Forgot your password?
typodupeerror

Comment: Re:Jobs Told IBM and Sony Where to Stick Cell (Score 5, Interesting) 457

by kc8apf (#44089297) Attached to: PlayStation 4 Will Be Running Modified FreeBSD

Having been part of the team that evaluated practically every processor being considered for Apple products from 2003-2009, Cell wasn't used because it sucks as a general purpose processor. The SPUs are interesting but you need to completely rewrite algorithms to use them effectively. While porting to Intel wasn't exactly easy (mostly due to the endian switch), it didn't involve rewriting every compute-heavy algorithm from scratch. Intel also had a roadmap while Cell was a point design.

Comment: Re:In related news (Score 5, Insightful) 460

by kc8apf (#36782994) Attached to: Lennart Poettering: BSD Isn't Relevant Anymore

It's interesting that the question implies that Linux is leading the charge in defining new APIs. Everything listed has a FreeBSD equivalent that predates the linux version:

cgroups -> jails
udev -> devfs
fanotify, timerfd, signalfd -> kqueue

Of course, the Linux developers decided to reinvent them all making compatibility impossible. I guess you could argue that the Linux versions offer some extra features over the FreeBSD versions, but from a user and developer perspective, the FreeBSD versions seem more complete and stable (see jails vs cgroups).

Comment: Re:Mac OS X (Score 1) 245

by kc8apf (#33320844) Attached to: Open-Source 2D, 3D Drivers For ATI Radeon HD 5000 Series

Intellectual property from other companies generally has to be stripped from the code base and those algorithms reimplemented in a different way. Yes, technically those other companies could open-source their code, but generally they don't. Sadly, that intellectual property is almost always used to get high performance.

Comment: Re:IO limited? (Score 2, Informative) 208

by kc8apf (#32868346) Attached to: Fastest Graphics Ever, Asus ARES Rips Benchmarks

Typically for graphics cards, the only data sent over PCIe is texture data, vertex lists, and commands. The bulk of the operations done by the card are running the commands over the vertex lists while bringing in texture data. The commands are almost always a multi-pass or pipeline so each vertex will be used in computations more than once. The result is the pushed to the monitor, not the PCIe. So, yes, in general, a graphics card will have more FLOPs than I/O bandwidth.

"Success covers a multitude of blunders." -- George Bernard Shaw

Working...