USB memory speed

  • Thread starter Thread starter Bill Ridgeway
  • Start date Start date
B

Bill Ridgeway

I routinely copy key files to a USB memory stick. My original 1Gb capacity
stick became rather full and files were being fragmented. I replaced this
with a 4Gb stick There is a lot of spare capacity so files are not being
fragmented (much). However, it takes a lot longer (about three times at a
guess) to copy the same amount of files.

Is there a technical difference between USB memory sticks which affect the
speed with which they will transfer files? Is this a question of the more
you pay the faster it will be?

Bill Ridgeway
 
Bill Ridgeway said this on 1/6/2009 12:06 PM:
I routinely copy key files to a USB memory stick. My original 1Gb capacity
stick became rather full and files were being fragmented. I replaced this
with a 4Gb stick There is a lot of spare capacity so files are not being
fragmented (much). However, it takes a lot longer (about three times at a
guess) to copy the same amount of files.

Is there a technical difference between USB memory sticks which affect the
speed with which they will transfer files? Is this a question of the more
you pay the faster it will be?

Bill Ridgeway
Click to expand...

Just observations.
Yes, my 256meg thumb is fast,
I have an 8 gig store generic brand and its slow (it was dirt cheap).
I have also excluded that drive from my AV and noticed an nice improvement.

PS. Since there is no head moving, fragmentation is not a concern. Its
direct read/write.
 
Bill Ridgeway said this on 1/6/2009 12:06 PM:
Click to expand...


Big_Al said:
Just observations.
Yes, my 256meg thumb is fast,
I have an 8 gig store generic brand and its slow (it was dirt cheap).
I have also excluded that drive from my AV and noticed an nice
improvement.

PS. Since there is no head moving, fragmentation is not a concern. Its
direct read/write.
Click to expand...


Bill & "Big_Al"...
Our experience with a fairly wide variety of flash drives indicate a wide
variance of data transfer speeds among the group. And we've found no
correlation between device capacity and data transfer rates. Sometimes the
smaller capacity flash device was faster (as in "Big_Al's" case); other
times the reverse was true (although the fastest device we used was also the
one with the largest capacity - see below).

Using MB/min (copying speed from HDD to flash device) as the yardstick,
we've found a range from 200 MB/min to a little over 700 MB/min - a fairly
wide variance, no? The fastest flash drive we've used is an OCZ Rally2 32 GB
model, the largest-capacity device we've used.

All in all it seems "Yer pays yer money and yer takes yer cherce".
Anna
 
Bill said:
I routinely copy key files to a USB memory stick. My original 1Gb capacity
stick became rather full and files were being fragmented. I replaced this
with a 4Gb stick There is a lot of spare capacity so files are not being
fragmented (much). However, it takes a lot longer (about three times at a
guess) to copy the same amount of files.

Is there a technical difference between USB memory sticks which affect the
speed with which they will transfer files? Is this a question of the more
you pay the faster it will be?

Bill Ridgeway
Click to expand...

There is a table here, with some benchmarks taken.
I've converted the numbers to MB/sec, for easier comparison.

http://www.1394ta.org/consumers/FAQ.html

Theoretical Actual Actual Efficiency
Cable limit Read Write (Average)

USB 2.0 60MB/sec 33MB/sec 26MB/sec 49%
1394a 50MB/sec 41.5 33 75%
1394b 100MB/sec 89 62 76%

That is for hard drives in enclosures, in a sustained transfer
condition. That should give you some idea as to the limitations
of the USB bus and its polled transfer protocol. The hard drive
doesn't limit the transfer rate, for USB2 rates, so gives you
some idea of the practical max you might expect from
a USB flash stick.

I have an OCZtechnology "dual-channel" USB stick, and it
gets 30MB/sec read and perhaps 18MB/sec write (it has changed
since I got it). The dual-channel is meant to imply there are
two flash chips, and one controller chip, inside the stick.
The two flash chips can work in parallel, to sustain a
higher transfer rate. The above rates are for large
files, and transferring a thousand small files is dog-slow.

The most miserable USB sticks you can find, transfer at
slightly more than 1MB/sec. Their transfer rate is high
enough, that they are not operating in USB 1.1 mode, and
the transfer is just above the USB1.1 practical limit.
So that is how bad it can get. I think I'm doing pretty
well, by getting 30MB/sec on sustained reads, but there
is still some room for improvements on write.

I've tried a few things with it, and it doesn't seem
impressively fast. Mine is an 8GB model.

Paul
 
Paul said this on 1/6/2009 1:16 PM:
There is a table here, with some benchmarks taken.
I've converted the numbers to MB/sec, for easier comparison.

http://www.1394ta.org/consumers/FAQ.html

Theoretical Actual Actual Efficiency
Cable limit Read Write (Average)

USB 2.0 60MB/sec 33MB/sec 26MB/sec 49%
1394a 50MB/sec 41.5 33 75%
1394b 100MB/sec 89 62 76%

That is for hard drives in enclosures, in a sustained transfer
condition. That should give you some idea as to the limitations
of the USB bus and its polled transfer protocol. The hard drive
doesn't limit the transfer rate, for USB2 rates, so gives you
some idea of the practical max you might expect from
a USB flash stick.

I have an OCZtechnology "dual-channel" USB stick, and it
gets 30MB/sec read and perhaps 18MB/sec write (it has changed
since I got it). The dual-channel is meant to imply there are
two flash chips, and one controller chip, inside the stick.
The two flash chips can work in parallel, to sustain a
higher transfer rate. The above rates are for large
files, and transferring a thousand small files is dog-slow.

The most miserable USB sticks you can find, transfer at
slightly more than 1MB/sec. Their transfer rate is high
enough, that they are not operating in USB 1.1 mode, and
the transfer is just above the USB1.1 practical limit.
So that is how bad it can get. I think I'm doing pretty
well, by getting 30MB/sec on sustained reads, but there
is still some room for improvements on write.

I've tried a few things with it, and it doesn't seem
impressively fast. Mine is an 8GB model.

Paul
Click to expand...

Okay so the next question is, what or how do you benchmark a thumb drive
with.

I think between 3 friends and I we have about 15 of these in various
sizes. Would be kinda cute to do some benchmarks.
 
JS said:
Anna,

Is there any correlation to Brand or Models that are
always dogs?
Click to expand...


JS said:
Anna,

Is there any correlation to Brand or Models that are
always dogs?
Click to expand...


JS:
I honestly can't say since I really don't believe I've worked with enough
different brands (or models within the same brand) to make that kind of an
evaluation. Frankly it seems more of a crapshoot than anything else. I do
remember that when we worked with a number of Sony flash drives we were less
than thrilled with their data transfer rates. But I have to hurriedly say
that our experience with the Sony flash drives was about three years ago so
I can't say the same results would hold today.

As I mentioned in my previous post, that OCZ model performed better than any
other flash drive device we've ever used. Whether that would hold true of
other OCZ models I don't know since it's the first OCZ model I've ever used.
I'm aware of at least two other users of that particular model who were
similarly impressed. On the other hand I seem to recall some user negative
reviews of that OCZ model at newegg, so who knows?
Anna
 
Big_Al said:
Bob I said this on 1/6/2009 1:49 PM:
Thanks. I got that much figured out. Of the 544 different benchmarks,
I was looking for suggestions on a reliable test, or at least a good
user interface. I'll be more specific in the future. I got one anyway.
Click to expand...

For read testing, I use the free version of HDTune (version 2.55) from
hdtune.com . For writes, I just do a big file transfer and time it.
HDTune comes in a pay version, which I think has a write test, but
I didn't bother with that. Read testing should be good enough for
an optimistic benchmark result (since the write will likely be
slower). Read testing is non-destructive, which is an advantage
if testing live disks.

I think I had a copy of an "Atto" tool at one time, but that may be
sitting on another computer here somewhere. That one tests block
transfer sizes and performance, so has a different objective than
HDTune. HDTune tests across the surface of the device, and on hard
drives, this gives a characteristic curve, with the fastest part
of the hard drive being near the origin. A USB flash, on the
other hand, is typically a flat line (and boring), so you can
stop the test before it finishes if you want.

Paul
 
Thanks for your time and the info.

Was planning to by a three flash drive(s) special
(2, 4 and 8GB) at Amazon but they have since deleted
the special from my cart.
 
Big_Al said:
Bill Ridgeway said this on 1/6/2009 12:06 PM:

Just observations.
Yes, my 256meg thumb is fast,
I have an 8 gig store generic brand and its slow (it was dirt cheap).
I have also excluded that drive from my AV and noticed an nice
improvement.

PS. Since there is no head moving, fragmentation is not a concern. Its
direct read/write.
Click to expand...

There may not be a read / write head but there is something that determines
where to write information and, if files are fragmented, it has decided
where to write that information which must take some resource and time -
albeit nanoseconds.

Bill Ridgeway
 
Big_Al said:
Bob I said this on 1/6/2009 1:49 PM:
Thanks. I got that much figured out. Of the 544 different
benchmarks, I was looking for suggestions on a reliable test, or at
least a good user interface. I'll be more specific in the future. I
got one anyway.
Click to expand...

The program Paul suggests is good and there is another free version
program called HDTach, which will test the read speed of any hard drive
and USB "thumb" drives. I have a 4 GB Cruzer Micro drive, which has a
constant read speed of 31 MB/s, which is about as fast as you'll get
with USB 2.0 interfaced drives. It can be downloaded here:
http://www.pcworld.com/downloads/file/fid,69656/description.html
 
Back
Top