FAQ FAQ  Forum Search   Events   Register Register  Login Login

Card speeds - Alpha mount bodies (updated for a99)

Page  123>
Author
Bob J View Drop Down
Admin Group
Admin Group
Dyxum Administrator

Joined: 23 December 2005
Country: United Kingdom
Location: London
Status: Offline
Posts: 27010
Post Options Post Options   Quote Bob J Quote  Post ReplyReply Direct Link To This Post Topic: Card speeds - Alpha mount bodies (updated for a99)
    Posted: 11 December 2010 at 16:50
Card speed graphs



Konica Minolta 5D (updated 7 Jan 2011):



Sandisk cards perform very well (without massive performance difference). The Kingston Ultimate performs less well than its speed rating would suggest, but the camera has serious issues with the Kingston Elite and Duracel (Dane-Elec)Pro Photo UDMA card.



Sony a700 (updated 8 June 2011):



As you can see cards rated at similar speed do perform in a fairly similar way (whether speed is expressed as 30MB/s or 200X), although there is an interesting difference between Sandisk’s Extreme and Ultra ranges, despite them having the same nominal MB/s rating. Note that the overall time for taking a series of 40 shots and clearing the buffer for cRAW can be between 60-75% of the time for RAW using the same camera/card combination – a good argument for cRAW on the a700 IMHO... Some issues with UDMA cards, even those from Sandisk (Ducati and Extreme 60).



Sony a850 (updated 20 Dec 2010)



UDMA card is more than twice as fast as the 20mb/s card..



Edited by Bob J - 09 June 2011 at 11:36
 



Back to Top
Bob J View Drop Down
Admin Group
Admin Group
Dyxum Administrator

Joined: 23 December 2005
Country: United Kingdom
Location: London
Status: Offline
Posts: 27010
Post Options Post Options   Quote Bob J Quote  Post ReplyReply Direct Link To This Post Posted: 14 December 2010 at 20:25
Why test and how to test

For some time I’ve been interested in mapping the relative performance of the memory cards we use in our cameras.

The key thing for me is to have an objective method of measurement that is easy and reproducible for a number of different people to gather data. I’ve been working on a method that involves taking a series of pictures of an on-line stopwatch.

In real-life situations the influence of the speed of a card is affected by the camera’s internal memory buffer – the a700 has a buffer that ensures that you can take at least 13 images at 5fps, even with the slowest cards – however, slow cards take a long time to clear their buffers, so recovery time is also a factor.

My latest method has given me data which has allowed me to produce the above graph, which shows how long each set of 5 shots in a series takes to be recorded and adds this together with the time taken to empty the buffer again.

I’m intending to add in data for other cards in the a700 as they cross my path, but I’m also interested in any ideas for refining/improving on this method.

I would like data from other a700 users with other cards and also in how cards perform in other bodies such as the a850, a900, a55, a33 and a5xx, 3xx and 2xx series cameras – so, if you would like to run through the tests below for your kit I’d be happy to see the results.

If you would like to contribute, read the directions below and forward me the data you collect in the format shown at the end, either by posting here, or by sending me a PM; I can then build the data into various spreadsheets and add extra graphs as soon as I have a reasonable amount of reliable data. Please stick to the method below where possible so that we know we have a level playing-field, otherwise the data will get too variable.

Setup for testing memory speed

•     Go to http://www.online-stopwatch.com/full-screen-stopwatch/ and select ‘Stopwatch’

•     Select a reasonably fast lens

•     Set the camera up to use aperture just below fully open (so diaphragm has to operate, but does not have to travel a great distance – for instance choose f1.8 or f2.0 if shooting with an f1.4 lens).

•     Set AS/SSS on

•     Set continuous autofocus on

•     Set to high-speed continuous shooting (ie the highest speed available on the body being tested)

•     Set to either cRAW or RAW (but not RAW+JPG or cRAW+JPG)

•     Choose an ISO value that gives you a shutter speed over the flash synch speed for the camera (so that frame rate is not affected by waiting for release of second curtain)

•     Make sure you can frame and focus the camera so that you can handhold and see the values for seconds and thousandths of seconds from the online stopwatch in viewfinder.

•     Start the stopwatch – you don’t need to start shooting as soon as the stopwatch starts as we can time subsequent shots relative to the stopwatch time shown on the 1st shot.

•     Start shooting; continue shooting at full speed until 40 shots have been taken (or card is full – whichever is sooner) – You will get a series of fast shots followed by some at a lower frame rate – keep your finger on the shutter release and keep shooting throughout - try rough counting as the shots are made and try to go a little over (although we can extrapolate from other slowed-down values if required)

•     Once you have finished shooting, wait for the card write light to go out before downloading

•     On downloading the card, note the stopwatch time shown on the first shot, the 5th, 10th, 15th, 20th, 25th, 30th, 35th and 40th.



•     Look at the EXIF data for the 40th shot (in Picture Motion Browser – right-click then select ‘Media File Information’) – Note the Date modified and Date Taken values – This gives you the time taken to clear the buffer (ie between the last picture being taken and written to the card) – this will only be accurate to within one second, but should be accurate enough.



Unfortunately people have been having problems in finding the buffer clearing speed - if you are unable to get the info using the above method notify instead the size of the buffer in the camera - this should be indicated by the number visible in the viewfinder between the exposure compensation scale and the anti-shake indicator with the camera at rest - in the case of the a700 this is 9, while for the KM5D it is 5...

Format for return of data (please leave any missing values blank rather than extrapolating):

Camera model = a700
RAW/cRAW = cRAW
Card brand/range/speed = Kingston Ultimate 266x
Stopwatch time 1st shot = 4.057
Stopwatch time 5th shot = 4.858
Stopwatch time 10th shot = 5.890
Stopwatch time 15th shot = 6.890
Stopwatch time 20th shot = 7.923
Stopwatch time 25th shot = 9.358
Stopwatch time 30th shot = 11.257
Stopwatch time 35th shot = 13.123
Stopwatch time 40th shot = 15.024
40th (or last*) shot, modified time = 1:26:26
40th (or last*) shot, shot taken time = 1:26:23

(*If you take less than 40 shots, use the modified and shot taken times from the last picture taken of the series, as long as this was taken after the camera slowed down for the buffer to clear.)

I had assumed that the 300x and 400x cards would have been the fastest yet for the a700, but it seems that the a700 cannot take advantage of UDMA - in fact the best of these UDMA cards only match a non-UDMA 200x card.

Does anyone out there have a different experience with UDMA cards in the a700?

Edited by Bob J - 09 June 2011 at 11:41
Back to Top
artuk View Drop Down
Senior Member
Senior Member

Joined: 06 July 2007
Country: United Kingdom
Location: United Kingdom
Status: Offline
Posts: 3751
Post Options Post Options   Quote artuk Quote  Post ReplyReply Direct Link To This Post Posted: 14 December 2010 at 21:24
Bob, I was pretty sure I saw something that the A700 does not support UDMA. The camera needs the appropriate protocol to take advantage of the UDMA circuitry in the card.
Art
Back to Top
Heidfirst View Drop Down
Senior Member
Senior Member

Joined: 31 August 2005
Country: Scotland
Location: Glasgow
Status: Offline
Posts: 1745
Post Options Post Options   Quote Heidfirst Quote  Post ReplyReply Direct Link To This Post Posted: 14 December 2010 at 21:32
iirc the A700 was 1 of the first (if not the first) UDMA capable DSLRs.
However, other tests that I have seen have always shown it to be limited to middling 30's mb/s (UDMA 2?) presumably by hardware somewhere along the line - of course when it was launched (& especially developed) there was no media available that was faster than that anyway.
With the later 30mb/s Sandisk Ultra being a relatively recent thing (replacing the Extreme IIIs which themselves went from 20mb/s to 30mb/s partway through their lives) perhaps the 30mb/s transfer speed is so close to the A700s hardware limit that the differences are negligible?
Back to Top
Allan Olesen View Drop Down
Senior Member
Senior Member

Joined: 23 October 2010
Country: Denmark
Status: Offline
Posts: 677
Post Options Post Options   Quote Allan Olesen Quote  Post ReplyReply Direct Link To This Post Posted: 15 December 2010 at 00:28
I wonder why two of the cards did not use any time for emptying the buffer. If they were the fastest cards in the test, I could understand it, but they were not - and the faster cards also use time for emptying the buffer.
Back to Top
Bob J View Drop Down
Admin Group
Admin Group
Dyxum Administrator

Joined: 23 December 2005
Country: United Kingdom
Location: London
Status: Offline
Posts: 27010
Post Options Post Options   Quote Bob J Quote  Post ReplyReply Direct Link To This Post Posted: 15 December 2010 at 00:42
I'm only showing buffer emptying speeds where I have the data - in that case The person was unable to get the figures I was after - you could assume the buffer emptying times will be similar to those either side.
 



Back to Top
Bob J View Drop Down
Admin Group
Admin Group
Dyxum Administrator

Joined: 23 December 2005
Country: United Kingdom
Location: London
Status: Offline
Posts: 27010
Post Options Post Options   Quote Bob J Quote  Post ReplyReply Direct Link To This Post Posted: 16 December 2010 at 00:20
Please see revised data for a700 in the first post.

Key new points: PNY UDMA 266x suffers less than Sandisk or Dane-Elec.

A PNY UDMA 266x card performs as well as a conventional 200x card. In comparison Sandisk UDMA cards (300x or 400x) both give speed equivalent to a 200x card. The Dane-Elec (Duracell) card only gives the same performance as a 100x card.

Notable is the performance of the PNY 266x UDMA 2GB card, which is faster than the Kingston 266x conventional card - Is this down to the lower capacity card's use of FAT16?

At the moment I'm thinking that the fastest combination for the a700 would be a 2GB (FAT16), 266x conventional card, recording cRAW shots...

Thoughts (or more data) anyone?
Back to Top
GrahamB View Drop Down
Senior Member
Senior Member

Joined: 13 October 2007
Country: United States
Location: Eastern NC
Status: Offline
Posts: 1465
Post Options Post Options   Quote GrahamB Quote  Post ReplyReply Direct Link To This Post Posted: 16 December 2010 at 00:42
Originally posted by Bob J Bob J wrote:

Please see revised data for a700 in the first post.

Key new points: PNY UDMA 266x suffers less than Sandisk or Dane-Elec.

A PNY UDMA 266x card performs as well as a conventional 200x card. In comparison Sandisk UDMA cards (300x or 400x) both give speed equivalent to a 200x card. The Dane-Elec (Duracell) card only gives the same performance as a 100x card.

Notable is the performance of the PNY 266x UDMA 2GB card, which is faster than the Kingston 266x conventional card - Is this down to the lower capacity card's use of FAT16?

At the moment I'm thinking that the fastest combination for the a700 would be a 2GB (FAT16), 266x conventional card, recording cRAW shots...

Thoughts (or more data) anyone?


Your data regarding the PNY v the Kingston appears to differ from Rob Galbraith. link
Although the Galbraith uses the a900, I think the UDMA performance should be very similar between a700 and a900.

Graham
Back to Top
Heidfirst View Drop Down
Senior Member
Senior Member

Joined: 31 August 2005
Country: Scotland
Location: Glasgow
Status: Offline
Posts: 1745
Post Options Post Options   Quote Heidfirst Quote  Post ReplyReply Direct Link To This Post Posted: 16 December 2010 at 00:49
I think that you need a lot more data befor you can really start drawing significant conclusions e.g. you have 2 examples of Sandisk Cf Extreme III 30mb/s RAW results but with a noticeable difference in times.
However ...
Imo Dane Elec (Toshiba iirc) have never been fast cards.
Interesting that the Sandisk 30mb/s, 45mb/s & 60mb/s all seem to have the same performance in an A700. I suspect that there's something structural in how the Sandisks operate that is preventing them from matching e.g. the Kingston.

btw, why select AF-C for the test rather than AF-S?
in fact why not use MF to eliminate any chance of AF variations affecting results?

Edited by Heidfirst - 16 December 2010 at 01:41
Back to Top
Bob J View Drop Down
Admin Group
Admin Group
Dyxum Administrator

Joined: 23 December 2005
Country: United Kingdom
Location: London
Status: Offline
Posts: 27010
Post Options Post Options   Quote Bob J Quote  Post ReplyReply Direct Link To This Post Posted: 16 December 2010 at 06:57
Originally posted by GrahamB GrahamB wrote:



Your data regarding the PNY v the Kingston appears to differ from Rob Galbraith. link
Although the Galbraith uses the a900, I think the UDMA performance should be very similar between a700 and a900.

Graham


I had thought so too, however the data I'm getting suggests otherwise - I note that the 'Product compatibility' tool on the Sandisk website here does not suggest UDMA cards for the a700, but does for the a850/900.

Originally posted by Heidfirst Heidfirst wrote:

I think that you need a lot more data befor you can really start drawing significant conclusions e.g. you have 2 examples of Sandisk Cf Extreme III 30mb/s RAW results but with a noticeable difference in times.
However ...
Imo Dane Elec (Toshiba iirc) have never been fast cards.
Interesting that the Sandisk 30mb/s, 45mb/s & 60mb/s all seem to have the same performance in an A700. I suspect that there's something structural in how the Sandisks operate that is preventing them from matching e.g. the Kingston.

btw, why select AF-C for the test rather than AF-S?
in fact why not use MF to eliminate any chance of AF variations affecting results?


Quite right about the need for more info - I'm still looking for a reaqson beind the variation for there Extreme III 30mb/s card - Please regard any observations as musings rather than facts.

No particular reason for AF-C, other than that I would expect it to be set for long bursts in a real-world situation.
Back to Top
GrahamB View Drop Down
Senior Member
Senior Member

Joined: 13 October 2007
Country: United States
Location: Eastern NC
Status: Offline
Posts: 1465
Post Options Post Options   Quote GrahamB Quote  Post ReplyReply Direct Link To This Post Posted: 16 December 2010 at 16:35
Originally posted by Bob J Bob J wrote:

Originally posted by GrahamB GrahamB wrote:



Your data regarding the PNY v the Kingston appears to differ from Rob Galbraith. link
Although the Galbraith uses the a900, I think the UDMA performance should be very similar between a700 and a900.

Graham


I had thought so too, however the data I'm getting suggests otherwise - I note that the 'Product compatibility' tool on the Sandisk website here does not suggest UDMA cards for the a700, but does for the a850/900.



The a700 features UDMA as this test from DP now confirms: link

Graham
Back to Top
Bob J View Drop Down
Admin Group
Admin Group
Dyxum Administrator

Joined: 23 December 2005
Country: United Kingdom
Location: London
Status: Offline
Posts: 27010
Post Options Post Options   Quote Bob J Quote  Post ReplyReply Direct Link To This Post Posted: 17 December 2010 at 10:53
Originally posted by GrahamB GrahamB wrote:

The a700 features UDMA as this test from DP now confirms: link

Graham


UDMA cards do work, but not as fast as you might expect from their speed rating - for instance the Lexar in the test you linked to gave 18 shots before slowing, which is only the same result as a Kingston 266x card. I've yet to see anything run faster than this in an a700 (but I'm very happy to get more data on UDMA cards that shows they can break that barrier).

BTW I think the Extreme III card he mentions in his comparison test is one of the original 15mb/s 100x cards, as the one I tested also gave 14 shots before slowing, but the newer 30mb/s 200x cards did 17 shots.
Back to Top
Bob J View Drop Down
Admin Group
Admin Group
Dyxum Administrator

Joined: 23 December 2005
Country: United Kingdom
Location: London
Status: Offline
Posts: 27010
Post Options Post Options   Quote Bob J Quote  Post ReplyReply Direct Link To This Post Posted: 19 December 2010 at 09:27
New table added for KM 5D and update done for a700. Still interested in data from other cameras.
Back to Top
Mink View Drop Down
Senior Member
Senior Member

Joined: 08 December 2005
Country: South Africa
Location: Cape Town
Status: Offline
Posts: 1228
Post Options Post Options   Quote Mink Quote  Post ReplyReply Direct Link To This Post Posted: 19 December 2010 at 11:44
Bob - your first bar-graph, for the 5D, lists the "Sandisk Extreme 30mb/s", while the second, for the a700, shows results for the "Sandisk Extreme III 30mb/s" (ie with the roman numerals). Are they in fact the same cards? I assume so since the new Extreme (without roman numerals) is rated at 60mb/s.
In my bag: a850 / M17-35G / CZ24-70 / S70-300G / M100m / KM5600HSD. Backup: a55 / KM11-18 / CZ16-80 / S75-300
Back to Top
Dyxum main page >  Forum Home > Equipment forums > Other photographic topics Page  123>

Forum Jump Forum Permissions View Drop Down



This page was generated in 0.078 seconds.

Monitor calibration strip

Dyxum.com - Home of the alpha system photographer

In memory of Cameron Hill - brettania

Feel free to contact us if needed.