Alan Cox <alan@xxxxxxxxxxxxxxxxxxx> writes: > On Thu, 08 Nov 2012 09:43:29 -0800 > "Wolfgang S. Rupprecht" <wolfgang.rupprecht@xxxxxxxxx> wrote: >> Can someone familiar with how the AHCI driver works confirm that data >> overruns on a SATA link can cause the driver to down-shift the SATA >> speed 6->3->1.5 Gbits/sec. I don't see any kprintf's but I notice that >> a high speed SSD acting as if the SATA were running at 1.5 Gbits/sec. > > It has a set of heuristics based upon error rate over time. > > I'm not sure what you mean by "data overruns" however The SSD in question is capable of 550 MByte/sec (Intel 520 SSD), while the PCIE controller is only an x1 single lane controller which probbly saturates the single lane at 480 MByte/sec. I'm assuming (perhaps incorrectly) that the controller is choking when it receives more data from the SATA than it can transmit on the PCIE. The bootup kprintf's show a 6Mbit/sec negotiation, but later when I try a transfer I see 130 MByte/sec transfers for a raw partition dd, hinting strongly that it is now talking at 1.5Mbit/sec on the SATA. I'm wondering if it is silently downshifting the SATA speed. In comparison, a slower SSD rated at 250MBytes/sec (which can stay within the PCIE x1 lane budget) runs at full speed, of slightly over 250MBytes/sec on the same controller. -wolfgang -- g+: https://plus.google.com/114566345864337108516/about -- users mailing list users@xxxxxxxxxxxxxxxxxxxxxxx To unsubscribe or change subscription options: https://admin.fedoraproject.org/mailman/listinfo/users Guidelines: http://fedoraproject.org/wiki/Mailing_list_guidelines Have a question? Ask away: http://ask.fedoraproject.org
