Date: Fri, 16 Jan 2015 22:21:07 +0200 From: Mihai-Alexandru Vintila <unixro@gmail.com> To: freebsd-stable@freebsd.org Subject: Re: Poor performance on Intel P3600 NVME driver Message-ID: <CAORpBLUYzHw%2BRRccRpkcxYrnjnckRMLyAc86J9UPpirRSeWxqQ@mail.gmail.com> In-Reply-To: <CAKAYmM%2BEpvOYfFnF1i02aoTC2vfVT%2Bq=6XvCPBnYg-mRQAVD1A@mail.gmail.com> References: <54B7F769.40605@gmail.com> <20150115175927.GA19071@zxy.spb.ru> <54B8C7E9.3030602@gmail.com> <CAKAYmM%2BEpvOYfFnF1i02aoTC2vfVT%2Bq=6XvCPBnYg-mRQAVD1A@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
@ Chuck Tuffli - error was caused by older firmware + the fact that the nvme controller didn't negotiate on x16 so i had to force the pcie to x16 from bios. With the update is fixed, i've just added it to be indexed since i couldn't find anything on google. @Barney Wolff it's a new pool with only changes recordsize=4k and compression=lz4 . On linux test is on ext4 with default values. Penalty is pretty high. Also there is a read penalty for read as well between ufs and zfs. Even on nvmecontrol perftest you can see the read penalty it's not normal to have same result for both write and read On Fri, Jan 16, 2015 at 8:19 PM, Chuck Tuffli <ctuffli@gmail.com> wrote: > On Fri, Jan 16, 2015 at 12:12 AM, Mihai Vintila <unixro@gmail.com> wrote: > > I've tried to increase it and obtained around 90k iops for write, but > now i > > only get : > > > > nvme0: WRITE sqid:6 cid:125 nsid:1 lba:3907028496 len:16 > > nvme0: WRITE FAULTS (02/80) sqid:6 cid:125 cdw0:0 > > nvme0: async event occurred (log page id=0x1) > > nvme0: WRITE sqid:6 cid:126 nsid:1 lba:528 len:16 > > nvme0: WRITE FAULTS (02/80) sqid:6 cid:126 cdw0:0 > > nvme0: async event occurred (log page id=0x1) > > nvme0: WRITE sqid:6 cid:127 nsid:1 lba:3907027984 len:16 > > nvme0: WRITE FAULTS (02/80) sqid:6 cid:127 cdw0:0 > > nvme0: async event occurred (log page id=0x1) > > If it helps, 02/80 means media error / LBA out of range, and log page > 0x1 contains command error information which will probably show that > the starting LBA field for the write command was the issue. > > --chuck >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAORpBLUYzHw%2BRRccRpkcxYrnjnckRMLyAc86J9UPpirRSeWxqQ>