8 Replies Latest reply on Aug 29, 2010 9:14 PM by BizSAR

    DP55WB BIOS bug: Samsung Spinpoint F3

    frans

      I have a DP55WB with latest BIOS 0144. Attached are 4 Samsung SpinPoint F3 drives (HD103SJ) in RAID10 config.

      Right after boot (into either Windows 7 x64 or openSuse 11.1 x64) all drives frequently click loudly (about 2 times per minute, almost synchronous together).

      I verified using smartctl under Linux that it is a head unloading/loading cycle (increasing SMART attribute 225).

      With 2 head unload/load cycles per minute these drive will wear excessively.

       

      Fortunately I found a "workaround". I noticed that the Intel Matrix Storage Technology *does* work correctly with these drives after a sleep/resume cycle.

       

      So to boot my system: I wait for the OS to load (I have heard 2 or 3 clicks by now), then put the system to sleep, and then wake it up again -> clicks are gone.

        • 1. Re: DP55WB BIOS bug: Samsung Spinpoint F3
          frans

          Update:

           

          - problem persists after BIOS update 0165

          - problem 100% reproducable for: cold boot, warm reboot

          - "workaround" 100% reproducable for: wake from hybrid sleep

           

          It looks like the BIOS uses a drive initialization procedure that is incompatible with these drives, but the BIOS code path for wake from sleep (which of course is a different code path than for a full reboot) is correct and the problem disappears. It is also clear that these drives *can* work correctly with this board and that it is essentially a software issue. It would be nice if the BIOS would be fixed so that the awkward boot procedure described earlier is no longer necessary.

          • 2. Re: DP55WB BIOS bug: Samsung Spinpoint F3
            ophir

            Have you been able to fix this?

            • 3. Re: DP55WB BIOS bug: Samsung Spinpoint F3
              frans

              Not entirely (but there is a workaround).

               

              The board will not initialize the drives correctly on cold or warm boot. The drives will work, but they will park their heads every minute (unload/load cycle). But this is *is* sufficient to boot the operating system. The Windows 7 RAID drivers (Matrix Storage Manager 8.9, but also the beta Rapid Storage Technology 9.5) will not correct the problem.

               

              But after a sleep/wake cycle the board *does* initialize the drives correctly, and the repeated head unload/load cycles are gone. At this point the drives (which I use in RAID10 mode) are perfect: >200 MiB / sec performance, no hick-ups.

               

              So the trick is to do a sleep/wake cycle right after boot, then all is well.

              • 4. Re: DP55WB BIOS bug: Samsung Spinpoint F3
                frans

                Update: bug persists after BIOS update 0190

                 

                (This update mentions changes regarding the handling of certain drive models when connected to port 5 or 6. However, in my case the four drives are on port 1 through 4, and in fact the issue remains unchanged.)

                • 5. Re: DP55WB BIOS bug: Samsung Spinpoint F3
                  steppinwolf

                  Frans' experience sounds similar to a problem that occured on my system yesterday. I have two Samsung SpinPoint F3 drives in RAID-1 array. These are used for data, temp files, backups, etc. The OS is running on an SSD RAID-1 array.

                   

                  Yesterday, I heard one or more drives clicking (parking) and spinning up approx. once every 10-20 seconds. This cycle would repeat maybe 10 times (I'll try to count more accurately if it happens again). Then it would stop for perhaps a minute or two, then start cycling again. After about 10 minutes of this, RST reported a status of Warning for the SpinPoint drive on SATA2. Perhaps only this drive was cycling through park/spin-up?

                   

                  Unlike the behavior Frans reported, this has only happened once so far (is not consistant and repeatable) and was resolved by powering down and back on. After waiting a while and no longer hearing the drive cycling, I put SATA2 back to status normal and the array went through verify and repair (verification errors 0; media errors 0).

                   

                  This is a newly built system. The SpinPoints had been running without any problems for almost a month. Biggest change was updating chipset, BIOS and Intel Controller to latest versions (details below), but that was more than a week ago.

                   

                   

                  Motherboard: Intel DP55KG

                  CPU: Intel i7 860

                  HD (system): Intel X25-M G2 80GB SSD x 2 (RAID-1) on SATA0 and 1

                  HD (data): Samsung F3 HD103SJ 1TB x 2 (RAID-1) on SATA2 and 3

                  BIOS: KGIBX10J.86A.4507.2010.0105.1706

                  Chipset P55, INF version 9.1.1.1022

                  ICH Driver RST 9.5.0.1037

                  OS: Windows 7 Pro x64

                  • 6. Re: DP55WB BIOS bug: Samsung Spinpoint F3
                    frans

                    Thanks for your reply, as you are the first to mention issues with RST + F3 RAID besides me.

                     

                    As you said: this *could* be a related issue.

                     

                    Our setups are very simmilar (same drives, same processor, same OS, same ICH driver version).

                    Our boards are different (you KG, me WB), but still: same P55 series.

                     

                    On the other hand, the symptoms I experience are *very* reproducable, it happens *every* time after boot. And the drives do not spin down, they just go through a unload/load cycle for the heads (I have verified this by checking the SMART values, the load/unload cycle counter increases, but not the start/stop cycle counter). Besides these "clicks", the system is operating normally, no warnings or verification errors for the RAID array. And a simple sleep/wake cycle stops the clicks, also 100% reproducable.

                     

                    If the root cause is an incompatiblilty between the F3 firmware and Intel RST (which is what I suspect), then the issues might be related.

                     

                    Anyone else using Spinpoint F3's in RAID using a P55 board ?

                    • 7. Re: DP55WB BIOS bug: Samsung Spinpoint F3
                      steppinwolf

                      Frans, it occurred to me that some users may be getting symptoms without realizing it. As we know, the sounds are quiet enough that one might not hear these drives parking/unparking and spinning up if it's a noisy PC/environment or the PC isn't close enough to someone who would recognize these abnormal sounds. Some may only notice occasional response time delays or RAID status messages.

                       

                      Wish I could read SMART values, but it seems MSM and RST drivers don't provide access to SMART values for drives in an array.

                       

                      Also, Intel documentation indicates known issues with RAID and some sleep modes.

                       

                      I recently found something that could have been causing this problem on my system. In January, I modified Power Settings Balanced Plan to put the computer to sleep in 30 minutes. Recently, I checked Advanced power settings and discovered Hard Disk was set to turn off after only 20 minutes (obviously incompatible settings). I haven't heard the repeated parking/spinup behavior since increasing hard disk sleep timer to 45 minutes, Would be nice if Windows warned you or automatically adjusted hard disk timer to be >= computer sleep timer.

                      • 8. Re: DP55WB BIOS bug: Samsung Spinpoint F3
                        BizSAR

                        I'm having a similar issue with the F3 and my DX58SO, running the latest BIOS, version 5417. Every few minutes the F3 clicks, just like it would if it had gone to sleep and parked the heads. It causes a short hang in games, etc., then all is fine for another 3-5 minutes.

                        I noticed this after I enabled AHCI, installed a SSD (using the F3 for non-OS stuff), and re-installed Win 7 64-bit. Since then, I'm also seeing a cursor in the upper-left corner for ~20 seconds upon power up, but before POST. They might be related.

                         

                        I'll try the sleep workaround and update here.

                        UPDATE: Yup, going into sleep mode and back does indeed correct the problem.

                        Are there any updates to this issue?

                         

                        Message was edited by: BizSAR