I use this bootloader: title Fiasco, Console, Loader, Tiny Shell (run) root (hd0,2) kernel /boot/fiasco_mod/bootstrap -modaddr 0x02000000 module /boot/fiasco_mod/fiasco -nowait -serial_esc -comspeed 115200 -comport 1 module /boot/fiasco_mod/sigma0 module /boot/fiasco_mod/roottask task modname "bmodfs" attached 3 modules module /boot/fiasco_mod/names module /boot/fiasco_mod/log module /boot/fiasco_mod/dm_phys module /boot/fiasco_mod/l4io module /boot/fiasco_mod/con -vbemode 0x117 module /boot/fiasco_mod/simple_ts -t 300 module /boot/fiasco_mod/bmodfs module /boot/fiasco_mod/run module /boot/fiasco_mod/libloader.s.so module /boot/fiasco_mod/libld-l4.s.so module /boot/fiasco_mod/loader --fprov=BMODFS run
I have to compare l4 with ecos. I want to use lmbench, bonnie++ and AIM-suite9, my questions are: - How can I use those benchmarks when I boot Fiasco?
I think lmbench, bonnie++ and AIM-suite9 are not suitable when you compare l4/fiasco with ecos although they are real-time operating systems. L4 itself is not linux binary compatible but l4linux could provide full linux binary comptability. And these benchmark tools must run on the linux-binary-compatible system such as BSD,UNIX, etc. So we could compare l4linux with linux by using lmbench, bonnie++ and AIM-suite9. Although ecos could provide many linux APIs but it is still not linux binary compatible. In common we can't compile these tools on the ecos so the benchmark is impossible. Additonal lmbench, bonnie++ and AIM-suite9 are designed for the general purpose operating system. From the result we could know the average performance of the system. Therefore, if you want to compare l4 with ecos you could take these two system as real-time systems and you should get some data showing real-time performance such as jitter, responsing time, etc. But there is no benchmark tool which could cover these two platforms : l4 and ecos. So you have to write your application for this goal. Thanks Cheng Guanghui