x
This website is using cookies. We use cookies to ensure that we give you the best experience on our website. More info. That's Fine
HPC:Factor Logo 
 
Latest Forum Activity

USB 2.0 vs 3.0

Rich Hawley Page Icon Posted 2015-02-06 8:31 PM
#
Avatar image of Rich Hawley
Global Moderator
H/PC Guru

Posts:
7,188
Location:
USA
Status:
So what gives?

As I understand it, USB 3.0 is supposed to be 10 times faster than 2.0.

I have two external 2.5 hard drives, exactly the same...both 250gb drives. One is in a 2.0 external case, the other in a 3.0 external case.

I just installed a PCI-E USB 3.0 card reader in my old Gateway E4610. All seems to work fine.

However, when I copy a 10gb folder from my desktop to the USB 2.0 drive, using a USB 2.0 port, I get transfer rates of about 25mb/sec.

Doing the same thing with the same folder, except using the USB 3.0 enclosure with the new 3.0 port, I only get transfer rates of about 61mb/sec.

How is this 10 times faster? It isn't...it is only a little better than double the speed....and I repeated the test several times.



(usb3.jpg)



Attachments
----------------
Attachments usb3.jpg (438KB - 0 downloads)
 Top of the page
C:Amie Page Icon Posted 2015-02-07 9:49 AM
#
Avatar image of C:Amie
Administrator
H/PC Oracle

Posts:
17,989
Location:
United Kingdom
Status:
Ah young Hawley, you have been duped by marketers I'm afraid. You're looking at the theoretical maximum line speed.

25 MB/s = 200Mbps
61MB/s = 488Mbps

USB 1 = 11Mbps
USB 2 = 480Mbps
USB 3 = 4000Mbps (ish)

SATAI = 150Mbps
SATAII = 300Mbps
SATAIII = 600Mbps

They are the theoretical maximum values supported by the standard, not the expected values supported by the standard. On top of the standards you have to factor
- The rotation speed of the disk
- The number of platters
- The number of heads
- The disk controller chipset
- How big the cache / write buffer is on the disk
- Whether write caching is enabled (which on a USB device is almost certainly will NOT be)
- Whether the drive is bus or mains powered
- How empty the drive is
- Whether the drive is fragmented
- Overheads from the file system
- Whether the underlying drive supports things like NCQ
- Overheads caused by USB cabling

You can throw 4Gbps of data at the hard drive, but I promise you that it can't swallow 4Gbps of data. With a 64MB write cache (which I doubt is turned on as it is USB) the disk cache at 4Gbps it will be full in less than 1.1 seconds while even a SATAIII drive running at SATAIII speeds (which doesn't exist) would take 8.4 seconds to clear that cache down to the spindle (assuming an over simplified design).

So what you actually have at 61Mbps is a drive performing somewhere between SATAII and SATAIII, basically at USB 2.0 speeds. Thus meaning that save for the extra 8Mbps that you obtained, USB 3 on hard drives' (n this case) wasn't worth it.
Stick a SSD in that drive caddy and we play a different game.
 Top of the page
Paianni Page Icon Posted 2015-02-07 11:39 AM
#
Avatar image of Paianni
H/PC Elite

Posts:
734
Location:
England, UK
Status:
As I was suspecting the performance of the magnetic media would surely be the weak link here.
 Top of the page
Rich Hawley Page Icon Posted 2015-02-07 1:18 PM
#
Avatar image of Rich Hawley
Global Moderator
H/PC Guru

Posts:
7,188
Location:
USA
Status:
Okay...guess I need to get the fastest USB 3.0 flash drive I can get and try it again...
 Top of the page
C:Amie Page Icon Posted 2015-02-07 2:47 PM
#
Avatar image of C:Amie
Administrator
H/PC Oracle

Posts:
17,989
Location:
United Kingdom
Status:
Again with a red alert, UFD flash != SSD

http://en.wikipedia.org/wiki/USB_flash_drive#Third_generation
In a typical well-conducted review of a number of high-performance USB 3.0 drives, a drive that could read large files at 68 MB/s and write at 46 MB/s, could only manage 14 MB/s and 0.3 MB/s with many small files. When combining streaming reads and writes the speed of another drive, that could read at 92 MB/s and write at 70 MB/s, was 8 MB/s.


An SSD can manage about 600MB/s
 Top of the page
Jump to forum:
Seconds to generate: 0.156 - Cached queries : 62 - Executed queries : 10