Check-out the read/write speed on this card

Discussion in 'Digital Cameras' started by RichA, Mar 1, 2014.

  1. RichA

    RichA Guest

    RichA, Mar 1, 2014
    #1
    1. Advertisements

  2. RichA

    Dave S Guest

    Dave S, Mar 1, 2014
    #2
    1. Advertisements

  3. RichA

    Alan Browne Guest

    It would depend on the camera hardware - it won't hurt, but you might
    not get the full benefit (probably won't given the gulf between the
    camera requirement and the card performance - eg: how old is your
    camera?). In the end you could find out what is the best card speed for
    your particular camera, and get that, rather than the fastest.

    Also depends on your shooting pace, propensity to burst shooting and
    buffer size on your camera. (For me, such is pretty useless).

    The read speed is beyond most serial interfaces, other than _maybe_ USB
    3.0 to transfer (in practical context). Indeed not many computer hard
    disk drives in other than recent computers can store at a sustained 280
    MB/s. Most SSD drives can handle it. But again it's the worst choke
    point that will set the actual speed.

    Thunderbolt can handle this. Such card readers claim around 500 MB/s.
    You would certainly need SSD storage to keep up with that.
     
    Alan Browne, Mar 1, 2014
    #3
  4. RichA

    David Taylor Guest

    On 01/03/2014 15:03, Dave S wrote:
    []
    Yes, if your PC is capable of reading at higher speeds. So far, with a
    USB 3.0 based card reader, I've seen large-file speeds up to 38.1 MB/s
    (write) and 41.0 MB/s (read) on a card which on an earlier PC logged
    18.7 MB/s (write) and 20.1 MB/s (read). Nothing like 280 MB/s, though.

    I would take the extra speed if it didn't cost me any more money.
     
    David Taylor, Mar 1, 2014
    #4
  5. RichA

    David Taylor Guest

    On 01/03/2014 15:38, Alan Browne wrote:
    []
    Have you measured this 500 MB/s yourself? I find that claims are often
    not substantiated in real-world tests, let alone real-world use.
     
    David Taylor, Mar 1, 2014
    #5
  6. RichA

    Alan Browne Guest

    No I haven't. See bottom for others results. [1][2] (not this reader).

    Thunderbolt is a hardware based interface with a bit rate of 10 Ghz. I
    would not be surprised that it could sustain 500 MB/s as long as the
    chain has no particular choke points. (The card, the reader, the
    computer, target disk ...).

    With the card under discussion, such a chain - including SSD as
    destination, should be pretty close. But add DNG conversion (as I do)
    and/or write-to disk encryption and that would choke it somewhat.

    With USB 3.0 I'm writing over 180 MB/s to a hard disk (SATA III) on an
    external dock. (rather: to a 2 TB drive: 165 MB/s; to a 3 TB drive 185
    MB/s). I haven't measured the readback yet as I wrote that program for
    a particular purpose).

    If the same drives are encrypted (Filevault 2), then the write rate
    drops about 35 - 50%. So that's an other consideration.

    It's important to note that USB (any flavour) always falls far short of
    its design bit rate in actual use (for several reasons); whereas
    hardware buffered and controlled approaches (Ethernet, Firewire,
    Thunderbolt) actually deliver what they claim (or close to it) as long
    as there are no other choke points (disk read/write speeds, other
    interfaces in the chain, etc).

    With gigabit ethernet I get sustained transfers of 750 - 800 Mb/s.
    75-80% - though the disks on each end have to be fast.

    Firewire 800 I get 80 - 90 MB/s - call it 85%.

    USB 2.0 I get about 35 MB/s - 58%

    USB 3.0 I get up to 185 MB/s - 37% ( out of 4 GB/s maximum data rate -
    it is absolutely not 5 Gb/s). I still have to measure pure readback
    which should be quicker. Some day soon.

    Thunderbolt quotes follow:
    [1] Gordon Ung at Maximum PC saw peak read transfer speeds of 931MB/s
    when reading from a RAID 0 of four 240GB SandForce SF-2281 SSDs in a
    Pegasus R4 chassis.

    [2] AnandTech actually got an SSD RAID in a Pegasus chassis right up to
    1002MB/s at its very peak, which seems to be right up at the practical
    limit of a single Thunderbolt channel, but that was using a RAID 0 of
    four 128GB 6Gbps SATA SSDs, running sustained 2MB reads at a queue depth
    of 10.

    [1],[2]:
    http://www.tested.com/tech/457440-theoretical-vs-actual-bandwidth-pci-express-and-thunderbolt/
     
    Alan Browne, Mar 1, 2014
    #6
  7. RichA

    Guest Guest

    it won't help in the camera, but it will help when you connect it to
    the computer, assuming the computer is capable of handling its speed.
     
    Guest, Mar 1, 2014
    #7
  8. RichA

    David Taylor Guest

    Thanks for your comments and references. I accept what you say about
    the different interfaces, but we are talking about a camera memory card
    here, and not SSDs.

    Therefore, I will remain sceptical until I see a result of testing that
    particular card in real hardware.
     
    David Taylor, Mar 1, 2014
    #8
  9. RichA

    RJH Guest

    That's been the most noticeable difference to me on buying faster cards.
     
    RJH, Mar 1, 2014
    #9
  10. RichA

    RichA Guest

     
    RichA, Mar 2, 2014
    #10
  11. RichA

    Whisky-dave Guest

     
    Whisky-dave, Mar 3, 2014
    #11
  12. RichA

    Dave S Guest

    Thanks for the replies below.
    I could summarize them as: no benefit in camera; possible benefit in
    transfers to computer, depending on interface.

    So would you characterize the advice given on Sandisk's web site as
    false or vastly inflationary?
    I use a Canon 60D (among others).
    When I use the Sandisk web site to find recommended cards for that
    camera their recommendation is to buy one of the Extreme class cards.
    In this category, the slowest are "up to 45 MB/s".

    Two classes lower, the Sandisk Ultra cards are all rated at "up to 30 MB/s".

    Dave S.
     
    Dave S, Mar 3, 2014
    #12
  13. RichA

    Guest Guest

    it's insignificant.

    it might have been a problem 15 years ago when processors were
    significantly slower then they are now, but it's not an issue anymore.

    plus, usb 3 has less cpu overhead than usb 2 does.
    no it isn't. firewire is faster because it's basically serial scsi.
    no it won't. usb has priority (and is an insignificant load anyway).
    that has more to do with the fat32 implementation and how full the card
    is. fat slows down as the volume gets full because it's not well
    designed.
     
    Guest, Mar 3, 2014
    #13
  14. RichA

    Savageduck Guest

    I have been using SanDisk Extreme UDMA CL cards rated at 60MB/s. My
    D300S has UDMA compliant specification, so it makes no sense to use
    anything with a slower write speed, especially if you are shooting at a
    high frame rate.

    I use a SanDisk FW800 CL card reader (no longer available) which
    provides more than adequate download speed for any of my Macs. I just
    don't see where using higher priced ultra high speed card are going to
    give me any tangible benefit beyond the choke points of the camera
    write speed and buffer size, and the card reader read/Tx speed.

    So why pay the premium price when it is tough to see the benefit beyond
    winning a pissing contest?
     
    Savageduck, Mar 3, 2014
    #14
  15. RichA

    Eric Stevens Guest

    There is another factor to consider in the matter of write speed.
    http://www.clarkvision.com/photoinfo/compact_flash_card_speed/index.html
    shows that this is very dependent on (a) how full the card is (b) the
    type of file being saved and (c) the camera.

    It appears as though while you might get no advantage from what might
    be considered an excessively fast card when it is empty, it will
    definitely be faster than a still adequate but slower card when the
    card starts to fill up.
     
    Eric Stevens, Mar 3, 2014
    #15
  16. RichA

    Guest Guest

    that's because it's a fat filesystem.

    the file type does not matter. it's just bits.

    a particular camera might have different speeds for different file
    types but that's not the card's fault.
     
    Guest, Mar 3, 2014
    #16
  17. RichA

    Eric Stevens Guest

    You should look again at the URL. It's the type of file being saved
    that matters.
    Look again at the URL.
     
    Eric Stevens, Mar 4, 2014
    #17
  18. RichA

    Guest Guest

    i did and it isn't.

    bits are bits. the file type does not matter.

    what can affect fat is the number of files in a single folder,
    something he mentions.

    maybe *you* should look again at the url.

    also note his blanket disclaimer:
    The following assumes these tests are a general property of compact
    flash cards. Clearly more testing is needed to confirm how general
    these conclusions are.

    in other words, the test is inconclusive.
     
    Guest, Mar 4, 2014
    #18
  19. RichA

    Eric Stevens Guest

    Have a look at the first figure.

    Raw files only on card, write raw, original format - speed is almost
    constant until the card is over 90% filled.

    Raw files only on card, write raw, camera formatted - speed falls off
    uniformly as card fills up, except that there is a peculiar bump
    between 60% and 75% filled. Speed falls from 7.5 to about 3.2
    "mbytes/second" when the card has filled up.

    JPEG + RAW files on card, write raw+jpeg, Camera formatted - speed
    falls off almost uniformly as card fills up until it has fallen from
    7.5 to 2.3 "mbytes/second" when the card has filled up.

    The other diagrams also show a fall off in speed as the card fills up.
    The manner of the handling does.
    And you can draw not even tentative conclusions from the work he has
    done so far.
     
    Eric Stevens, Mar 4, 2014
    #19
  20. RichA

    Whisky-dave Guest

    Depends who you talk to, I've tested it and found it significant to me.
    Significant enough for video cameras to prefer firewire over USB.

    Have you every run tests copying 100 1Mb files over USB and firewire ?.

    Have you every run tests copying 1,000 100K files over USB and firewire

    Have you every run tests copying 10,000 1K files over USB and firewire


    I didn't say it was an issue but a technical fact.

    Renmemebr FAT 32 is far more than a 15 year old format.

    USB3.0 is faster thant is the point of it it can also provide more power to devices.

    That isn't the reason, USB requires the processor to host beteeen the USB devices and that uses far more processing cycles where as firewire does NOT need to use teh hosts PC to communicate.

    priority in what way.
    So in teh end are the card speeds specified in FAT format and how to they copy when filling up. Are the quoted speeds for an empty FAT 32 card or are they the fastest possible ?
     
    Whisky-dave, Mar 4, 2014
    #20
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.