript performance (as it's the only thing they can actually measure - anything claiming to measure drawing performance is a lie, whether or not they realize it, but I digress), and JavaScript is all single-threaded.

There is going to be painting and such while the benchmark runs, which will use a second thread, but that's about it. And that's going to be relatively light compared to actual usage where you're scrolling around and such.

The majority of the other threads of a browser are I/O bound (networking, disk, etc...) and won't be using much of the CPU.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: