Here's a rambling question. Basically I want to understand the relationship\nbetween IRQs, SATA, and APICs, on mainboards that use the Intel 865/875\nchipset.\n\nI have an Abit IS7, running Linux 2.4.21 and Win2K. My focus right now is\non Linux. (a.c.p.m.asus readers, please don't flame me-\- my question\napplies equally well to any 865/875 mobo, and I figure there are quite a few\nowners of those reading your NG.)\n\nI'm considering adding a SATA drive to my rig. I've seen reports in the\nnewsgroups that some people are having trouble getting good performance out\nof these drives, especially in Linux, using mobos that have the 865/875\nchipset. The problem seems to be related to IRQ sharing-\- the SATA\nchannels are multiplexed with other devices on a single IRQ (e.g.\n[URL]http://lists.insecure.org/lists/linux-kernel/2003/Jul/0429.html[/URL]). Here's\nthe situation on my box:\n\n$ cat /proc/interrupts\nCPU0\n0: 153043 IO-APIC-edge timer\n1: 2693 IO-APIC-edge keyboard\n2: 0 XT-PIC cascade\n8: 3 IO-APIC-edge rtc\n9: 0 XT-PIC acpi\n12: 68748 IO-APIC-edge PS/2 Mouse\n14: 18451 IO-APIC-edge ide0\n15: 6 IO-APIC-edge ide1\n16: 90301 IO-APIC-level usb-uhci, usb-uhci, nvidia\n18: 16555 IO-APIC-level usb-uhci, SysKonnect SK-98xx\n19: 0 IO-APIC-level usb-uhci\n20: 21 IO-APIC-level DC395x_TRM\n23: 0 IO-APIC-level ehci-hcd\nNMI: 0\nLOC: 152994\nERR: 0\nMIS: 0\n\nMy boot log tells me that both SATA channels will go onto IRQ 18, sharing\nthat one IRQ with a USB 1.1 device and my ethernet controller. So it seems\nthat I may be headed for the sort of IRQ logjam that others have\nreported.\n\nSeveral questions arise:\n\n- Is it likely to be a problem to share one IRQ between two SATA drives, a\nUSB 1.1 device, and my ethernet controller? The USB is a slow device, but\nthe Ethernet could run at 100 or even 1000 Mb/s if I use it on my LAN.\n\n- Do I really have 24 IRQs now? Or do I still really have 16, which are\nlogically presented as 24? In the old days, there were only 16 IRQs (15\nreally), and it was sometimes a problem that there weren't enough to go\naround. But the FOLDOC definition of an APIC\n([URL]http://foldoc.doc.ic.ac.uk/foldoc/foldoc.cgi?query=apic&action=Search[/URL])\nsays that it "usually has more available interrupt lines that a typical\nPIC."\n\n- Can any of you report success or problems with throughput of SATA drives,\nusing the 865/875 chipset, especially in Linux? What does your IRQ\narrangement look like?\n\n- Has anyone successfully rearranged their IRQ usage to avoid logjams? Did\nyou do it in the BIOS, or in the OS (which OS?)? Unfortunately, the IS7's\nBIOS doesn't offer any explicit control over IRQs that I can find.\n\nThanks in advance for your time.\nAndrew.