Hi Frederick! * Frederick Page writes: >Hi Joerg, >Joerg Schilling wrote on Sat, Oct 12 2002:
>>The fact that Linux does not use DMA with ATAPI has ben reported by >>may people! >For example in the Linux Kernel Mailing List, and one would assume, >that those guys know what's going on. Yes, they should, but it seems, that it's not the case. Or how would you explain my results? And then agaion it wasn't many people on LKML. >>In addition, it may be that Linux does not report the load correctly >>when doing PIO to the drive. >This (sadly) is another issue, when using e.g. "top", one should >*not* look at the CPU-consumption of cdrecord, but have a look at >the "idle" time. You will find, that cdrecord e.g. uses 5% resources, >the other tasks maybe another 5%, but that idle-time (should be around >90% in this example) is only 10% or less. Well, I don't think this behaviour is wrong actually. What really eats up CPU time in PIO mode is the interrupt handling. So it's not cdrecord, that uses the CPU, but the system and that's they way it is shown in top. As system time. >Meaning: the ressources shown in "top" do not add up when the kernel >falls back to PIO mode. It's quite easy to get confused about the real >ressources-consumption, when just watching the processes and/or >system/user times. Again: the idle time is important. I'd say system time is important. High system time normally means that something is wrong. In this case PIO is used instead of DMA. regards Markus -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
