To view parent comment, click here.
To read all comments associated with this story, please click here.
That's what benchmarks are for
Already, the Haiku kernel performs many operations faster than BeOS did.
Often times reading the code does not reveal its relative performance, that actually takes real-world tests anyhow. The "performance" of BeOS wasn't necessarily a result of any special coding tricks that only Be, Inc. coders knew, but rather certain design decisions that Haiku also follows in many cases.
I think if there was ever a comparison made between Haiku's code and BeOS' code, we'd find that the Haiku code might be cleaner and better-written largely because it's FOSS, and built from the ground up to re-implement the same functionality as BeOS.
The kernel Haiku forked from was NewOS, which was written by an ex-Be engineer originally.
The BeOS kernel has really poor performance compared to even old Linux. It just had better design in some places (that Linux still didn't pick up), and so has the upper layers. The overall multithreading gives the impression it's faster, while the kernel is much slower.
Some parts of the BeOS kernel had gross hacks that allowed it to take shortcuts but wouldn't scale to today hardware.
Haiku has already much better code quality overall because it received a lot more peer review than BeOS ever had.
Ok. Thanx for your help. Ive heard lots about Haiku. Maybe Haiku on the desktop and Solaris on the home server with ZFS would be a nice combo.
My fear was that BeOS kernel was super engineered because everyone says it is so good and now we try to imitate it with Haiku. But what says that the imitation takes the same good design decisions as BeOS? Haiku is compatible, but it could maybe have a bad design which makes it inferior to BeOS. But benchmarks reveals that Haiku is as quick as BeOS on the same hardware. This means that Haiku should not be that bad as I feared.
Just because BeOS was super, does that automatically imply that Haiku is super too? No it doesnt. But now my fears are qualmed. Thanx guys. It will be very interesting to try out Haiku under VirtualBox!




Member since:
2007-07-27
Ok thanx for explaining this. One more question. The BeOS kernel might be totally totally different from the Haiku kernel. Is that a problem? We know that BeOS had great performance. The Haiku kernel, could it be built less optimal and have bad performance but still obey the BeOS api and therefore be binary compatible? Could BeOS kernel be super engineered whereas the Haiku kernel could be poorly engineered and have terrible performance? Could it be so? How do we know Haiku kernel is as good as BeOS kernel? We dont have the BeOS code?