Flash Card Write Speeds

I just ran across this article (referenced in Slashdot) that lists actual
test results for for various flash cards:
http://www.anandtech.com/printarticle.aspx?i=2654
In some respects the tests compare apples and oranges in that the some cards
are tested in their high speed versions and other cards tested are tested
only in their low or mid speed versions. The piece does however provide a
good reference to real world card speeds that we we are likely to see.
Chuck

C Wright wrote:
I just ran across this article (referenced in Slashdot) that lists
actual test results for for various flash cards:
http://www.anandtech.com/printarticle.aspx?i=2654
In some respects the tests compare apples and oranges in that the
some cards are tested in their high speed versions and other cards
tested are tested only in their low or mid speed versions. The piece
does however provide a good reference to real world card speeds that
we we are likely to see. Chuck

The wide difference in rank in the different tests make me query the set
of results. Also, there was no in-camera testing - the actual situation
where the cards will be used most of the time.

My initial feeling is of a flawed test.

David

Flawed test. Very flawed test.
Don't give a crap about how fast a card is plugged into a desk top
computer. Set it to copy, go get a drink, take a wizz and come back.

The real test is firing off actual PICTURES IN THE CAMERA.
This person has next to no ability to understand computers and should
not have been given permission to do such a deceptive test.

On Sat, 24 Dec 2005 14:01:12 -0600, No Sale wrote:

Flawed test. Very flawed test.
Don't give a crap about how fast a card is plugged into a desk top
computer. Set it to copy, go get a drink, take a wizz and come back.

The real test is firing off actual PICTURES IN THE CAMERA.
This person has next to no ability to understand computers and should
not have been given permission to do such a deceptive test.


Flawed in that they didn't runn all cards through all cameras. That
would have been near impossible.

Valid in that taking the camera buffer and electronics out of the
equation gives a fair comparison of each cards speed in read/write and
useful number of cycles before wearing out.

Correct me if I am wrong, but there is no card that cares whether it
is in a camera or ina card reader. I mean there is no proprietary
function of a card to tell it thatit is in a camera or in a computer.

I found the 1,5 and 10 meg file size read/write times very interesting
in that they all showed differences in effeciency when give different
file sizes to crunch on.

Robert Strom - not a memory card scientist

Robert Strom wrote:
[]
Correct me if I am wrong, but there is no card that cares whether it
is in a camera or ina card reader. I mean there is no proprietary
function of a card to tell it thatit is in a camera or in a computer.
[]
Robert Strom - not a memory card scientist

Doesn't the write-acceleration depend on the firmware's capability to
support it? That would be different between reader and camera. I find
the test also flawed in that no attempt was made (IIRC) to explain the
widely different rankings on the different tests.

David

Robert Strom wrote:

Correct me if I am wrong, but there is no card that cares whether it
is in a camera or ina card reader. I mean there is no proprietary
function of a card to tell it thatit is in a camera or in a computer.

However, there are issues with particular cards in particular cameras.
My colleague is selling a 2 GB card that is slow as hell in his K-M A2.
The card was on the official K-M list of compatible cards and was
later removed from it. It works normally in other cameras and
in a card reader.

I bought a CF with stated write speed of 16 MB/s for my K-M 7D.
In reality it writes one 9 MB raw in 8 seconds. Measuring
method: burst shooting until the internal buffer (9 images)
is full and waiting until the light goes off gives - this
gives 75 seconds or so. Folks at dpreview were able to write
the same RAW in 4 seconds somehow, with theoretically slower
cards. Same camera, same firmware. Of course it is much faster
than this in a card reader.

I still have some 128 MB cards that work everywhere but in the
Minolta 7Hi I owned before.

So from my experience the only thing that counts is the actual
write speed in the particular camera. The CF card is electrically
an IDE interface and all that stuff with PIO/UDMA/... applies
to it. There are many more variables in the whole chain than
the stated speed...

Regards
--
Stano

Robert Strom - not a memory card scientist- writes ...

Flawed in that they didn't runn all cards through all cameras.
That would have been near impossible.

Good point, but here's a site that has run a large # of cards thru most
Nikon and Canon models and calculated actual performance, and for the
most part Rob's numbers sort of agree with the tests discussed in the
link that started this thread ...
http://www.robgalbraith.com/bins/mul...e.asp?cid=6007 ... click on
the drop-down menu to choose your camera model.

Correct me if I am wrong, but there is no card that cares whether it
is in a camera or ina card reader. I mean there is no proprietary
function of a card to tell it thatit is in a camera or in a computer.

But sometimes you see a big difference between cards of the same
nominal speed when they are used in different cameras, or even the same
camera with different firmware ... the best example I've seen of that
is when I tested my Lexar 80x Pro cards against SanDisk 66x Extreme and
Ultra cards in my Canon 1D Mark II and with the original firmware I got
almost identical write times, 24-25x with all of them ... then Canon
came out with a firmware change that supposedly sped writes up for
"certain cards" and with this 1.1 firmware patch the Lexar improved to
29x but the SanDisk cards jumped to 45x.

So clearly this was a case where the performance cards of nominally
similar speed was up to 50% better in one camera just due to a firmware
update. That's why the tests should be run in real cameras, I feel ...

Bill