Very slow i/o to external hard drive's HFS+

Very slow i/o to external hard drive's HFS+

[Rob Stevens]

Hello

I have a removable hard drive sporting an HFS+ connected to a USB3 port.
The drive is a Seagate Slim Line 2T.
I am using a driver from Paragon Sofware to write/read from this drive.
I am running Windows10 build 19041.508 (ie up to date)

Writing to this file system using the unix cp command is incredibly slow:

cp my_file /cygdrive/h/some_destination
The same write, using copy paste in Windows Explorer is much faster.

I have not precisely timed it, but it is way more than an order of magnitude
faster (a 1G file took approx 10 mins: the graphical display shows just
a few seconds).

(In Windows Device Manager I have the policy for the drive set to be
"Better Performance"  and not "Quick Removal" -- apparently
the latter can lead to dramatic degradation in performance).

Can anyone suggest what might be the cause?

Rob Stevens

Re: Very slow i/o to external hard drive's HFS+

[Thomas Wolff]

Am 16.09.2020 um 17:16 schrieb Rob Stevens:
> Hello
>
> I have a removable hard drive sporting an HFS+ connected to a USB3 port.
> The drive is a Seagate Slim Line 2T.
> I am using a driver from Paragon Sofware to write/read from this drive.
> I am running Windows10 build 19041.508 (ie up to date)
>
> Writing to this file system using the unix cp command is incredibly slow:
>
> cp my_file /cygdrive/h/some_destination
> The same write, using copy paste in Windows Explorer is much faster.
>
> I have not precisely timed it, but it is way more than an order of magnitude
> faster (a 1G file took approx 10 mins: the graphical display shows just
> a few seconds).
>
> (In Windows Device Manager I have the policy for the drive set to be
> "Better Performance"  and not "Quick Removal" -- apparently
> the latter can lead to dramatic degradation in performance).
>
> Can anyone suggest what might be the cause?
"Better Performance" means that the write cache is active, so that the 
copy could be reported finished before it actually is. So your 
measurement may be deceptive. I assume cp applies some final flush, so 
its result would be real while the Windows copy isn't. Please try to 
change that setting for a reliable measurement.
Thomas