How to determine whether a USB 3.0 device is operating at SuperSpeed? USB 3.0 introduces a new operating speed called SuperSpeed. Compared to USB 2.0 bandwidth of 480 Mbps, SuperSpeed supports 5.0 Gbps making it 10 times faster than USB 2.0. … USB 3.0 controllers are required to work with all existing USB devices.Nov. 9, 2020
As noted above, USB 3.0 SS ports are actually marked as such with the SS symbol. All of the ports on most recent desktops are 3.0, but usually only two of them are SS.
The USB 3.0 theoretical limit is 5.0GB/s but only SS ports actually achieve this.
To be honest from what I saw and read in some other threads SS doesn’t matter.
Whats way more important is if it’s either USB 3.0 Gen 1 or Gen 2.
USB 3.0 Gen 1 is rated at 5GB/s, the Gen 2 on the other Hand is archiving up to 10GB/s.
I think the SS you are talking about in every single thread on the forum doesn’t matter at all…
Just my personal opinion…
The reason I started using the SS designation in the first place was because that is how the products are usually marketed. Also, most PC’s SS ports are marked as such. If the product says Super Speed, SS, or 5GB/s then you are getting the right thing. If it just says USB 3.0, it is probably about 1GB/s. If it says 3.1 it may be SS but you don’t know for sure. If it says 3.2 it is almost certainly not compatible or is over provisioned compared to the rest of your system.
The 3.0, 3.1, and 3.2 meanings and names were all changed on the fly making it very difficult to use them for purchasing the right thing. 3.2 actually supports 10 and 20GB/s, but as far as I know, none of the standard arrays have better than SS, and I think 3.2 is mainly being used in apple world as, “Thunderbolt”.
USB 2.0 is about .5GB/s. Standard USB 3.0 is about 1 GB/s. USB 3.0 SS is 5GB/s. 5GB/s is what I recommend using throughout, whatever name you call it by.
When I connected my third array I had run out of marked SS ports and used a regular 3.0 port. My average response time immediately went way up and if writing a plot I was missing challenges. The final plot transfer took 2.5 hours instead of my usual 30 minutes.
I bought a USB 3.0 SS powered hub, plugged it in to one of my PC’s SS ports and plugged all of my arrays into the hub, My response and write times immediately returned to normal.
To check your drives, get Crystal Disk Info It will give you SMART info about all your drives. The lines to watch for are:
Read Error Rate
Reallocated Sectors Count
Seek Error Rate
Reallocation Event Count
Uncorrectable Sector Count
UltraDMA CRC Error Count
Actually, “Power-On Hours” is another param that is worth to check. All your drives should have about the same number of hours on them. Maybe that one drive is an old drive, and thus you see some odd things there.
Not all those attributes are on all drives, but just look for those with Errors there. Most of them should have all zeros on the far right side.
If you see anything that you don’t like, download SmartMonTools, It gives the same SMART data, but can also run tests (short - minutes, long - hours). Once the test is done, you will see some basic status at the end of the full report. Also, this utility is also available for Linux (originated as a Linux tool). I would really recommend running it.
If you see bad reports, rerun the test, as some errors are recoverable. If you will keep seeing those errors, time to get a new drive.
If you don’t see any errors, than those HDs are not the source of your problems.
By the way, that SmartMonTools is waking up drives to get SMART data. If the drive is active, the output is instantaneous. If you see a second/two delays, your drives were sleeping. Therefore, it can be used to periodically touch your drives to make them active without actually touching any platters.
If they are doing is reading the same few bytes from every file those will get cached, and yes, they will go to sleep until a request for new data (proof) comes.
it gets cached in ram like all IO on modern operating systems. You can probably turn off read caching somewhere, i never have and i wouldn’t recommend it. There are easier ways to keep your disks alive. Just write a txt file with a timestamp every 10 minutes to each drive if you are having issues with them going to sleep you cant solve with OS tools.
You can use that SmartMonTools I have mentioned above. If your drives are USB, you may need to run your command prompt with admin rights, or use some extra params (sometimes it gives headaches, when things don’t work). Also, not all USB adapters cooperate well with that tool.
smartctl --scan // shows you what drives you have
smartctl -a /dev/sdb // will give you a good info about your drive
smartctl -A /dev/sdb // will give you only SMART info, what is usually good enough to check on your drives from time to time
smartctl -n active /dev/sdb // will let you know the state of that device
smartctl -t short /dev/sdb // will run few minutes worth of test
smartctl -t long /dev/sdb // will take hours, but does a good job of testing your drives
A sample response from my drive
C:\>smartctl -n idle /dev/sdc
smartctl 7.3 2021-09-14 r5236 [x86_64-w64-mingw32-w10-21H1] (CircleCI)
Copyright (C) 2002-21, Bruce Allen, Christian Franke, www.smartmontools.org
Device is in ACTIVE or IDLE mode
My understanding is that as long as it is not in STANDBY mode, you are good, as IDLE to ACTIVE transition is short (sub-second ?).
Also, it may be that when you run that command, it will trigger your drive to move out of the STANDBY state, but I am not sure about that. I think that I have seen it when asking for all info (-a switch). So, potentially that -A switch can be used from a script to make sure that drives are not going to STANDBY.
It is one of the best tools for working with HDs. Runs a bit better on Linux, though.
Yep. If those drives would be sleeping, you would have those problems all the time, not just every other hour, or even less often.
One thing that I did is excluded all those drives from being scanned by Windows Security, and also from being indexed. Those processes are invoked from time to time and are taking a bit of CPU time. IMO there is no need to check those drives.
I’m just not convinced, maybe it’s correct.
My drives were set to sleep @ 20 mins, but best I can tell they never have done.
Edit. Actually mine were set to turn off and not just sleep.
I was going through man pages for that smartctl command. It looks really promising. You can run:
smartctl -g apm /dev/sdb // gives you the current apm level (1-254),
smartctl -s apm,128 /dev/sdb // sets the drive to the quiet mode, but prevents spinning down
smartctl -s apm,off / dev/sdb // disables APM, but the results are vendor specific - rather avoid that
I cannot say that I fully understand that, but maybe that -s apm,128 is enough to make sure that drives will not be sleeping, thus no need for any other methods to “keep them running.” Also, they will be in the “quiet” mode - less aggressive with heads movements when in IDLE, I think.
I think that this settings overrides whatever Windows sets (that 20 minutes power saving feature), but again, I am not sure about that. I will try to play with that.
This tool is an open source code for the past 100 years or so. Therefore, no reason to think that it may do something malicious.
