1,401 karma · joined December 23, 2008
Alternately, you can try out various settings such as number of threads. You can also run the "Misc" tests which are a variation of the STREAM benchmarks. The "Simple" simply tests the memcpy performance.
edit: Some scores. 1.25 GB/s on Samsung Hummingbird processor. 1.8GB/s on the previous-gen dual-core Snapdragon.
Btw you may want to know that there is a bug in 4.0.3 that sometimes loads armeabi code instead of armv7-a code on some ARMv7 systems like HTC One S.
This is a CERN project and it uses Clang from the LLVM project. The idea is simple: use the clang to generate LLVM and then use LLVM just-in-time compiler.
when people use iPads they end up just using technology to consume things instead of making things. With a computer you can make things. You can code, you can make things and create things that have never before existed and do things that have never been done before.
I think closed devices are really a step backwards. I learnt programming (and a whole lot more, like mathematics and logic), when I picked up a computer as a kid and discovered Python and started hacking around. I could look around for tutorials, download code or try out code from books and nobody stopped me.
On a somewhat similar vein, I am also sad that desktops are slowly going away. I learnt so much more about computers (and how it is not magic) by building my own.
I think something needs to be said about systems that allow you to explore and hack around for the fun of it.
The physics and the economics of smaller process nodes is becoming harder and harder, to the point where anything beyond 10nm or so looks extremely difficult.
There is already an Android port of Qt in progress here, with a lot of things (including Qt Quick and QWidgets) already working: http://sourceforge.net/p/necessitas/home/necessitas/ Several apps using Qt have already been published in Google Play.
1. Using raw pointers as opposed to vectors gives 2x the performance on my machine.
2. This is a microbenchmark with extremely simple array indexing pattern. JVMs can optimize away the array bounds checks penalty in this case but those become hard to do once your access patterns become anything other than "i" or "i-1". For example, even more general linear subscripts are hard to correctly optimize for array bounds checks, let alone anything non-linear. How do I know? Compiler researcher here :)
However, still impressive to see JVMs advance.
Sure, it is still not a one-click task, but that can never be the case. I think Qt has a bright future simply because it is essentially unkillable at this point. (Open-source licensing, open governance already in place, highly portable).
Took me a while to realize that you can actually apply effects to the video element!
edit: I think we can safely conclude that hackers are about as human and as fallible as any other group. We have our own fashions and our own tabloids.
For one, it is a rumor. Apple Insider isn't exactly known for well placed sources inside RIM.
Second, even if we were to take this at face value, read the text: "The layoffs will affect the company's legal, marketing, sales, operations, and human resources divisions, a source said." Not that engineering isn't mentioned. That is a good cutting of cruft.
I must say I am a little disappointed that rants like these are now on HN front page.
However, I speculate even the 2nd restriction is partly for business reasons because it prevents some third-party toolkits (such as those compiling to LLVM) from achieving high performance. You might remember that Adobe was compiling Flash to LLVM to allow devs to port their Flash apps to iOS, but since LLVM cannot do proper JIT compilation on iOS, it will reduce performance.
FWIW: WinRT imposes both the restrictions above. Android technically doesn't impose either for sideloaded apps. But if you want to distribute app through Google Play, the 1st restriction is indeed imposed.