My USB 3.0 effective speed is only 50MB/s

[TijmenDal]

Hi,

Yesterday I bought a Seagate Expansion 3 TB drive as a second off-site backup disk. It's my first USB 3.0 drive, but I'm only getting 50MB/s speeds.

Now, I ran BlackMagic's Disk Speed test and here's the weird thing:
When my target drive is the HD inside my Macbook I'm getting ~40MB/s seconds, whereas when I select the external drive I'm getting ~130-150MB/s readings.
Obviously the bottle-neck is inside the computer, which is really weird. Can anyone help me out here?

Thanks a ton.

 

[domemvs]

What exact mbp model do you have?

 

[simonsi]

Assuming you are reading from your internal drive then that will likely limit the data transfer rate to the USB3 external....Blackmagic reads and writes files that are in RAM, not on a drive that would limit the results artificially.

 

[Steve121178]

Surely the speed depends on the type of data you are copying?

 

[g4cube]

If you are copying from your internal drive to the external drive, of course you are limited to the speed of the internal drive.

When you run a benchmark, speed of other devices are removed from the calculation, and in Blackmagic case, test file is read/written from RAM data which is orders of magnitude faster.

Also, copying many small files will be slower than copying few large files. When copying small files there are data writes to the file, and updating of the directory info and journaling info.

 

[TijmenDal]

If you are copying from your internal drive to the external drive, of course you are limited to the speed of the internal drive.

When you run a benchmark, speed of other devices are removed from the calculation, and in Blackmagic case, test file is read/written from RAM data which is orders of magnitude faster.

I get that I'm limited to the speed of the internal drive, but surely my internal drive is faster than 50MB/s, right?

Assuming you are reading from your internal drive then that will likely limit the data transfer rate to the USB3 external....Blackmagic reads and writes files that are in RAM, not on a drive that would limit the results artificially.

What do you mean by 'files that are in RAM'?

What exact mbp model do you have?

9,2

Surely the speed depends on the type of data you are copying?

It's through Time Machine. In 90 minutes I did about 270GB.

 

[simonsi]

I get that I'm limited to the speed of the internal drive, but surely my internal drive is faster than 50MB/s, right?



What do you mean by 'files that are in RAM'?



9,2



It's through Time Machine. In 90 minutes I did about 270GB.

A test tool such as Blackmagic uses it own test data to write to disk and read back but it moves the data to and from RAM to ensure it is measuring the disk speed as far as possible.

As you say, copy activity can only run at the speed of the slowest link in the chain, be it source drive, target drive or an interface inbetween.

Your internal testing at 50MB/s - not impossible for a relatively low density 5400rpm drive. Disk data speed is governed by its rotational speed and data density so for instance a 2TB 3.5" 5400 rpm has more data passing under the read/write heads per revolution of the disk compared to an otherwise-identical 1TB drive (the additional data capacity comes from having both more data tracks and more data per track). As there is more data passing under the heads the 2TB drive will both read and write data faster than its 1TB version.

A track is the circle of data a head can access in a single revolution of the disk platter, tracks are arranged as concentric rings on the platters and the heads move in and out to access each track. If a particular disk has equal data density then also the outer tracks on the disk can/will achieve a faster data rate as again there is more physical data passing under the heads per sec.

Now my 7200rpm internal 1TB achieves around 70MB/s IIRC so if you have a drive in the 500GB and 5400rpm your 50MB/s rate would sound about right.

 

[g4cube]

Slowing things further is Time Machine which adds another layer of complexity. TM does not write files and folders directly to the backup drive. Instead it mounts a disc image file, mounts the volume, and then writes to that virtual volume. There is additional time to calculate what files and folders need to be in the backup.

A tool such as SuperDuper or Carbon Copy Cloner might indicate higher thruput when making a backup.

 

[TijmenDal]

Thanks a lot for the responses guys! Of course it's still better than the measly 20-25MB's I was getting with my old USB drive. Sounds like it's time to put an SSD in my laptop.