hi,
some surface tests cannot be performed or are very slow when using more than 1 drive at a time
testing on 2 drives:
*Read = ok (speed 150 MB/s)
*Write = ok
*Write+read = ok
*Read+write+read = sloow (speed only 20 MB/s)
*Re-init = speed 20 MB/s
*Disk repair = once 2nd drive starts , (only) that one gives access denied al the time
options like "extensive read write test" and "drive lock" do not seem to make any difference
i think i saw a post saying disk repair cannot run on >1 drive....
i temporarily switched to a lsi hba controller to eliminate the mobo controller.
are these speeds normal?
(using v5.61.12B/prof)
Surface Test : some tests are very slow (multi drive)
-
hdsentinel
- Site Admin
- Posts: 3115
- Joined: 2008.07.27. 17:00
- Location: Hungary
- Contact:
Re: Surface Test : some tests are very slow (multi drive)
Yes, this is completely normal and expected situation: multiple disk drives connected to a disk controller (or the motherboard integrated controller) share the total available bandwidth.
You can very easily verify: if you start multiple disk tests and you see a test slowing down then (in the other window) click on any small square of the disk surface map to view the contents of the sector. While that window is open, that particular test is paused - so you may notice that the other disk test will be faster.
With 1-2 hard disks, this is not really noticeable, but if you'd test higher number of hard disks (or especially SSDs) on the same controller, then it may be noticeable. Then yes, using a different controller for some drives (and/or by connecting via USB docking stations) can help.
The Re-init disk surface test (as it is a very intensive test, performing multiple overwrite passes before clearing data and verification) would require approximately 5 x time to complete compared to a simple read test. With other words, if your disk drive would perform with 150 MByte/sec in the Read test, then the Re-init disk surface test would perform around 30 MByte/sec only (even if only that particular disk would be tested, so it does not really depend on other disks/tests).
The Disk Repair test can be used on 1 (one) disk drive at any given time (yes, probably mentioned in a different topic). During the Disk Repair, you can launch any number of other tests on other disk drives - but the Disk Repair test can be used on a single drive only.
> Options like "extensive read write test" and "drive lock" do not seem to make any difference
The "Extensive read+write+read test" can make difference: it controls the behaviour of the Read+Write+Read (Refresh disk area) test. It enables an additional write step before writing back the actual data, so disabling that option could speed up the procedure somehow.
The "Drive Lock" option attempts to lock the drive for the duration of the test (to prevent other software and Windows accessing the drive). Yes, this does not increase the speed of the test.
> i temporarily switched to a lsi hba controller to eliminate the mobo controller.
Yes, this can be a good idea.
> are these speeds normal?
Yes, I can confirm that these are completely normal speeds. Please check the Help for detailed information about the different type of tests:
https://www.hdsentinel.com/help/en/61_surfacetest.html
These describe the steps performed by the different tests. Both the Read+Write+Read test and the Re-init disk surface test functions perform multiple read/write operations on each sectors, just like if you'd perform multiple complete read/write tests. So yes, if a simple Read test performs with 150 MByte/sec, then you can expect 20-30 MByte/sec from the Read+Write+Read test and slightly lower from the Re-init disk surface test.
If you prefer higher speed, you can:
1) select Disk menu -> Surface test, select the disk drive to be tested and the test type.
2) then in the Configuration tab, you can adjust options of the the disk tests, for example
- disable the Extensive read+write+read test option for the Read+write+test. This may increase the speed with about 20-30%
- lower the Surface reinitialization level to 2 or 1, to increase speed with 20-40%
Or you can use the simple Disk menu -> Surface test -> Write test, which performs the fastest complete erase: this test fills all sectors with zeroes (but on the mentioned Configuration tab, you can also adjust the write pattern).
You can very easily verify: if you start multiple disk tests and you see a test slowing down then (in the other window) click on any small square of the disk surface map to view the contents of the sector. While that window is open, that particular test is paused - so you may notice that the other disk test will be faster.
With 1-2 hard disks, this is not really noticeable, but if you'd test higher number of hard disks (or especially SSDs) on the same controller, then it may be noticeable. Then yes, using a different controller for some drives (and/or by connecting via USB docking stations) can help.
The Re-init disk surface test (as it is a very intensive test, performing multiple overwrite passes before clearing data and verification) would require approximately 5 x time to complete compared to a simple read test. With other words, if your disk drive would perform with 150 MByte/sec in the Read test, then the Re-init disk surface test would perform around 30 MByte/sec only (even if only that particular disk would be tested, so it does not really depend on other disks/tests).
The Disk Repair test can be used on 1 (one) disk drive at any given time (yes, probably mentioned in a different topic). During the Disk Repair, you can launch any number of other tests on other disk drives - but the Disk Repair test can be used on a single drive only.
> Options like "extensive read write test" and "drive lock" do not seem to make any difference
The "Extensive read+write+read test" can make difference: it controls the behaviour of the Read+Write+Read (Refresh disk area) test. It enables an additional write step before writing back the actual data, so disabling that option could speed up the procedure somehow.
The "Drive Lock" option attempts to lock the drive for the duration of the test (to prevent other software and Windows accessing the drive). Yes, this does not increase the speed of the test.
> i temporarily switched to a lsi hba controller to eliminate the mobo controller.
Yes, this can be a good idea.
> are these speeds normal?
Yes, I can confirm that these are completely normal speeds. Please check the Help for detailed information about the different type of tests:
https://www.hdsentinel.com/help/en/61_surfacetest.html
These describe the steps performed by the different tests. Both the Read+Write+Read test and the Re-init disk surface test functions perform multiple read/write operations on each sectors, just like if you'd perform multiple complete read/write tests. So yes, if a simple Read test performs with 150 MByte/sec, then you can expect 20-30 MByte/sec from the Read+Write+Read test and slightly lower from the Re-init disk surface test.
If you prefer higher speed, you can:
1) select Disk menu -> Surface test, select the disk drive to be tested and the test type.
2) then in the Configuration tab, you can adjust options of the the disk tests, for example
- disable the Extensive read+write+read test option for the Read+write+test. This may increase the speed with about 20-30%
- lower the Surface reinitialization level to 2 or 1, to increase speed with 20-40%
Or you can use the simple Disk menu -> Surface test -> Write test, which performs the fastest complete erase: this test fills all sectors with zeroes (but on the mentioned Configuration tab, you can also adjust the write pattern).