EM1mk2 - FPS and card speed.... tests

Discussion in 'Olympus Cameras' started by AussiePhil, May 16, 2017.

  1. AussiePhil

    AussiePhil Mu-43 All-Pro

    Jun 1, 2014
    Canberra, ACT, Aust
    Phil
    Due to another thread I decided to test how different cards could affect the number of images buffered when shooting 60fps.
    Camera settings
    Lens 12-100 (and 50-200 later) both wide open
    Single AF
    Shutter speed at least 1/200th
    E-shutter 60fps (silent)
    superfine Jpg resulting in 11Mb file
    the target this time was a stop watch running on windows 10 with 1/100th accuracy
    Each card captured one burst, this was brought into adobe bridge so i could lay out 10 across to make counting easier
    Firstly the results
    16Gb Adata SDHC card
    16GBSDHC.PNG
    The camera literally stopped taking images at image 48 as the buffer was full and the not one image had been written to the card, it cleared the buffer in roughly a minute.

    Next Sandisk 32Gb Extreme Pro
    Sandisk 32Gb extreme pro.PNG
    at 48 images it slowed down but continued to capture images as the buffer was cleared

    Next Lexar 1000x
    Lexar1000x.PNG
    Sorry i lifted the finger as i seen it slow down, and missed the couple extra frames but as you may have guessed.... exactly 48 before slowdown again.

    Next Lexar 2000x
    Lexar2000x.PNG
    Strangely no surprise here but the buffer did clear noticeable faster

    -----------------
    Now for @Phocal@Phocal we threw the 50-200 onto the camera, i just had to test this as i had only used it in CAF previously
    SAF, and wide open aperture
    firstly 60fps at the wider end
    50-200 SAF wide open 60fps setting.PNG
    We averaged close to 11fps.....

    mmm ok lets try this on silent low 18fps and at 200mm, still wide open
    50-200 at 200 wide open 10fps.PNG
    Clearly i was seeing just under 10fps.

    --------------------
    Some conclusions
    * Card write speed does NOT influence the images captured to fill the buffer
    * Card write speed is critical in clearing the buffer, faster is better.
    * A fast card will let you keep shooting at a reduced rate.

    * the 50-200 can NOT achieve full burst rates the camera is capable of. mind you 10fps is not to be sneezed at :)

    If anyone would like variations on these tests i'm happy to oblige

    just for reference the four cards used
    P5161643.JPG
     
    • Informative Informative x 6
  2. nstelemark

    nstelemark Mu-43 Hall of Famer Subscribing Member

    May 28, 2013
    Halifax, Nova Scotia Canada
    Larry
    Sooooo, will the 50-200 do caf+tracking. Mine won't focus properly in that mode.
     
  3. AussiePhil

    AussiePhil Mu-43 All-Pro

    Jun 1, 2014
    Canberra, ACT, Aust
    Phil
    From memory mine worked fine but I'll need to retest to be 100%
     
  4. Hypilein

    Hypilein Mu-43 Top Veteran

    562
    Mar 18, 2015
    Thank you for doing this. Quite interesting.
     
  5. nstelemark

    nstelemark Mu-43 Hall of Famer Subscribing Member

    May 28, 2013
    Halifax, Nova Scotia Canada
    Larry
    Thanks for these tests BTW. I find the buffer to only be adequate on the E-M1mkII (raw+jpg). Currently I shoot jpg to card 1 and RAW to card 2. But, I was thinking of switching this around to shoot RAW to a higher speed card in slot 1. This tells me that other than buffer clearing it won't help.
     
  6. Growltiger

    Growltiger Mu-43 Top Veteran

    965
    Mar 26, 2014
    UK
    Note that the SanDisk Extreme Pro you have is the old version that isn't UHS II. It is perfect for the E-M1, not for the E-M1 II.
    The SanDisk Extreme Pro cards I use are marked 300MB/s, a lot faster than 95MB/s.

    Surely if you had a card where the write speed was very fast, it should be able to write some images to the card while the initial 48 are filling the buffer, thus allowing a few more to be taken before slowing down? Why isn't this the case?
     
  7. AussiePhil

    AussiePhil Mu-43 All-Pro

    Jun 1, 2014
    Canberra, ACT, Aust
    Phil
    With a smile :) note I know it's a uhs1 card and is part of the card set for my mk1, I picked it so I had spread of card types and speed to clearly illustrate that at 60fps the card write speed is irrelevant
    Now let's say we clear the buffer at a few frames per second for a fast uhs2 card like the Lexar 2000x as I have observed but not timed empathically, the buffer fills in .75-.80 of a second... at best it might write 2 images but I'll bet during that first 1/2 second the camera is preparing the images to write to the card and doesn't start writing immediately... mmm might need to slow down to 5fps a test
    I'll rerun a set at slower fps and see what happens
     
  8. Growltiger

    Growltiger Mu-43 Top Veteran

    965
    Mar 26, 2014
    UK
    That makes sense. I would be interested to see the results at about 10fps.
     
  9. AussiePhil

    AussiePhil Mu-43 All-Pro

    Jun 1, 2014
    Canberra, ACT, Aust
    Phil
    NOTE: Single card in slot 1 in all cases.....

    All settings as per first test including SuperFine JPG just changed the FPS
    Now it's starts to get interesting and the speed of the SD card matters when the fps is set lower and the camera can stream to the card at the same time.
    For reference here is the super slow 16Gb card results at 5fps - 58 frames captured.
    5fps-slowSD.PNG
    I then threw the 32Gb 2000x UHSii card in and started at 5fps..... gave up after a minute and 300 images :)

    Changed the frame rate only to 10FPS Gave up around 40 seconds and 400 images with no sign of a slow down.

    So next changed to RAW only at 10FPS ..... finally at 190 images, 19 seconds, the buffer was no longer being cleared fast enough and the fps fell off.... smart people may be able to work out the effective write speed.

    ---------------------
    Not unexpected that a nice fast cards allows more total images to be captured as the buffer is parallel cleared to the SD card.

    I don't believe this in any way changes my initial statements that the card does NOT impact the actual initial buffer filling FPS .

    Cheers
     
    • Informative Informative x 3
  10. nstelemark

    nstelemark Mu-43 Hall of Famer Subscribing Member

    May 28, 2013
    Halifax, Nova Scotia Canada
    Larry
    This is useful, so a UHSII card in slot one will actually help the buffer situation. Thanks!

    NB at 6FPS and a fast card in slot one the buffer depth is effectively infinite. Nice.
     
  11. ThomD

    ThomD Mu-43 Top Veteran

    575
    Jun 1, 2013
    SF Bay Area
    Interesting tests. I did a quick set of tests with my Lexar (same as yours) and a San Disk (same as yours but 128G). 47/48 images into the buffer before it slows down. I tested the time to clear the buffer by running to 2 seconds total and then watching for buffer clearing time.The Lex in slot 1 clears 2 seconds of shooting (call it around 55ish) in 14 seconds after I stop shooting. The San Disk in slot 1 clears the buffer in 25 seconds after stop shotting. Shooting only raw.


    Inspired by this,

    Olympus E-M1 II SD Card Speed Comparison Test - Fastest UHS-II and UHS-I Memory Cards by Write Speed and Continuous Shooting Results in the Olympus OM-D E-M1 Mark II - Camera Memory Speed Comparison & Performance tests for SD and CF cards

    NOTE: While that report is interesting, I cannot replicate their results. My results align with AussiePhil's so I would not get my hopes up on performance based on that page. Provided for entertain purposes only.

    So I did a second test - how many shots in 30 seconds at 60fps - 153 for the lex, 110 for the SanDisk. SD took 24 seconds to clear the buffer, Lex took 19. Both cards (lex in 1, SC in 2) gives me 109 shots, with 24 seconds to clear.

    at 5 fps (high speed silent) the Lex gave me 150 shots in 30 seconds, with 16 seconds to clear the buffer. The SD gave me 104, with 22 seconds to clear. Not surprising, once the buffer is full card write speeds take over.

    What have we learned - card speed matters for buffer clearing time, but not for buffer depth at high speeds. I could imagine a whole series of tests to figure out how many fps each card could sustain indefinitely - matching buffer write volume to card ability.
     
    Last edited: May 16, 2017
    • Like Like x 1
  12. ThomD

    ThomD Mu-43 Top Veteran

    575
    Jun 1, 2013
    SF Bay Area
    One more thing, that test says they got 311 shots in 30 seconds at 15 fps using what I think is the same card I have (amazon says it is), but I only get 151ish. Hmm.

    And the say, "When shooting RAW, the E-M1 II could reach up to 119 shots at full frame rate with the fastest UHS-II card. ", but that isn't our experience at all.
     
  13. AussiePhil

    AussiePhil Mu-43 All-Pro

    Jun 1, 2014
    Canberra, ACT, Aust
    Phil
    Agreed. it's why i didn't link that page as i've not found it to match the real world
     
  14. ThomD

    ThomD Mu-43 Top Veteran

    575
    Jun 1, 2013
    SF Bay Area
    I'm going to amend my post.
     
  15. kbouk

    kbouk Mu-43 Regular

    36
    Dec 21, 2013
    Greece
    Just curious, did you try it with MF (as they did), manual exposure and IS Off?
     
  16. ThomD

    ThomD Mu-43 Top Veteran

    575
    Jun 1, 2013
    SF Bay Area
    Good question. Yes - MF, Manual mode, no IS.
     
  17. Bytesmiths

    Bytesmiths Mu-43 Veteran

    You may want to reverse that. Recall that the highest-speed cards (don't remember the TLA, UHSII?) only work in slot 1.
     
    Last edited: May 18, 2017 at 10:28 PM