Interesting SSD speed issue

[Jerry1]

I got an external SSD on Prime Day. I'm going to use it for DW's machine where she stores her pictures. She's currently using a WD 1TB drive. I've been backing that up to my main computer using backup software, but I also clone that drive to an identical WD drive. So, I took her working drive and went to clone it onto the SSD. No problems. Then, I went ahead and made another clone of her working drive to the other WD drive since it had been awhile since I last backed it up. Now I have three identical drives. No problem.

However, when I was cloning the first time which was WD 1TB HD to the SSD, the speed of copying was about 50 to 60 MB/s. When I cloned the same WD 1TB HD to the other identical WD drive, the speed of copying was just over 100 MB/s. Why or how could it be that HD to HD is faster than HD to SSD?

Besides the drives being different, the SSD drive has a cable that is USB-C (plugged into the drive) to USB 3.0 (plugged into the computer). The WD hard drive has a USB 3.0 cable also going into the computer, but the end going into the drive is Micro B. Is the cable making the difference? I thought USB-C was supposed to be really fast. I've had Micro B ended cables on drives now for quite a few years. Not new technology.

Any ideas?

[HI Bill]

Assuming your USB port is really USB 3 (blue plastic inside), then I'd rack it up to Windows 10 (assuming that's what you're using). My transfer rates to SSD from HDDs range from 10 kbps to 190 mbps. Disconnecting the wi-fi and turning off the antivirus seems to help, most of the time.

[ERD50]

I agree that the USB-C should be as fast/faster than USB 3.0.

But then devil is in the details. How much RAM buffer does each drive have? A flash/SSD writes in 'blocks'. Even if you change one byte, it has to re-write everything in the block. So it might depend on how the copy software is controlling the writing.

A drive may have every different rates for one long data stream, versus one short data, or a series of short data writes. But "access times" for SSD are usually good, no waiting for a mechanical head to move/settle.

Do you have any specs for those drives? It might not tell the whole picture, but it might give a clue.

edit/add: what HNL_Bill said about the blue plastic. My laptop has 3 USB ports, but only one s blue (3.0). Don't know about the other stuff, couldn't hurt to minimize other resources.

-ERD50

[Oz investor]

on your 'spinning rust ' ( physical hard-drive ) there is some fairly efficient hardware inside and a sizable memory buffer .. , multiple heads for reading etc. ( hard drive speeds very consireably between styles/colours at WD )

solid state media has a reputation of quick reading but much slower writing ( or rewriting )

https://www.westerndigital.com/produ...#hard-disk-hdd

[Jerry1]

My USB ports are the Blue USB 3.0 ports.

I can't find any technical specs on the WD other than size and what comes with it. No data transfer spec. However, the SSD has a data transfer rate of 540 MB/s maximum.

I'm thinking it may indeed be the details. Is there a program or app that can just test each drive individually as is done with internet speed? If may also be the cloning software (AOMEI). Just seems wrong. It would suck to spend that much more for 1TB of storage (about $80 more than a HD) and actually have a slower drive in real terms. However, not all would be lost because one reason for wanting the disk (the SSD) is that is should prove a little more durable - less susceptible to damage from being dropped, etc.

[davebarnes]

To learn more, a lot more, about this, you need to attend https://flashmemorysummit.com/Englis...ance_2019.html

Please publish your trip report here.

[ERD50]

I think Oz_investor is on the right track. Your " transfer rate of 540 MB/s maximum" is undoubtedly read speed, so they don't even spec the write speed?

And I always get a kick out of a spec like " transfer rate of 540 MB/s maximum". So if it ever happened to transfer at 541 MB/s, you could ask for your money back? A maximum speed guarantee, but what you really want is a must meet some minimum speed guarantee!

A search should bring up some apps for testing speed, but you can do it yourself with a large file. But again, a large file versus a bunch of small files can be way different. The testing apps probably have a mix.

-ERD50

[Jerry1]

Quote:

Originally Posted by davebarnes To learn more, a lot more, about this, you need to attend https://flashmemorysummit.com/Englis...ance_2019.html Please publish your trip report here.

Thanks, but no thanks. I had enough of computer/IT in college to know that it's not my thing. I appreciated it enough that I could converse with people who worked for me, but to go to a seminar at this point. Rather go to the beach.

[target2019]

ATTO Disk Benchmark might be worth a try. I just installed it and ran a test on C:
https://www.atto.com/disk-benchmark/

You can change the parameters to what you need, then run, and save the report for comparison.

There are other tools out there, but you need to read carefully, so you don't download something malware-ish.

[Jerry1]

Thanks.

[boog]

Two thoughts. All USB-C cables are not alike. Some are inferior and can’t support full speed. If possible, try another cable on the SSD drive. I wasn’t getting the expected speed on my SSD added to my Mac and discovered that TRIM wasn’t enabled by default in the operating system. After enabling, I saw the expected speed. I think modern Windows enables TRIM by default but you can check to make sure. You can Google how to check.

[Jerry1]

Quote:

Originally Posted by boog Two thoughts. All USB-C cables are not alike. Some are inferior and can’t support full speed. If possible, try another cable on the SSD drive. I wasn’t getting the expected speed on my SSD added to my Mac and discovered that TRIM wasn’t enabled by default in the operating system. After enabling, I saw the expected speed. I think modern Windows enables TRIM by default but you can check to make sure. You can Google how to check.

The cable came with the drive. I certainly hope they didn’t cheap out there.

The computer I have has an SSD as the main drive. I certainly hope that it would know to enable TRIM, but I will check. Is TRIM something that would be drive specific? Or, once on, would it be on for any SSD that the operating system encounters.

[target2019]

Learned something new, and it isn't 8am yet!

https://www.windowscentral.com/how-e...sd-performance

So, DisableDelete is off, meaning TRIM is on. I don't have any USB drives plugged in at the moment.

PS C:\WINDOWS\system32> fsutil behavior query DisableDeleteNotify
NTFS DisableDeleteNotify = 0  (Disabled)
ReFS DisableDeleteNotify = 0  (Disabled)
PS C:\WINDOWS\system32>

[braumeister]

Quote:

Originally Posted by boog All USB-C cables are not alike.

Not just USB-C cables either.
I've had problems that were solved by getting a new, higher quality Ethernet cable as well. Cables are probably the last thing most people suspect of causing a problem, but the easiest and cheapest fix in some cases.

[SecondCor521]

Quote:

Originally Posted by Jerry1 The cable came with the drive. I certainly hope they didn’t cheap out there. The computer I have has an SSD as the main drive. I certainly hope that it would know to enable TRIM, but I will check. Is TRIM something that would be drive specific? Or, once on, would it be on for any SSD that the operating system encounters.

TRIM is an ATA command that is defined in a particular revision of the ATA specification (and later). Most SSDs in the last several years support the command, meaning they will respond properly if the OS sends the command.

TRIM is not "on" or "off". TRIM is a command that the host can send to the drive to say, "Hey, as far as I'm concerned, sectors 1, 5, 287, and 3,784 are empty, so you can do whatever you want with them." The drive can then essentially garbage collect those sectors, which can be advantageous for various reasons, but generally means the drive can perform more efficiently.

Whether the OS sends the command is something that is up to the OS to decide. The OS would have to decide that the drive was an SSD [1] and that it supports the TRIM command [2] and the OS has to decide to send the command periodically. Recent versions of Windows and *nix support the TRIM command, but google your specific OS for the particulars of whether and how it supports the command.

Since the command can be drive-by-drive specific, the OS could send TRIM commands to drives that support the command and not send it to drives that don't support it. The OS could even send TRIM commands to drives that don't support the command. That would be somewhat nonsensical but I think according to the ATA spec the drive would just abort the command and do nothing; the drive would probably not be damaged.

[1] Usually this is done by the OS noticing a "0" in the RPM field in the IDENTIFY command response.

[2] I think the IDENTIFY command response data also indicates support for various levels of ATA specifications. This way the host knows what commands the drive understands. All ATA drives are required to support the IDENTIFY command.

[SecondCor521]

Quote:

Originally Posted by ERD50 A flash/SSD writes in 'blocks'. Even if you change one byte, it has to re-write everything in the block. So it might depend on how the copy software is controlling the writing.

SSD flash drives typically write data in pages, which are 8KB, 16KB, or 32KB depending on the underlying flash being used. Blocks consist of 256 or 512 or 1024 pages. Erasing is done at the block level. So changing one byte would generally result in writing one new page, although the OS would probably cache the write in some way.

I'm not 100% certain, but most copy programs are probably going to copy in either the smaller of the source or destination page size, or the native OS read/write size, or maybe some least common multiple of those three numbers. A while ago Windows was a native 4KB read/write, but I think there was talk/hope of it going to 16KB. At any rate, most drives should be able to handle sustained sequential 4/8/16/32KB reading and writing quite well.

[SecondCor521]

Oh, and as to the original question, the only thing that makes sense is that the cable to the SSD is bad. I'd try swapping cables and/or running a disk benchmark (ATTO as someone else mentioned is decent. There are a few others but I've forgotten their names.)

[Oz investor]

Quote:

Originally Posted by boog Two thoughts. All USB-C cables are not alike. Some are inferior and can’t support full speed. If possible, try another cable on the SSD drive. I wasn’t getting the expected speed on my SSD added to my Mac and discovered that TRIM wasn’t enabled by default in the operating system. After enabling, I saw the expected speed. I think modern Windows enables TRIM by default but you can check to make sure. You can Google how to check.

previously i have had a hobby in audio/music and in computer servers and can vouch cables can vary greatly ( sometimes individully in 'identical ' cables
), i would assume external USB cables would be no exception .

also do some research on the computer hardware you would shocked at where i have been severely disappointed ( where makers have cut corners )

some 'smart' controllers have proved to be dumber than dirt ( and it was nothing to do with drivers/software/BIOS , the software was doing ALL the work ) .

ALSO check your drive has a funtional Error-Correcting Chip ( ECC ) that can make a BIG difference

[explanade]

Speeds vary among SSDs.

Also SSD may be faster than the USB bus.

I bought a 512 GB SSD and speeds varied when trying different ports and copy scenarios.

[Oz investor]

Quote:

Originally Posted by explanade Speeds vary among SSDs. Also SSD may be faster than the USB bus. I bought a 512 GB SSD and speeds varied when trying different ports and copy scenarios.

i was experimenting with a ( quality ) SSD internal drive and it was lightning fast at reading ( plugged in to the normal harddrive bus ) , reading/writing/re-writing ( as you would with a normal hard drive in use ) was not so impressive , however this was a couple of years back and SSD durability ( life-cycle ) still had some unknowns .