"da0: 40.000MB/s transfers" What was rationale behind pegging USB 2.0 at 40MB/s?

classic Classic list List threaded Threaded
17 messages Options
Reply | Threaded
Open this post in threaded view
|

"da0: 40.000MB/s transfers" What was rationale behind pegging USB 2.0 at 40MB/s?

Jakub Lach
Hi,

I was fortunate enough to buy USB 3.0 pendrive,
which actually works with FreeBSD just ripped from
package! (normal _empty_ FAT filesystem, no
garbageware added, no need to format).

It actually bounces from 40MB/s limit when reading
from it.

Writing is about 18MB/s.

Device is supposed to be "467x" which should
be about 70MB/s.

And USB 2.0 hard limit is 60MB/s.

Wouldn't be nice to squeeze few additional
MB/s?
Reply | Threaded
Open this post in threaded view
|

Re: "da0: 40.000MB/s transfers" What was rationale behind pegging USB 2.0 at 40MB/s?

Jakub Lach
This could be deducted, but I will add for clarity, that I
bought USB 3.0 pendrive to use in 2.0 port, to take
advantage of 2.0 to the fullest (as 2.0 pendrives have
slow flashes inside).
Reply | Threaded
Open this post in threaded view
|

Re: "da0: 40.000MB/s transfers" What was rationale behind pegging USB 2.0 at 40MB/s?

Wojciech Puchar-5
In reply to this post by Jakub Lach
> which actually works with FreeBSD just ripped from
> package! (normal _empty_ FAT filesystem, no
> garbageware added, no need to format).

zero difference. newfs_msdos take a moment.

> It actually bounces from 40MB/s limit when reading
> from it.
>
> Writing is about 18MB/s.
>
> Device is supposed to be "467x" which should
> be about 70MB/s.

don't treat all advertised data seriously

>
> And USB 2.0 hard limit is 60MB/s.
>
> Wouldn't be nice to squeeze few additional
> MB/s?

how did you measured it.

dd if=/dev/da0 of=/dev/null bs=64k
?

dd if=/dev/zero of=/dev/da0 bs=64k

via msdosfs
via mtools

?

_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[hidden email]"
Reply | Threaded
Open this post in threaded view
|

Re: "da0: 40.000MB/s transfers" What was rationale behind pegging USB 2.0 at 40MB/s?

Jakub Lach
18MB/s write is figure from few

dd if=/dev/zero of=/dev/da0 bs=1 to 15M

runs, 13-14MB/s from actual files copied
in mc to flash and 36-39MB/s file copied
from flash to hdd in mc.

dd if=/dev/da0 of=/dev/zero bs=15m

gives 33MB/s read.

Speaking of advertisements, yes I know
but USB 3.0 drives with nice flashes are
capable of speeds well above 2.0 limits,
and that's the point anyway.

Speaking of formatting, I can't agree, as
I bought such awfully formatted drive, that it
had to be FAT formatted in Windows  to be
even recognized in FreeBSD as device.

And I don't usually have Windows around.
Reply | Threaded
Open this post in threaded view
|

Re: "da0: 40.000MB/s transfers" What was rationale behind pegging USB 2.0 at 40MB/s?

Adam Vande More
In reply to this post by Jakub Lach
On Sun, Jul 22, 2012 at 4:20 PM, Jakub Lach <[hidden email]> wrote:

> Hi,
>
> I was fortunate enough to buy USB 3.0 pendrive,
> which actually works with FreeBSD just ripped from
> package! (normal _empty_ FAT filesystem, no
> garbageware added, no need to format).
>
> It actually bounces from 40MB/s limit when reading
> from it.
>
> Writing is about 18MB/s.
>
> Device is supposed to be "467x" which should
> be about 70MB/s.
>
> And USB 2.0 hard limit is 60MB/s.
>
> Wouldn't be nice to squeeze few additional
> MB/s?
>

You are suffering a misunderstanding of how USB 2 works.  There is a lot of
overhead to it.  If you want USB 3 speeds, buy a USB 3 controller.
--
Adam Vande More
_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[hidden email]"
Reply | Threaded
Open this post in threaded view
|

Re: "da0: 40.000MB/s transfers" What was rationale behind pegging USB 2.0 at 40MB/s?

Jakub Lach
Speaking of misunderstanding, that's certainly
possible.

How much overhead is "normal" and alternatively,
why in FreeBSD USB 2.0 reports as "40MB/s" and
not other arbitrary number.

I hope I didn't sound like "PLEASE HELP I WANT
USB 3.0 SPEEDS ON USB 2.0"...

Reply | Threaded
Open this post in threaded view
|

Re: "da0: 40.000MB/s transfers" What was rationale behind pegging USB 2.0 at 40MB/s?

Adam Vande More
On Sun, Jul 22, 2012 at 6:05 PM, Jakub Lach <[hidden email]> wrote:

> Speaking of misunderstanding, that's certainly
> possible.
>
> How much overhead is "normal" and alternatively,
> why in FreeBSD USB 2.0 reports as "40MB/s" and
> not other arbitrary number.
>

The overhead includes many different things including hardware latency.
 However the big one is USB communication itself.  That is static, you
can't change it and it doesn't vary(assuming same communication type).
 Your reported speeds are typical, and in all likelihood would be very
similar under another OS.



--
Adam Vande More
_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[hidden email]"
Reply | Threaded
Open this post in threaded view
|

Re: "da0: 40.000MB/s transfers" What was rationale behind pegging USB 2.0 at 40MB/s?

Jakub Lach
Apparently my speeds are pretty decent,
as this this advertised speed relates to
read speed, and write one is pretty weak.

People are reporting

62-70MB/s read and
17-31MB/s write.

Are you saying that disk clearly bumping
from 40MB/s read barrier (as I saw in midnight
commander is my imagination or it's cause is totally
unrelated to OS?

I thought it's worth investigating, as FreeBSD
coincidentally reports USB 2.0 ports as such.
Reply | Threaded
Open this post in threaded view
|

Re: "da0: 40.000MB/s transfers" What was rationale behind pegging USB 2.0 at 40MB/s?

Wojciech Puchar-5
In reply to this post by Jakub Lach
> dd if=/dev/zero of=/dev/da0 bs=1 to 15M
>
> runs, 13-14MB/s from actual files copied
> in mc to flash and 36-39MB/s file copied
> from flash to hdd in mc.
>
> dd if=/dev/da0 of=/dev/zero bs=15m
>
> gives 33MB/s read.
and this is incredibly high as of any pendrive.

Consider few things:

- pendrive flash mappers are incredibly dumb, they work in usable speed
when you do linear write only.

- today lies are standard. never believe in stated performance unless you
measure it.


> even recognized in FreeBSD as device.
what i do with any pendrive:

dd if=/dev/zero of=/dev/da0 bs=64k count=1
newfs_msdosfs /dev/da0 (if i want FAT filesystem)

partition table is not needed in any OS, including Windows starting from
XP.


msdosfs in FreeBSD isn't high performance too.

try writing small amount of huge files.
_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[hidden email]"
Reply | Threaded
Open this post in threaded view
|

Re: "da0: 40.000MB/s transfers" What was rationale behind pegging USB 2.0 at 40MB/s?

Wojciech Puchar-5
In reply to this post by Jakub Lach
> And USB 2.0 hard limit is 60MB/s.
i have never seen USB 2.0 exceeding 35MB/s write and 40MB/s read.

even when connecting SATA disk over USB-SATA bridge.

60MB/s is wire speed. USB have enormous protocol overhead.
_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[hidden email]"
Reply | Threaded
Open this post in threaded view
|

Re: "da0: 40.000MB/s transfers" What was rationale behind pegging USB 2.0 at 40MB/s?

Jakub Lach
> i have never seen USB 2.0 exceeding 35MB/s write and 40MB/s read.

That means I essentially got what I wanted- as
high read output as possible on USB 2.0. Thanks.

Indeed 35MB/s-40MB/s is common reported maximum
throughput.

> dd if=/dev/zero of=/dev/da0 bs=64k count=1
> newfs_msdosfs /dev/da0

Apart from bs= that's exactly what I did (Well, there
was one /dev/random/ run prior.)

What I previously meant is that I had such pendrive, that
without former formatting in Windows, didn't even show
up as device in FreeBSD- was completely useless.

That does not mean I didn't newfs_msdosfsed it after
that in FreeBSD (worked perfectly fine since) :)
Reply | Threaded
Open this post in threaded view
|

Re: "da0: 40.000MB/s transfers" What was rationale behind pegging USB 2.0 at 40MB/s?

Jerry
On Mon, 23 Jul 2012 03:02:56 -0700 (PDT)
Jakub Lach articulated:

> What I previously meant is that I had such pendrive, that
> without former formatting in Windows, didn't even show
> up as device in FreeBSD- was completely useless.
>
> That does not mean I didn't newfs_msdosfsed it after
> that in FreeBSD (worked perfectly fine since) :)

I experienced that phenomena of a drive not being recognized once also.
However, after formatting it in Windows why duplicate it again in
FreeBSD? It serves no purpose that I am aware of. By the way, it is too
bad that FreeBSD is not able to take advantage of the "exFat" format
like other distributions do.

--
Jerry ♔

Disclaimer: off-list followups get on-list replies or get ignored.
Please do not ignore the Reply-To header.
__________________________________________________________________
Paul Revere was a tattle-tale.
_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[hidden email]"
Reply | Threaded
Open this post in threaded view
|

Re: "da0: 40.000MB/s transfers" What was rationale behind pegging USB 2.0 at 40MB/s?

Jakub Lach
> However, after formatting it in Windows why duplicate it again in
> FreeBSD?

Just to check if it works as should, also trim sectors and whatever.

Format without partition table?

But in this case, no 1 reason was probably most important.
Reply | Threaded
Open this post in threaded view
|

Re: "da0: 40.000MB/s transfers" What was rationale behind pegging USB 2.0 at 40MB/s?

Wojciech Puchar-5
In reply to this post by Jerry
> However, after formatting it in Windows why duplicate it again in
> FreeBSD? It serves no purpose that I am aware of. By the way, it is too
> bad that FreeBSD is not able to take advantage of the "exFat" format


fusefs-exfat in ports

still i don't really care, i would reformat in as FAT32 anyway.
_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[hidden email]"
Reply | Threaded
Open this post in threaded view
|

Re: "da0: 40.000MB/s transfers" What was rationale behind pegging USB 2.0 at 40MB/s?

Wojciech Puchar-5
In reply to this post by Jakub Lach
> Format without partition table?

yes. Windows recognizes it properly except Win98/95 (which doesn't work
with large USB drives anyway).

_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[hidden email]"
Reply | Threaded
Open this post in threaded view
|

Re: "da0: 40.000MB/s transfers" What was rationale behind pegging USB 2.0 at 40MB/s?

Wojciech Puchar-5
In reply to this post by Jakub Lach
> What I previously meant is that I had such pendrive, that
> without former formatting in Windows, didn't even show
> up as device in FreeBSD- was completely useless.

the result of XXI century "way" of programming - flash translator firmware
in that case.

They don't even read specs about USB storage, just it is fine if it works
(seems to work) in windoze.
_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[hidden email]"
Reply | Threaded
Open this post in threaded view
|

Re: "da0: 40.000MB/s transfers" What was rationale behind pegging USB 2.0 at 40MB/s?

Erich Dollansky-2
In reply to this post by Jerry
Hi,

On Monday 23 July 2012 17:48:50 Jerry wrote:

> On Mon, 23 Jul 2012 03:02:56 -0700 (PDT)
> Jakub Lach articulated:
>
> > What I previously meant is that I had such pendrive, that
> > without former formatting in Windows, didn't even show
> > up as device in FreeBSD- was completely useless.
> >
> > That does not mean I didn't newfs_msdosfsed it after
> > that in FreeBSD (worked perfectly fine since) :)
>
> I experienced that phenomena of a drive not being recognized once also.
> However, after formatting it in Windows why duplicate it again in
> FreeBSD? It serves no purpose that I am aware of. By the way, it is too
> bad that FreeBSD is not able to take advantage of the "exFat" format
> like other distributions do.

isn't there support for it via fuse?

Erich
_______________________________________________
[hidden email] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "[hidden email]"