[SOLVED] Very slow HDD got 2x faster when I switched its power connector. Bad PSU?

[JamesFallon]

~4 years ago a Win10 Inspiron PC got slow after I physically moved it across town. I eventually realized the HDD was the culprit; it was getting only about 4MB/s write speed and could have easily been damaged in the move.

The PC has been a reliable media server ever since, just very slow.

Fast forward to today when I finally got a replacement HDD. But when I plugged it in, it ALSO got only 4MB/s write speed. Swapped SATA cables. No change. The new drive is also getting 4MB/s write speed, same as the old drive.

On a whim I plugged the other power connecter from the same power cable into the HDD (from P3 to P5 in the picture below). That caused write speed to jump to ~8MB/s. WTF?

8MB/s is still horrendously slow, but how could a different power connector on the same power cable cause a disk speed increase? Does that indicate a PSU problem? The PC has been fine in the years since the disk slowness first occurred; no unexpected shutdowns that I'd normally associate with a bad PSU. I'm struggling to make sense of this. Other than swapping PSUs (I don't have another one), ideas for how to figure out what's actually broken here?

 

[kerberos_20]

~4 years ago a Win10 Inspiron PC got slow after I physically moved it across town. I eventually realized the HDD was the culprit; it was getting only about 4MB/s write speed and could have easily been damaged in the move.

The PC has been a reliable media server ever since, just very slow.

Fast forward to today when I finally got a replacement HDD. But when I plugged it in, it ALSO got only 4MB/s write speed. Swapped SATA cables. No change. The new drive is also getting 4MB/s write speed, same as the old drive.

On a whim I plugged the other power connecter from the same power cable into the HDD (from P3 to P5 in the picture below). That caused write speed to jump to ~8MB/s. WTF?

8MB/s is still horrendously slow, but how could a different power connector on the same power cable cause a disk speed increase? Does that indicate a PSU problem? The PC has been fine in the years since the disk slowness first occurred; no unexpected shutdowns that I'd normally associate with a bad PSU. I'm struggling to make sense of this. Other than swapping PSUs (I don't have another one), ideas for how to figure out what's actually broken here?

that sounds fishy
well technically, drive receives voltage to spin motor, if voltage is lower, motor would run slower
tbh that PSU seems to be quite old, orange cable on sata power has been removed ages ago

 

[fzabkar]

Are these SMR drives?

BTW, spin speed is regulated by the motor controller, so lower voltages would not affect this. In any case, if the read speed is normal, then it can't be an RPM issue.

 

[arif1752002m4a1]

Well same thing I have seen with my WD Blue (10years old) WD3200AAJS HDD.... Actually what is happening.. older the drive gets.. more power it requires to spin..
So if your PSU is bad.. (if it's 12v is less than 11.8volts and 5v is less than 4.9 volts) you can see these kind of issues... With some drives (like 10-8% Disks only)
And if your HDD is going out.. same thing can be seen too... So I guess check the voltages of PSU... And if it's good.. time. To change your Drive.. (it's not proven fact just I have seen this so am telling.. if anyone knows exact answer... Surely I would like to see that too)

 

[Grobe]

I think for now, it's only speculation whether the motherboard, CPU or PSU is to blame for this.

I think that only by physically replace the PSU, you'd get an definitive answer if the PSU was indeed faulty or not.

Anyway - what OS do you run on that computer, still W10? Can you provide the HDD s.m.a.r.t. data and post it here (you can use the program CrystalDiskInfo, coy to clipboard, then paste here and put code tag).

 

[JamesFallon]

Are these SMR drives?

Original HDD is not, I believe. It is a 1TB Seagate from ~2015, which I think is before SMR was being put on 1TB consumer drives. But new HDD (BarraCuda 2TB) is, according to https://www.seagate.com/internal-hard-drives/cmr-smr-list/

I'd never heard of SMR before your question, but after reading https://linuxreviews.org/Shingled_magnetic_recording ... looks terrible. I might return the new HDD. Could SMR storage be the bottleneck if the network is reading files at 20MB/s for 20 minutes and storing them to disk? Seems like maybe. That would be unfortunate.

Anyway, I double checked my observations about switching power connectors sped up the disk. After doing so, I was completely wrong about my current disk speed. Today I ran the default CrystalDiskMark tests on the old HDD. Do these speeds indicate a potential disk issue or are they plausibly normal?

[Read]
Sequential 1MiB (Q=  8, T= 1):   154.920 MB/s [    147.7 IOPS] < 53753.89 us>
Sequential 1MiB (Q=  1, T= 1):   152.667 MB/s [    145.6 IOPS] <  6856.26 us>
    Random 4KiB (Q= 32, T=16):     0.701 MB/s [    171.1 IOPS] <690290.99 us>
    Random 4KiB (Q=  1, T= 1):     0.314 MB/s [     76.7 IOPS] < 13015.13 us>

[Write]
Sequential 1MiB (Q=  8, T= 1):   107.588 MB/s [    102.6 IOPS] < 75112.42 us>
Sequential 1MiB (Q=  1, T= 1):    81.972 MB/s [     78.2 IOPS] < 12765.85 us>
    Random 4KiB (Q= 32, T=16):     0.760 MB/s [    185.5 IOPS] <583862.15 us>
    Random 4KiB (Q=  1, T= 1):     0.751 MB/s [    183.3 IOPS] <  5441.90 us>

Profile: Default
   Test: 1 GiB (x5) [Interval: 5 sec] <DefaultAffinity=DISABLED>
   Date: 2021/10/26 2:56:42
     OS: Windows 10  [10.0 Build 19043] (x64)

My old estimation method: when computer was being slow I'd open Task Manager processes tab and see which of CPU, Memory, and Disk was pegged. It was always Disk at 100%, and the processes with the highest Disk speed would usually add up to about 4MB/s.

New HDD has ubuntu on it, so when it was being slow installing a few GB of packages, I ran iostat -dx 5 which showed the following. Back-of-the-envelope calculation of wkB/s / %util gave ~3-5.5MB/s. Roughly similar to 4MB/s. Perhaps meaningless though?

Linux 5.4.0-89-generic (ara2)     10/24/2021     _x86_64_    (4 CPU)

Device            r/s     rkB/s   rrqm/s  %rrqm r_await rareq-sz     w/s     wkB/s   wrqm/s  %wrqm w_await wareq-sz     d/s     dkB/s   drqm/s  %drqm d_await dareq-sz  aqu-sz  %util
sdb             12.44    591.26     3.09  19.91    4.54    47.54   60.71   3104.79    75.77  55.52   26.95    51.14    0.00      0.00     0.00   0.00    0.00     0.00    1.60  66.33

Device            r/s     rkB/s   rrqm/s  %rrqm r_await rareq-sz     w/s     wkB/s   wrqm/s  %wrqm w_await wareq-sz     d/s     dkB/s   drqm/s  %drqm d_await dareq-sz  aqu-sz  %util
sdb              3.20    360.00     0.00   0.00    2.50   112.50  483.40   3985.60    55.20  10.25    4.72     8.24    0.00      0.00     0.00   0.00    0.00     0.00    1.90 100.08

Device            r/s     rkB/s   rrqm/s  %rrqm r_await rareq-sz     w/s     wkB/s   wrqm/s  %wrqm w_await wareq-sz     d/s     dkB/s   drqm/s  %drqm d_await dareq-sz  aqu-sz  %util
sdb              0.80     38.40     0.00   0.00    2.75    48.00  757.20   3351.20    87.20  10.33    2.21     4.43    0.00      0.00     0.00   0.00    0.00     0.00    0.68  99.76

Anyway - what OS do you run on that computer, still W10? Can you provide the HDD s.m.a.r.t. data and post it here (you can use the program CrystalDiskInfo, coy to clipboard, then paste here and put code tag).

Yeah, old HDD has Win10 still. CrystalDiskInfo below. Does anything stand out at you?

----------------------------------------------------------------------------
CrystalDiskInfo 8.12.12 (C) 2008-2021 hiyohiyo
                                Crystal Dew World: https://crystalmark.info/
----------------------------------------------------------------------------

    OS : Windows 10  [10.0 Build 19043] (x64)
  Date : 2021/10/25 16:58:25

-- Controller Map ----------------------------------------------------------
 + Intel(R) 8 Series/C220 Chipset Family SATA AHCI Controller [ATA]
   - ST1000DM003-1ER162
 - Microsoft Storage Spaces Controller [SCSI]

-- Disk List ---------------------------------------------------------------
 (01) ST1000DM003-1ER162 : 1000.2 GB [0/0/0, pd1]

----------------------------------------------------------------------------
 (01) ST1000DM003-1ER162
----------------------------------------------------------------------------
           Model : ST1000DM003-1ER162
        Firmware : CC45
   Serial Number : Z4Y8732D
       Disk Size : 1000.2 GB (8.4/137.4/1000.2/----)
     Buffer Size : Unknown
     Queue Depth : 32
    # of Sectors : 1953525168
   Rotation Rate : 7200 RPM
       Interface : Serial ATA
   Major Version : ACS-2
   Minor Version : ACS-3 Revision 3b
   Transfer Mode : SATA/600 | SATA/600
  Power On Hours : 50520 hours
  Power On Count : 382 count
     Temperature : 36 C (96 F)
   Health Status : Good
        Features : S.M.A.R.T., APM, NCQ
       APM Level : 8080h [ON]
       AAM Level : ----
    Drive Letter : C:

-- S.M.A.R.T. --------------------------------------------------------------
ID Cur Wor Thr RawValues(6) Attribute Name
01 115 _99 __6 000005593D10 Read Error Rate
03 _97 _97 __0 000000000000 Spin-Up Time
04 100 100 _20 00000000017E Start/Stop Count
05 100 100 _10 000000000000 Reallocated Sectors Count
07 _84 _60 _30 000013498B8E Seek Error Rate
09 _43 _43 __0 00000000C558 Power-On Hours
0A 100 100 _97 000000000000 Spin Retry Count
0C 100 100 _20 00000000017E Power Cycle Count
B7 100 100 __0 000000000000 Vendor Specific
B8 100 100 _99 000000000000 End-to-End Error
BB 100 100 __0 000000000000 Reported Uncorrectable Errors
BC 100 100 __0 000000000000 Command Timeout
BD 100 100 __0 000000000000 High Fly Writes
BE _64 _50 _45 000026150024 Airflow Temperature
BF 100 100 __0 000000000000 G-Sense Error Rate
C0 100 100 __0 000000000012 Power-off Retract Count
C1 _98 _98 __0 00000000159C Load/Unload Cycle Count
C2 _36 _50 __0 000F00000024 Temperature
C5 100 100 __0 000000000000 Current Pending Sector Count
C6 100 100 __0 000000000000 Uncorrectable Sector Count
C7 200 200 __0 000000000000 UltraDMA CRC Error Count
F0 100 253 __0 C2190000C55B Head Flying Hours
F1 100 253 __0 00131356E4E2 Total Host Writes
F2 100 253 __0 01A50F4890AA Total Host Reads

 

[kerberos_20]

4k test is a bit slow, but smart seek error rate is high, that explains it...that is if drive misshits, so it has to try again
it doesnt mean drive is bad, no data loss from this, just some performance is lost

 

[fzabkar]

The SMART report looks clean. The number of seek errors is actually 0.

http://www.hddoracle.com/viewtopic.php?f=6&t=89