[gentoo-user] {OT} USB 3.0 hard drive speed test

[gentoo-user] {OT} USB 3.0 hard drive speed test

[Grant]

I'm testing this USB 3.0 bus-powered hard drive:

http://www.amazon.com/gp/product/B0041OSQ9S

and I get:

# hdparm -tT /dev/sdb
/dev/sdb:
Timing cached reads:   8006 MB in  2.00 seconds = 4004.33 MB/sec
Timing buffered disk reads: 252 MB in  3.01 seconds =  83.63 MB/sec

# hdparm -tT /dev/sdb
/dev/sdb:
Timing cached reads:   8230 MB in  2.00 seconds = 4116.54 MB/sec
Timing buffered disk reads: 252 MB in  3.02 seconds =  83.55 MB/sec

# hdparm -tT /dev/sdb
/dev/sdb:
Timing cached reads:   8446 MB in  2.00 seconds = 4224.36 MB/sec
Timing buffered disk reads: 230 MB in  3.02 seconds =  76.28 MB/sec

Wikipedia says USB 3.0 has transmission speeds of up to 5 Gbit/s.
Doesn't MB/sec denote mega*bytes* per second?

- Grant

Re: [gentoo-user] {OT} USB 3.0 hard drive speed test

[covici]

Grant <emailgrant@gmail.com> wrote:

> I'm testing this USB 3.0 bus-powered hard drive:
> 
> http://www.amazon.com/gp/product/B0041OSQ9S
> 
> and I get:
> 
> # hdparm -tT /dev/sdb
> /dev/sdb:
> Timing cached reads:   8006 MB in  2.00 seconds = 4004.33 MB/sec
> Timing buffered disk reads: 252 MB in  3.01 seconds =  83.63 MB/sec
> 
> # hdparm -tT /dev/sdb
> /dev/sdb:
> Timing cached reads:   8230 MB in  2.00 seconds = 4116.54 MB/sec
> Timing buffered disk reads: 252 MB in  3.02 seconds =  83.55 MB/sec
> 
> # hdparm -tT /dev/sdb
> /dev/sdb:
> Timing cached reads:   8446 MB in  2.00 seconds = 4224.36 MB/sec
> Timing buffered disk reads: 230 MB in  3.02 seconds =  76.28 MB/sec
> 
> Wikipedia says USB 3.0 has transmission speeds of up to 5 Gbit/s.
> Doesn't MB/sec denote mega*bytes* per second?
What usb3 is supported by Linux?  Is it a pci card?

-- 
Your life is like a penny.  You're going to lose it.  The question is:
How do
you spend it?

         John Covici
         covici@ccs.covici.com

Re: [gentoo-user] {OT} USB 3.0 hard drive speed test

[Mark Knecht]

On Thu, Aug 11, 2011 at 10:12 AM, Grant <emailgrant@gmail.com> wrote:
> I'm testing this USB 3.0 bus-powered hard drive:
>
> http://www.amazon.com/gp/product/B0041OSQ9S
>
> and I get:
>
> # hdparm -tT /dev/sdb
> /dev/sdb:
> Timing cached reads:   8006 MB in  2.00 seconds = 4004.33 MB/sec
> Timing buffered disk reads: 252 MB in  3.01 seconds =  83.63 MB/sec
>
> # hdparm -tT /dev/sdb
> /dev/sdb:
> Timing cached reads:   8230 MB in  2.00 seconds = 4116.54 MB/sec
> Timing buffered disk reads: 252 MB in  3.02 seconds =  83.55 MB/sec
>
> # hdparm -tT /dev/sdb
> /dev/sdb:
> Timing cached reads:   8446 MB in  2.00 seconds = 4224.36 MB/sec
> Timing buffered disk reads: 230 MB in  3.02 seconds =  76.28 MB/sec
>
> Wikipedia says USB 3.0 has transmission speeds of up to 5 Gbit/s.
> Doesn't MB/sec denote mega*bytes* per second?
>
> - Grant

4000MB/s = 4Gb/s

Re: [gentoo-user] {OT} USB 3.0 hard drive speed test

[Mark Knecht]

On Thu, Aug 11, 2011 at 10:27 AM,  <covici@ccs.covici.com> wrote:
<SNIP>
> What usb3 is supported by Linux?  Is it a pci card?
>
> --
> Your life is like a penny.  You're going to lose it.  The question is:
> How do
> you spend it?
>
>         John Covici
>         covici@ccs.covici.com
>
>

I have it on my Asus laptop.

- Mark

Re: [gentoo-user] {OT} USB 3.0 hard drive speed test

[Volker Armin Hemmann]

Am Donnerstag 11 August 2011, 10:12:54 schrieb Grant:
> I'm testing this USB 3.0 bus-powered hard drive:
> 
> http://www.amazon.com/gp/product/B0041OSQ9S
> 
> and I get:
> 
> # hdparm -tT /dev/sdb
> /dev/sdb:
> Timing cached reads:   8006 MB in  2.00 seconds = 4004.33 MB/sec
> Timing buffered disk reads: 252 MB in  3.01 seconds =  83.63 MB/sec
> 
> # hdparm -tT /dev/sdb
> /dev/sdb:
> Timing cached reads:   8230 MB in  2.00 seconds = 4116.54 MB/sec
> Timing buffered disk reads: 252 MB in  3.02 seconds =  83.55 MB/sec
> 
> # hdparm -tT /dev/sdb
> /dev/sdb:
> Timing cached reads:   8446 MB in  2.00 seconds = 4224.36 MB/sec
> Timing buffered disk reads: 230 MB in  3.02 seconds =  76.28 MB/sec
> 
> Wikipedia says USB 3.0 has transmission speeds of up to 5 Gbit/s.
> Doesn't MB/sec denote mega*bytes* per second?
> 
> - Grant

so what?

first you are measuring the access to the in-memory cache and then a harddisk 
- and for a harddisk 80mb/sec is really, really good.


-- 
#163933

Re: [gentoo-user] {OT} USB 3.0 hard drive speed test

[Volker Armin Hemmann]

Am Donnerstag 11 August 2011, 10:30:04 schrieb Mark Knecht:
> On Thu, Aug 11, 2011 at 10:12 AM, Grant <emailgrant@gmail.com> wrote:
> > I'm testing this USB 3.0 bus-powered hard drive:
> > 
> > http://www.amazon.com/gp/product/B0041OSQ9S
> > 
> > and I get:
> > 
> > # hdparm -tT /dev/sdb
> > /dev/sdb:
> > Timing cached reads:   8006 MB in  2.00 seconds = 4004.33 MB/sec
> > Timing buffered disk reads: 252 MB in  3.01 seconds =  83.63 MB/sec
> > 
> > # hdparm -tT /dev/sdb
> > /dev/sdb:
> > Timing cached reads:   8230 MB in  2.00 seconds = 4116.54 MB/sec
> > Timing buffered disk reads: 252 MB in  3.02 seconds =  83.55 MB/sec
> > 
> > # hdparm -tT /dev/sdb
> > /dev/sdb:
> > Timing cached reads:   8446 MB in  2.00 seconds = 4224.36 MB/sec
> > Timing buffered disk reads: 230 MB in  3.02 seconds =  76.28 MB/sec
> > 
> > Wikipedia says USB 3.0 has transmission speeds of up to 5 Gbit/s.
> > Doesn't MB/sec denote mega*bytes* per second?
> > 
> > - Grant
> 
> 4000MB/s = 4Gb/s


please read man hdparm

  -T     Perform timings of cache reads for benchmark and comparison purposes.  
For meaningful results,  this  operation
              should be repeated 2-3 times on an otherwise inactive system (no 
other active processes) with at least a couple
              of megabytes of free memory.  This displays the speed of reading 
directly from the Linux buffer  cache  without
              disk access.  This measurement is essentially an indication of 
the throughput of the processor, cache, and mem-
              ory of the system under test.


as you can see, those numbers have nothing to do with the transport.

And 80mb/sec for a harddisk is really, really good.

-- 
#163933

Re: [gentoo-user] {OT} USB 3.0 hard drive speed test

[Simon]

[-- Attachment #1: Type: text/plain, Size: 2052 bytes --]

It is my opinion that benchmarks should be done with a real benchmark tool.

Try with bonnie++
This will really show you the strengths and weaknesses of your setup.

Good luck,
  Simon

On Thu, Aug 11, 2011 at 1:49 PM, Volker Armin Hemmann <
volkerarmin@googlemail.com> wrote:

> Am Donnerstag 11 August 2011, 10:30:04 schrieb Mark Knecht:
> > On Thu, Aug 11, 2011 at 10:12 AM, Grant <emailgrant@gmail.com> wrote:
> > > I'm testing this USB 3.0 bus-powered hard drive:
> > >
> > > http://www.amazon.com/gp/product/B0041OSQ9S
> > >
> > > and I get:
> > >
> > > # hdparm -tT /dev/sdb
> > > /dev/sdb:
> > > Timing cached reads:   8006 MB in  2.00 seconds = 4004.33 MB/sec
> > > Timing buffered disk reads: 252 MB in  3.01 seconds =  83.63 MB/sec
> > >
> > > # hdparm -tT /dev/sdb
> > > /dev/sdb:
> > > Timing cached reads:   8230 MB in  2.00 seconds = 4116.54 MB/sec
> > > Timing buffered disk reads: 252 MB in  3.02 seconds =  83.55 MB/sec
> > >
> > > # hdparm -tT /dev/sdb
> > > /dev/sdb:
> > > Timing cached reads:   8446 MB in  2.00 seconds = 4224.36 MB/sec
> > > Timing buffered disk reads: 230 MB in  3.02 seconds =  76.28 MB/sec
> > >
> > > Wikipedia says USB 3.0 has transmission speeds of up to 5 Gbit/s.
> > > Doesn't MB/sec denote mega*bytes* per second?
> > >
> > > - Grant
> >
> > 4000MB/s = 4Gb/s
>
>
> please read man hdparm
>
>  -T     Perform timings of cache reads for benchmark and comparison
> purposes.
> For meaningful results,  this  operation
>              should be repeated 2-3 times on an otherwise inactive system
> (no
> other active processes) with at least a couple
>              of megabytes of free memory.  This displays the speed of
> reading
> directly from the Linux buffer  cache  without
>              disk access.  This measurement is essentially an indication of
> the throughput of the processor, cache, and mem-
>              ory of the system under test.
>
>
> as you can see, those numbers have nothing to do with the transport.
>
> And 80mb/sec for a harddisk is really, really good.
>
> --
> #163933
>
>

[-- Attachment #2: Type: text/html, Size: 2831 bytes --]

Re: [gentoo-user] {OT} USB 3.0 hard drive speed test

[Grant]

>> I'm testing this USB 3.0 bus-powered hard drive:
>>
>> http://www.amazon.com/gp/product/B0041OSQ9S
>>
>> and I get:
>>
>> # hdparm -tT /dev/sdb
>> /dev/sdb:
>> Timing cached reads:   8006 MB in  2.00 seconds = 4004.33 MB/sec
>> Timing buffered disk reads: 252 MB in  3.01 seconds =  83.63 MB/sec
>>
>> # hdparm -tT /dev/sdb
>> /dev/sdb:
>> Timing cached reads:   8230 MB in  2.00 seconds = 4116.54 MB/sec
>> Timing buffered disk reads: 252 MB in  3.02 seconds =  83.55 MB/sec
>>
>> # hdparm -tT /dev/sdb
>> /dev/sdb:
>> Timing cached reads:   8446 MB in  2.00 seconds = 4224.36 MB/sec
>> Timing buffered disk reads: 230 MB in  3.02 seconds =  76.28 MB/sec
>>
>> Wikipedia says USB 3.0 has transmission speeds of up to 5 Gbit/s.
>> Doesn't MB/sec denote mega*bytes* per second?
> What usb3 is supported by Linux?  Is it a pci card?

I'm on 2.6.39-hardened-r10 and I'm using this motherboard with onboard USB 3.0:

http://www.newegg.com/Product/Product.aspx?Item=N82E16813128490

- Grant

Re: [gentoo-user] {OT} USB 3.0 hard drive speed test

[Grant]

>> I'm testing this USB 3.0 bus-powered hard drive:
>>
>> http://www.amazon.com/gp/product/B0041OSQ9S
>>
>> and I get:
>>
>> # hdparm -tT /dev/sdb
>> /dev/sdb:
>> Timing cached reads:   8006 MB in  2.00 seconds = 4004.33 MB/sec
>> Timing buffered disk reads: 252 MB in  3.01 seconds =  83.63 MB/sec
>>
>> # hdparm -tT /dev/sdb
>> /dev/sdb:
>> Timing cached reads:   8230 MB in  2.00 seconds = 4116.54 MB/sec
>> Timing buffered disk reads: 252 MB in  3.02 seconds =  83.55 MB/sec
>>
>> # hdparm -tT /dev/sdb
>> /dev/sdb:
>> Timing cached reads:   8446 MB in  2.00 seconds = 4224.36 MB/sec
>> Timing buffered disk reads: 230 MB in  3.02 seconds =  76.28 MB/sec
>>
>> Wikipedia says USB 3.0 has transmission speeds of up to 5 Gbit/s.
>> Doesn't MB/sec denote mega*bytes* per second?
>>
>> - Grant
>
> 4000MB/s = 4Gb/s

OK, I thought "B" designated bytes and "b" designated bits.  So when
you see something like MB/s or Mb/s there's no way to know if it's
referring to megabits or megabytes?

- Grant

Re: [gentoo-user] {OT} USB 3.0 hard drive speed test

[Mark Knecht]

On Thu, Aug 11, 2011 at 11:27 AM, Grant <emailgrant@gmail.com> wrote:
>>> I'm testing this USB 3.0 bus-powered hard drive:
>>>
>>> http://www.amazon.com/gp/product/B0041OSQ9S
>>>
>>> and I get:
>>>
>>> # hdparm -tT /dev/sdb
>>> /dev/sdb:
>>> Timing cached reads:   8006 MB in  2.00 seconds = 4004.33 MB/sec
>>> Timing buffered disk reads: 252 MB in  3.01 seconds =  83.63 MB/sec
>>>
>>> # hdparm -tT /dev/sdb
>>> /dev/sdb:
>>> Timing cached reads:   8230 MB in  2.00 seconds = 4116.54 MB/sec
>>> Timing buffered disk reads: 252 MB in  3.02 seconds =  83.55 MB/sec
>>>
>>> # hdparm -tT /dev/sdb
>>> /dev/sdb:
>>> Timing cached reads:   8446 MB in  2.00 seconds = 4224.36 MB/sec
>>> Timing buffered disk reads: 230 MB in  3.02 seconds =  76.28 MB/sec
>>>
>>> Wikipedia says USB 3.0 has transmission speeds of up to 5 Gbit/s.
>>> Doesn't MB/sec denote mega*bytes* per second?
>>>
>>> - Grant
>>
>> 4000MB/s = 4Gb/s
>
> OK, I thought "B" designated bytes and "b" designated bits.  So when
> you see something like MB/s or Mb/s there's no way to know if it's
> referring to megabits or megabytes?
>
> - Grant

Nahh, I made a mistake there. 4000MB/S = 4GB/S  - You are correct, at
least as far as I'm concerned.

Of course Volker, the man with the answer to everything, just told you
that 80 mega-bit / second was a 'really, really good hard drive' so I
guess we all make mistakes. ;-)

I was really trying to point out that you ARE getting 80% of the USB
3.0 spec in the first number which is the speed of communications
across the USB cable talking to the drive interface. Sustained storage
rates of the drive don't have anything to do with that though. That
depends on the drive in the case.

Cheers,
Mark

Re: [gentoo-user] {OT} USB 3.0 hard drive speed test

[Grant]

>> > I'm testing this USB 3.0 bus-powered hard drive:
>> >
>> > http://www.amazon.com/gp/product/B0041OSQ9S
>> >
>> > and I get:
>> >
>> > # hdparm -tT /dev/sdb
>> > /dev/sdb:
>> > Timing cached reads:   8006 MB in  2.00 seconds = 4004.33 MB/sec
>> > Timing buffered disk reads: 252 MB in  3.01 seconds =  83.63 MB/sec
>> >
>> > # hdparm -tT /dev/sdb
>> > /dev/sdb:
>> > Timing cached reads:   8230 MB in  2.00 seconds = 4116.54 MB/sec
>> > Timing buffered disk reads: 252 MB in  3.02 seconds =  83.55 MB/sec
>> >
>> > # hdparm -tT /dev/sdb
>> > /dev/sdb:
>> > Timing cached reads:   8446 MB in  2.00 seconds = 4224.36 MB/sec
>> > Timing buffered disk reads: 230 MB in  3.02 seconds =  76.28 MB/sec
>> >
>> > Wikipedia says USB 3.0 has transmission speeds of up to 5 Gbit/s.
>> > Doesn't MB/sec denote mega*bytes* per second?
>> >
>> > - Grant
>>
>> 4000MB/s = 4Gb/s
>
>
> please read man hdparm
>
>  -T     Perform timings of cache reads for benchmark and comparison purposes.
> For meaningful results,  this  operation
>              should be repeated 2-3 times on an otherwise inactive system (no
> other active processes) with at least a couple
>              of megabytes of free memory.  This displays the speed of reading
> directly from the Linux buffer  cache  without
>              disk access.  This measurement is essentially an indication of
> the throughput of the processor, cache, and mem-
>              ory of the system under test.
>
>
> as you can see, those numbers have nothing to do with the transport.
>
> And 80mb/sec for a harddisk is really, really good.

Here's what I get from the same hard drive plugged into a USB 2.0 port:

# hdparm -t /dev/sdb
/dev/sdb:
Timing buffered disk reads: 102 MB in  3.01 seconds =  33.90 MB/sec

# hdparm -t /dev/sdb
/dev/sdb:
Timing buffered disk reads:  92 MB in  3.00 seconds =  30.66 MB/sec

# hdparm -t /dev/sdb
/dev/sdb:
Timing buffered disk reads: 102 MB in  3.03 seconds =  33.63 MB/sec

So USB 2.0 throughput is obviously creating a bottleneck.  USB 2.0
throughput is said to be 60 MB/s so I'm surprised I'm not doing much
better than 30 MB/s there.

USB 3.0 throughput is said to be 625 MB/s so I must be running up
against the speed of the disk itself in USB 3.0 mode, correct?  Here's
what I get from my internal SATA hard drive, but it is surely a much
faster disk:

# hdparm -t /dev/sda
/dev/sda:
Timing buffered disk reads: 412 MB in  3.01 seconds = 136.99 MB/sec

# hdparm -t /dev/sda
/dev/sda:
Timing buffered disk reads: 412 MB in  3.01 seconds = 136.75 MB/sec

# hdparm -t /dev/sda
/dev/sda:
Timing buffered disk reads: 414 MB in  3.01 seconds = 137.55 MB/sec

- Grant

Re: [gentoo-user] {OT} USB 3.0 hard drive speed test

[Grant]

>>>> I'm testing this USB 3.0 bus-powered hard drive:
>>>>
>>>> http://www.amazon.com/gp/product/B0041OSQ9S
>>>>
>>>> and I get:
>>>>
>>>> # hdparm -tT /dev/sdb
>>>> /dev/sdb:
>>>> Timing cached reads:   8006 MB in  2.00 seconds = 4004.33 MB/sec
>>>> Timing buffered disk reads: 252 MB in  3.01 seconds =  83.63 MB/sec
>>>>
>>>> # hdparm -tT /dev/sdb
>>>> /dev/sdb:
>>>> Timing cached reads:   8230 MB in  2.00 seconds = 4116.54 MB/sec
>>>> Timing buffered disk reads: 252 MB in  3.02 seconds =  83.55 MB/sec
>>>>
>>>> # hdparm -tT /dev/sdb
>>>> /dev/sdb:
>>>> Timing cached reads:   8446 MB in  2.00 seconds = 4224.36 MB/sec
>>>> Timing buffered disk reads: 230 MB in  3.02 seconds =  76.28 MB/sec
>>>>
>>>> Wikipedia says USB 3.0 has transmission speeds of up to 5 Gbit/s.
>>>> Doesn't MB/sec denote mega*bytes* per second?
>>>>
>>>> - Grant
>>>
>>> 4000MB/s = 4Gb/s
>>
>> OK, I thought "B" designated bytes and "b" designated bits.  So when
>> you see something like MB/s or Mb/s there's no way to know if it's
>> referring to megabits or megabytes?
>>
>> - Grant
>
> Nahh, I made a mistake there. 4000MB/S = 4GB/S  - You are correct, at
> least as far as I'm concerned.
>
> Of course Volker, the man with the answer to everything, just told you
> that 80 mega-bit / second was a 'really, really good hard drive' so I
> guess we all make mistakes. ;-)
>
> I was really trying to point out that you ARE getting 80% of the USB
> 3.0 spec in the first number which is the speed of communications
> across the USB cable talking to the drive interface. Sustained storage

As Volker pointed out, I don't think hdparm -T interacts with the USB
system at all.  I get the same rates whether the USB hard drive is
plugged into a 3.0 port or 2.0 port.

- Grant


> rates of the drive don't have anything to do with that though. That
> depends on the drive in the case.
>
> Cheers,
> Mark

Re: [gentoo-user] {OT} USB 3.0 hard drive speed test

[Alan McKinnon]

On Thu 11 August 2011 11:50:17 Grant did opine thusly:
> USB 3.0 throughput is said to be 625 MB/s so I must be running up
> against the speed of the disk itself in USB 3.0 mode,
> correct?  Here's what I get from my internal SATA hard drive, but
> it is surely a much faster disk:
> 
> # hdparm -t /dev/sda
> /dev/sda:
> Timing buffered disk reads: 412 MB in  3.01 seconds = 136.99 MB/sec

Correct. The drive platters are still only going to get data off as 
fast as they ever did. Even with USB 3.0 that speed will not change. 


-- 
alan dot mckinnon at gmail dot com

Re: [gentoo-user] {OT} USB 3.0 hard drive speed test

[Alan McKinnon]

On Thu 11 August 2011 11:27:13 Grant did opine thusly:
> >> I'm testing this USB 3.0 bus-powered hard drive:
> >> 
> >> http://www.amazon.com/gp/product/B0041OSQ9S
> >> 
> >> and I get:
> >> 
> >> # hdparm -tT /dev/sdb
> >> /dev/sdb:
> >> Timing cached reads:   8006 MB in  2.00 seconds = 4004.33
> >> MB/sec
> >> Timing buffered disk reads: 252 MB in  3.01 seconds =  83.63
> >> MB/sec
> >> 
> >> # hdparm -tT /dev/sdb
> >> /dev/sdb:
> >> Timing cached reads:   8230 MB in  2.00 seconds = 4116.54
> >> MB/sec
> >> Timing buffered disk reads: 252 MB in  3.02 seconds =  83.55
> >> MB/sec
> >> 
> >> # hdparm -tT /dev/sdb
> >> /dev/sdb:
> >> Timing cached reads:   8446 MB in  2.00 seconds = 4224.36
> >> MB/sec
> >> Timing buffered disk reads: 230 MB in  3.02 seconds =  76.28
> >> MB/sec
> >> 
> >> Wikipedia says USB 3.0 has transmission speeds of up to 5
> >> Gbit/s. Doesn't MB/sec denote mega*bytes* per second?
> >> 
> >> - Grant
> > 
> > 4000MB/s = 4Gb/s
> 
> OK, I thought "B" designated bytes and "b" designated bits.  So when
> you see something like MB/s or Mb/s there's no way to know if it's
> referring to megabits or megabytes?

B and b mean whatever the fuck the manufacturer felt like having them 
mean on any given day[1]. There's no rhyme or reason with disk 
manufacturers, so you have to resort to independant testers who will 
tell you that disk have usable capacity of X units, where units is an 
exactly defined thing that we all understand.

[1] There are standards, but not enforceable so manufacturers can 
often do anything they want. Japanese sound amplifiers are the same - 
I've seen portable Sony units advertised as 5,000W!!!!!

That's two hot water geysers, a stove and a few kettles of power. All 
out of size D torch batteries. yeah right.

-- 
alan dot mckinnon at gmail dot com

Re: [gentoo-user] {OT} USB 3.0 hard drive speed test

[Paul Hartman]

On Thu, Aug 11, 2011 at 1:50 PM, Grant <emailgrant@gmail.com> wrote:
>>> > I'm testing this USB 3.0 bus-powered hard drive:
>>> >
>>> > http://www.amazon.com/gp/product/B0041OSQ9S
>>> >
>>> > and I get:
>>> >
>>> > # hdparm -tT /dev/sdb
>>> > /dev/sdb:
>>> > Timing cached reads:   8006 MB in  2.00 seconds = 4004.33 MB/sec
>>> > Timing buffered disk reads: 252 MB in  3.01 seconds =  83.63 MB/sec
>>> >
>>> > # hdparm -tT /dev/sdb
>>> > /dev/sdb:
>>> > Timing cached reads:   8230 MB in  2.00 seconds = 4116.54 MB/sec
>>> > Timing buffered disk reads: 252 MB in  3.02 seconds =  83.55 MB/sec
>>> >
>>> > # hdparm -tT /dev/sdb
>>> > /dev/sdb:
>>> > Timing cached reads:   8446 MB in  2.00 seconds = 4224.36 MB/sec
>>> > Timing buffered disk reads: 230 MB in  3.02 seconds =  76.28 MB/sec
>>> >
>>> > Wikipedia says USB 3.0 has transmission speeds of up to 5 Gbit/s.
>>> > Doesn't MB/sec denote mega*bytes* per second?

Those speeds seem good and normal to me, much faster than USB 2.0
would have provided. And really good for a low-power drive. Faster
than my USB 3.0 32GB flash drive. :)

Here's a benchmark of 2.5" USB 3.0 external drives, in fact yours is
one of the tested disks. Your speeds above are actually faster than
the benchmarked speeds:
http://www.everythingusb.com/images/list/portable-drive-large-file-copy-benchmark.png

> Here's what I get from the same hard drive plugged into a USB 2.0 port:
>
> # hdparm -t /dev/sdb
> /dev/sdb:
> Timing buffered disk reads: 102 MB in  3.01 seconds =  33.90 MB/sec
>
> # hdparm -t /dev/sdb
> /dev/sdb:
> Timing buffered disk reads:  92 MB in  3.00 seconds =  30.66 MB/sec
>
> # hdparm -t /dev/sdb
> /dev/sdb:
> Timing buffered disk reads: 102 MB in  3.03 seconds =  33.63 MB/sec
>
> So USB 2.0 throughput is obviously creating a bottleneck.  USB 2.0
> throughput is said to be 60 MB/s so I'm surprised I'm not doing much
> better than 30 MB/s there.

USB 2.0 theoretical max is 480Mbps but realistically it is more like
320Mbps, which means your speeds above are just about as fast as
anyone is ever going to get on USB 2.0, and they are in line with the
fastest speeds I've ever gotten personally on USB 2.0. Roughly about
32 MB/sec is "maximum speed" on USB 2.0 in my experience.

> USB 3.0 throughput is said to be 625 MB/s so I must be running up
> against the speed of the disk itself in USB 3.0 mode, correct?

Processing power of the external USB<->SATA controller chip could also
come into play. But in your case I think you're getting the maximum
speed possible from the drive. Be happy. :)

>Here's
> what I get from my internal SATA hard drive, but it is surely a much
> faster disk:
>
> # hdparm -t /dev/sda
> /dev/sda:
> Timing buffered disk reads: 412 MB in  3.01 seconds = 136.99 MB/sec
>
> # hdparm -t /dev/sda
> /dev/sda:
> Timing buffered disk reads: 412 MB in  3.01 seconds = 136.75 MB/sec
>
> # hdparm -t /dev/sda
> /dev/sda:
> Timing buffered disk reads: 414 MB in  3.01 seconds = 137.55 MB/sec

Surely a faster/higher-powered disk and I would guess that USB 3 maybe
has higher latency than the SATA controller on your laptop's
motherboard. I know USB 2.0 has latency problems (and why the audio
nerds* prefer firewire equipment).

* used as a term of endearment :)

Re: [gentoo-user] {OT} USB 3.0 hard drive speed test

[Grant]

>> USB 3.0 throughput is said to be 625 MB/s so I must be running up
>> against the speed of the disk itself in USB 3.0 mode, correct?
>
> Processing power of the external USB<->SATA controller chip could also
> come into play. But in your case I think you're getting the maximum
> speed possible from the drive. Be happy. :)

I'm happy.  I just like to know where I stand relative to the spec.

> Surely a faster/higher-powered disk and I would guess that USB 3 maybe
> has higher latency than the SATA controller on your laptop's
> motherboard. I know USB 2.0 has latency problems (and why the audio
> nerds* prefer firewire equipment).
>
> * used as a term of endearment :)

For audio playback, USB is now just as good as Firewire thanks to the
inception of the asynchronous (as opposed to synchronous or adaptive)
USB DAC.  In an asynchronous implementation, the clock is in the DAC
itself and operates independently of the computer.  However, if
Firewire has lower latency I can see how it would be better than USB
for audio recoding and production.

- Grant

Re: [gentoo-user] {OT} USB 3.0 hard drive speed test

[Peter Humphrey]

On Thursday 11 August 2011 19:50:17 Grant wrote:

> So USB 2.0 throughput is obviously creating a bottleneck.

That might be obvious to you, but it isn't to me. You ran different tests in 
the two cases, differing in -T, as Volker pointed out.

-- 
Rgds
Peter		Linux Counter 5290, 1994-04-23

Re: [gentoo-user] {OT} USB 3.0 hard drive speed test

[Michael Mol]

On Thu, Aug 11, 2011 at 1:27 PM,  <covici@ccs.covici.com> wrote:
> Grant <emailgrant@gmail.com> wrote:
> What usb3 is supported by Linux?  Is it a pci card?

The UHCI interface standard provides support for USB 1.1. EHCI, USB 2.0.

XHCI supports USB 3.0, as well as 2.0 and 1.1. Any chipset
manufacturer which wants to support USB 3.0 will be using XHCI, which
the Linux kernel has an option for.

I've now built two systems that have USB 3.0 on the motherboard. A
quick search of Newegg for XHCI shows that you can PCIe expansion
cards to add USB 3.0 to your system.

-- 
:wq

Re: [gentoo-user] {OT} USB 3.0 hard drive speed test

[covici]

Michael Mol <mikemol@gmail.com> wrote:

> On Thu, Aug 11, 2011 at 1:27 PM,  <covici@ccs.covici.com> wrote:
> > Grant <emailgrant@gmail.com> wrote:
> > What usb3 is supported by Linux?  Is it a pci card?
> 
> The UHCI interface standard provides support for USB 1.1. EHCI, USB 2.0.
> 
> XHCI supports USB 3.0, as well as 2.0 and 1.1. Any chipset
> manufacturer which wants to support USB 3.0 will be using XHCI, which
> the Linux kernel has an option for.
> 
> I've now built two systems that have USB 3.0 on the motherboard. A
> quick search of Newegg for XHCI shows that you can PCIe expansion
> cards to add USB 3.0 to your system.

Thanks -- I did not know about XHCI.

-- 
Your life is like a penny.  You're going to lose it.  The question is:
How do
you spend it?

         John Covici
         covici@ccs.covici.com

Re: [gentoo-user] {OT} USB 3.0 hard drive speed test

[Grant]

>>> I'm testing this USB 3.0 bus-powered hard drive:
>>>
>>> http://www.amazon.com/gp/product/B0041OSQ9S
>>>
>>> and I get:
>>>
>>> # hdparm -tT /dev/sdb
>>> /dev/sdb:
>>> Timing cached reads:   8006 MB in  2.00 seconds = 4004.33 MB/sec
>>> Timing buffered disk reads: 252 MB in  3.01 seconds =  83.63 MB/sec
>>>
>>> # hdparm -tT /dev/sdb
>>> /dev/sdb:
>>> Timing cached reads:   8230 MB in  2.00 seconds = 4116.54 MB/sec
>>> Timing buffered disk reads: 252 MB in  3.02 seconds =  83.55 MB/sec
>>>
>>> # hdparm -tT /dev/sdb
>>> /dev/sdb:
>>> Timing cached reads:   8446 MB in  2.00 seconds = 4224.36 MB/sec
>>> Timing buffered disk reads: 230 MB in  3.02 seconds =  76.28 MB/sec
>>>
>>> Wikipedia says USB 3.0 has transmission speeds of up to 5 Gbit/s.
>>> Doesn't MB/sec denote mega*bytes* per second?
>> What usb3 is supported by Linux?  Is it a pci card?
>
> I'm on 2.6.39-hardened-r10 and I'm using this motherboard with onboard USB 3.0:
>
> http://www.newegg.com/Product/Product.aspx?Item=N82E16813128490

Just thought I'd mention that one of my USB 3.0 ports works and the
other doesn't.  The non-working port lights up the USB drive but the
drive isn't picked up by the system in dmesg at all.  I don't know if
this is a hardware or software issue.

- Grant

>
> - Grant

Re: [gentoo-user] {OT} USB 3.0 hard drive speed test

[Paul Hartman]

On Thu, Aug 18, 2011 at 2:46 PM, Grant <emailgrant@gmail.com> wrote:
> Just thought I'd mention that one of my USB 3.0 ports works and the
> other doesn't.  The non-working port lights up the USB drive but the
> drive isn't picked up by the system in dmesg at all.  I don't know if
> this is a hardware or software issue.

When I was building my PC, one of the USB ports on the front of my
case didn't work. After looking at it closely, I saw that one of the
pins on the port was bent inward on itself, so it never made contact
when I plugged devices into it.

Re: [gentoo-user] {OT} USB 3.0 hard drive speed test

[Peter Humphrey]

On Thursday 18 August 2011 23:46:30 Paul Hartman wrote:

>  I saw that one of the pins on the port was bent inward on itself, so it
>  never made contact when I plugged devices into it.

And when you tried to straighten it, it broke off, no? That's been my 
experience.

-- 
Rgds
Peter		Linux Counter 5290, 1994-04-23

Re: [gentoo-user] {OT} USB 3.0 hard drive speed test

[Paul Hartman]

On Thu, Aug 18, 2011 at 6:32 PM, Peter Humphrey
<peter@humphrey.ukfsn.org> wrote:
> On Thursday 18 August 2011 23:46:30 Paul Hartman wrote:
>
>>  I saw that one of the pins on the port was bent inward on itself, so it
>>  never made contact when I plugged devices into it.
>
> And when you tried to straighten it, it broke off, no? That's been my
> experience.

It was bent in such an awkward way that I couldn't get it to move...
I'm sure it would have snapped off if I did, though. Luckily, it was a
port on the case, not the motherboard, and the front panel was modular
-- so the manufacturer (Antec) sent me a new one free of charge.