Bas232, out of interest, what chipset is your motherboard based on?
Bas232, out of interest, what chipset is your motherboard based on?
MSI MediaLive nVidia MCP51 (well according to flashrom).
I checked with OCZ about the Vertex, but they couldn't say more then it should work with AMD chipset, and to check with MSI
So I talked to MSI and they told me nVidia isn't SSD verified by Intel SSD.
As Windows XP runs fine on either SSD, Vertex or Intel X25, I'm suspecting the Kernel has something to do with it or the ATA drivers.
The ncq is a longshot, but somebody else mentioned it a long time ago too, so I figured why not?
So far 9.10 (with all updates, but no version upgrade) works fine and no hickups or anything.
And yes, I have /var /tmp /home and swap stored on a normal harddisk, not on the SSD.
Interesting... mine is a totally different chipset (Intel P55). Also, I didn't try Ubuntu 9.x, only 10.10.
How do you test the system for stability? When I was actively experimenting with this, I found that doing a flat-out multi-gigabyte write onto the disk (for example, using "dd") would consistently trigger the fault.
That is one way.
I did a lot of installing and/or running W7 in a Virtual-box will trigger it too.
Simply put; create a lot of I/O on the SSD and it locks.
It could well be NCQ, as using NCQ on an SSD is simply stupid.
NCQ tries to order the read/write command in such a way that the heads of a HARDDISK don't need to wait rotations to read/write data.
But as an SSD doesn't have heads, nor does it spin; it access the data straight away without waiting, as such NCQ is unneeded and unwanted when you use an SSD.
My idea was that NCQ tried to rearrange the commands properly for a harddisk, but the kernel gets confused as to why the SSD delivers the data in the wrong order (or too fast).
So I believe the SSD / controller simply locks due to wrong communications.
I could be totally wrong, time will tell, but that is my idea behind it.
I believe an SSD doesn't know NCQ or is too fast for Linux to do NCQ.
If it didn't crash with this kernel:
Then I will upgrade to 10.04LTS tomorrow and do more testing.bas@workstation:~$ uname -r
2.6.31-22-generic
But I doubt it matters what version you run, I think it's a kernel/driver matter as 9.10 crashed the other day real hard.
And for others, no it's not the computer going wrong, as it does Linux/Windows fine from harddisk.
But kept crashing on 3 different SSD's, until now CROSSING MY FINGERS SO HARD THEY START TURNING PURLE
Still running fine, no hickups or crashes.
Aix1, did you test the kernel param on your machine already?
I am running into what seems like a similar issue, I am attempting to install Ubuntu 10.10 (64-bit) on a Crucial 256GB SSD. Currently I have a working install of Ubuntu running on the board off of a spare Hitachi 250GB drive however I ultimately want my primary boot partition on the SSD for performance reasons.
Tyan Thunder S2915-E
2 x Opteron 2378 / 32GB DDR667-RE
EVGA GeForce 275 GTX / Tuniq 1000W PSU
Crucial 256GB M225 SSD (boot)
ARC-1120 PCI-X / 3 x 2TB Hitachi 7200 RAID5 (data)
I am having persistent issues getting Ubuntu to recognize (correctly) the Crucial 256GB SSD. Although the SSD is recognized in the BIOS the Ubuntu installer fails to register it as a viable partition on it's first pass. If I reflash the BIOS of the SSD (using the "reflash" switch on front) the drive is now recognized as a 256.1 GB partition (which I think is erroneous, it should be more along the lines of 238 GB).
I've found a thread that seems to indicate that this might be due to a bug:
http://ubuntuforums.org/showthread.p...95#post8690795
...but I'm having some trouble implementing an immediate solution from the details of the thread - at least, I don't really know how to pass the kernel boot parameters described in the thread, if indeed this is the workaround the author is suggesting.
Any advice?
Lenovo Thinkpad W500 / Ubuntu 10.10 64-bit
Intel Core 2 Duo T9600 / 6GB PC3-8500
Intel GMA 4500WHD / ATI FireGL V5700 512MB
80GB Intel SSD
Go a few posts up and apply the kernel parameter.
I'm still running well on SSD without any problems.
nano /etc/default/grubI added the kernel parameter libata.force=noncq after quiet splash.
After update-grub (all with sudo or root)
So new line reads:
And with the installer, add the option before you boot the installer.GRUB_CMDLINE_LINUX_DEFAULT="quiet splash libata.force=noncq"
Last edited by Bas232; February 28th, 2011 at 07:01 AM.
And if you try it, please let us know whether the kernel parameter helped. Thanks.
have the same issue with Vertex3 in ubuntu 12.04
i'll try to apply kernel parameter when I get home, but I also have two hdd drives apart from the ssd and I'm afraid it'll negatively affect hdd performance. is it possible to disable ncq only for ssd?
Bookmarks