** Test Report ** Compact Flash Card speeds in 10D

[CyberDyneSystems]

TESTING CF WRITE SPEEDS OR "X-SPEEDS" IN 10D

I had done a test like this before,. but today I had access to two more cards,. and I also have a digital stop watch whcih I did not have before, so I thought I'd give it another go to "re-certify" my findings.

CARD MODELS TESTED
I tested a new Viking 1 Gig "standard" CF Card,
My Transcend 1Gig "30X" CF Card,
My Sandisk 256MB "standard" CF Card,
And a freinds Lexar 512 "24X" CF Card

HOW I TESTED
I set the 10D to "review off", Manual exposure, manual focus.
I tested only full buffer flushes of 9 shots in burst mode.
I did this with Large Fine jpeg (new) and with RAW as in my previous test.

The times listed are from the time I began shooting the burst. Each burst of nine was exactly 3 seconds,. so subtract three seconds from the scores to get the time from just after the 9th shutter release


RESULTS

CARD MODEL ______________ Jpeg Time __________ RAW Time

Transcend 1Gig "30X" _______ 20 seconds __________ 46 seconds

Lexar 512 "24X" ___________ 21 seconds __________ 46 seconds

Viking 1 Gig "standard" _____ 21 seconds __________ 48 seconds

Sandisk 256MB "standard" __ 23 seconds __________ 54 seconds


CONCLUSIONS

No suprises here really,. just a reminder that the 10D's CF card write speed is very limited. In fact to such a degree that there is essentially no value what so ever to high write speed cards in this particular camera.

The 10D's two stage buffereing algorithms on the other hand are pretty quick,. it is the write from the buffer to the card that is slow enough that it simply can not write to the card any faster than "standard" CF card speed.

The Lesson?

Once again,. I urge you,. high speed cards are not worth there premium. Yes you will get MUCH faster downloads in a fast Card reader that will take full advantage of the cards speed. But they won't offer ANY advantage writing in the 10D,.. which is when it counts most.

[Vegas Poboy]

Great info, Thanks but how come no lexar cards? I'm sure it does not matter just had to ask :)

[Tom W]

Interesting - as you said, the camera seems to determine the download, though the standard sandisk was a little slower. I wonder how they compare when uploading pictures via a USB 2.0 or firewire reader. That may, for now, be the only reason that a faster card might be useful. Of course, future cameras may take advantage of the faster cards in the future (unless "24X" doesn't really mean 24X).

[Flint350]

Seems to me, since the 10D doesn't support WA, why spend the extra bucks for it. The test results are mostly within the admitted "insignificant" difference as long as you stay in the top half of the chart. I'm sure my Sandisk Ultra II, while milliseconds slower than a Lexar 45X, was a better value.

[sjprg]

But if you get a new camera that supports USB 2.0 or firewire and can write to the WA specs, now your stuck with a bunch of slow CF cards. Plan ahead.
Paul

[FotoPhreak]

But if you get a new camera that supports USB 2.0 or firewire and can write to the WA specs, now your stuck with a bunch of slow CF cards. Plan ahead.
Paul

Couple of things ...

Is the above likely to happen anytime soon?

Can anyone offer a cost comparison between different models/brands vs their so called 'better' write speeds?

BTW, CDS you said that the better write speeds do come into effect when downloading the photos ... do you have any times for these? (with what card reader, or doesn't it matter?)

[dtrayers]

BTW, CDS you said that the better write speeds do come into effect when downloading the photos ... do you have any times for these? (with what card reader, or doesn't it matter?)

I have. The results are below:

http://home.comcast.net/~dtrayers/photos/CFSpeeds.jpg

I was interested in testing the download speed of a standard PCMCIA reader vs. my new Delkin Cardbus32. You see the large difference between the Sandisk Ultra 256 and the Ultra II 512.

I did an in-camera write test similar to CDS's and found that the Ultra II was slightly faster.

At buy.com there is only $15 difference between the 512 Ultra and the 512 Ultra II, and $45 between the 512 standard and the UltraII. The $45 is worth it to me for a 5x increase in download speed.

[FotoPhreak]

At buy.com there is only $15 difference between the 512 Ultra and the 512 Ultra II, and $45 between the 512 standard and the UltraII. The $45 is worth it to me for a 5x increase in download speed.

I agree, definately worth it!

Thanks for the post.

I just needed reassuring.

[vsolanoy]

The Lesson?

Once again,. I urge you,. high speed cards are not worth there premium. Yes you will get MUCH faster downloads in a fast Card reader that will take full advantage of the cards speed. But they won't offer ANY advantage writing in the 10D,.. which is when it counts most.

I learned that when I purchased a SanDisk Ultra II 256 meg card. I had a standard SanDisk 256 Meg card and when I just did observational tests, both cards performed identically on the 10D.... the camera's buffering seems to make up for any media shortfallings.

However, the Ultra II does come with a "lifetime" warranty... that could be worth the extra $20... either that or I'm just trying to justify my own gullability :oops:

[JoeTampa]

I decided to try the same test with my 1G Microdrives:


JPG: 18 seconds

RAW: 54 seconds

[CyberDyneSystems]

Great info, Thanks but how come no lexar cards? I'm sure it does not matter just had to ask :)

Errr???

RESULTS

CARD MODEL ______________ Jpeg Time __________ RAW Time

Transcend 1Gig "30X" _______ 20 seconds __________ 46 seconds

Lexar 512 "24X" ___________ 21 seconds __________ 46 seconds

Viking 1 Gig "standard" _____ 21 seconds __________ 48 seconds

Sandisk 256MB "standard" __ 23 seconds __________ 54 seconds

[CyberDyneSystems]

There are reasons that someone might want to spend the extra $$ for the premium cars,. I'll agree to that. Warranty,. download speed,. use in future cameras.. etc.. all reasons to buy the "Ultra" or "High X" cards.

But for us Canon owners... (actually I don't know about the 1D models?? Anybody?) ...for us Canon Non-pro body owners,. shooting speed is not one of them.

I just received my second Viking card tonight which prompted this second test. I allready own a Viking 512. It was when I got the Transcend 30X in June or so that I did my first test and like so manuy others learned the hard way that the extra "X" does nothing in my Camera. (lots of people actually won't even admit it! :) )

The Viking was $169.00 compared to a Lexar at Double the cost. The Viking has a 5 year warranty... not bad? assume in 5 years this 1 gig card will be about as usefull as the 4 and 8 MB Smart media cards that once impressed that now collect dust in the bottom of a drawer.

[J.A.F. Doorhof]

Strange I switched from a 512MB Xtra HS to a 1GB Prof. Speed from the same brand and noticed a rather high profit in speed ??
I shot with a burst of 9 frames in RAW, the speedincrease was arround 15%.

Greetings,
Frank

[CyberDyneSystems]

Really?? WOW! I'm getting no difference to speak of???

Still,. only 15% increase when the card claims to be 12X, 24X, 30X or 45X faster is triffling.

[J.A.F. Doorhof]

I looked it up.
9 frames on XsHigh speed 39 seconds.
On the pro-speed 29 seconds.

So 25% speed increase.

Greetings,
Frank

[CyberDyneSystems]

O-kay,. now I'm beginning to think it is just my 10D :( :( :( :(

[J.A.F. Doorhof]

Don't.
Maybe it's luck (I did the test 2 times ;D).
Remember it was with 9 frames in RAW, on one file this should not be noticable I guess.

Maybe you had a very fast card to begin with ?

I tested the same brand but different speeds.

Greetings,
Frank

[robertwgross]

Before you try to do a standard write test to a CF card, I would suggest that you do a complete format in the computer, i.e. scrape out any bad sectors. Then format in the camera.

---Bob Gross---

[J.A.F. Doorhof]

But offcourse, I did that. :D

[Tomsk]

Take a look here
http://www.robgalbraith.com/bins/multi_page.asp?cid=6007

I've done a RAW test using my 300D and an Integral 256MB card and I get results of ~650K/sec or about ~40 sces to write 4 RAW files to the CF.

(I'll be supplemnting the Integral with a faster and bigger card 'real soon now' :D )

[CyberDyneSystems]

I am well aware of the Rob Galbraith tests,. and though he has my infinite and undying respect,.. I did my own as I frankly do not know what the heck he is testing.

kbs transfer rate?

what? How does he arrive at those numbers. What instrumnet measures it.

I want :
A: Real world tests. IE: How long does it take to flush the buffer
B: Results posted in seconds,. not theoretical transfer rates.

He even has a disclaimer stating that "your real world use may vary"

Well the test should real world. What test can you possibly do to be given a number in KBS?????

If the KBS rats IS higher from one card to the next,. what does that translate to in seconds? Is the Fastest card in the 10D 30 seconds faster than the slowest card in a 9 frame burst of RAW files, or is it 1 second faster????

The Galbraith test does not say,.

So the test tells me nothing of interest.

I use a stopwatch with the camera aimed at it to guarantee I have started the burst at the same time for each test.

The time I give are from the moment I press the shutter untill the red "write light" finally stops (ie: the complete 56 MB or so of RAW data is written from the buffer to the CF card.

I did each test twice. Again, Manual focus,. Manual exposure, no playback etc...

In fact,. it is such a miserable cold here when I get out of work I may try it all over again! (although I won't have my freainds Lexar Card this time)

[Tomsk]

This is how I did it.

300D mounted on a tripod with a camcorder focused on the back.
I then took 4 sets of 4 pictures and from the camcorder tape I got the total time that the write LED was lit up. The CF is placed in a reader attached to my PC and the files sizes (16 each of .CRW & .THM) are added up. From this info I get KB/s, which I figure to be how fast the camera is writing to the CF card.

A further test I've just done is to interrupt the writing process by half pressing the shutter button. I did two seperate tests, first with just one 10 second interrupt and the second with 3 interrupts of 5, 10 and 15seconds. From my videos the total time taken to write, continuous write time is <1sec faster than interrupted.

Times for the above tests ~36 seconds for four frames. Slightly faster than my previous test, 'cos the subject was brighter (different TV programme!)

[slin100]

I want :
A: Real world tests. IE: How long does it take to flush the buffer
B: Results posted in seconds,. not theoretical transfer rates.

The problem with quoting numbers in seconds is that it's meaningless unless you also provide the amount of data being transferred. That is effectively what Rob is doing by providing a kbps value. Perhaps, the raw data (e.g. number of files, sizes of each file, and total transfer time in seconds) might have been more useful.

[AJSJones]

I frankly do not know what the heck he is testing.

kbs transfer rate?

what? How does he arrive at those numbers. What instrumnet measures it.

I want :
A: Real world tests. IE: How long does it take to flush the buffer
B: Results posted in seconds,. not theoretical transfer rates.

He even has a disclaimer stating that "your real world use may vary"

Well the test should real world. What test can you possibly do to be given a number in KBS?????
So, CDS - did you read his description of what tests he does and how he does them? Or did you just not understand them, and therefore dismiss them? His listed rates are not theoretical, they are a standardized way of describing read and write rates.

If the KBS rats IS higher from one card to the next,. what does that translate to in seconds? Is the Fastest card in the 10D 30 seconds faster than the slowest card in a 9 frame burst of RAW files, or is it 1 second faster????

The Galbraith test does not say,.

So the test tells me nothing of interest.


KB/second is the only way to represent read/write speeds in a meaningful way, and they are even averages of triplicate measurements. It's simple arithmetic to calculate how long it will take to write a specific number of KB to the card : at 1000 KB/sec it will take 2 seconds to write 2000 KB. Number of seconds to write 9 raw (or jpg) files will depend on the size of the raw (or jpg) files and is not meaningful unless (as pointed out above) you know the size of your files. You don't say what size your files are so your "test tells me nothing of interest." A jpg of a shot with a lot of blue sky compresses into a much smaller file than one with lots of detail, and the same is true for crw files, so # of jpgs per second doesn't help anyone.

I recommend you follow the link above and read his methodology. Once you understand that, and how thorough a job he does, you will appreciate what a great resource his database is. Not intending to "preach" but no-one is influenced by someone who says "I don't understand it, therefore it's bunk"

Andy

[CyberDyneSystems]

Hrmmm..

O-kay, first mistake,. I read the first pages of his testing three times,. and did not see exactly what was done to arrive at the KBs numbers,..

So,. I blew it there, of course it is described on the 10D page,. not on the first page,. as it needs to be because it will be different for different cameras.

My mistake.

2Nd mistake,. I did not look at the fact that despite my beleif that what I was testing was a generous representation of faster and slower cards,. in fact of the three cards I tested that Rob tested as well,. they all tested very similar on his chart. So my cards do NOT represent a good cross section of fast to slow cards,. they all seem about the same speed.

RE: Kbs Vs. How long it took to flush the buffer.

I am fully capable of doing the math. My problem is that my tests didn't agree with the conclusions he draws from his tests.

But doing the math,. the fastest 1366KBS cards willl clear a 48MB buffer in about 35 seconds.

The "normal" cards at about 1048KBs will do it in about 45 seconds,. which is close to what I was measuring with all of my cards. (indeed a "noticeable" difference... but still not huge. Not when someone is marketing the card as 45X faster :) )

Now here's the rub,. if my cards behave like Rob's cards,. then that puts my Lexar 12X right where it should be in his line up,. IE: not really any faster than a plain vanilla card. In fact the Lexar 12X is a dog in the 10D for some reason. The Microdrive beats it. The Transcedn 30X is also rated pretty slow,. especially for a "30X" but according to Rob's chart,. it is indeed the faste of the cards that I had of the three he has tested as well. So it turns out that the Viking "plain vanilla" is pretty peppy for a plain vanilla in a 10D.


O-kay so here's where I eat crow,. I was definately wrong to disparage the data that Rob listed.

It is his conclusion here that bothers me;
Our strong recommendation is to consider only the fastest cards available for this camera,
I'll get to why I disagree with this in a second,. but I still find this conclusion to be in tiotal contradiction to these statements ,. statements which I agree with 100%

...though it should be noted that the performance difference between the top 10 or so cards in the 10D is relatively narrow.

And
The camera writes relatively slowly, regardless of card.

Hers the where his own data favors my argumant.

The difference shot to shot from the fastes cards to the slowest cards is minimal for the 10D because it won't take advantage of the cards speed.

The "averadge" cards speed is about 1048KBS in the 10D,. the fastest is 1366KBS,.

Now in the 1D the difference between the "normal" card is about 1300KBs for the SLOWEST card,. and a whopping 3183KBs for the fastest.

In the D2h the slower cards are again about 1300 (or less)
But the fastest cards will earn you 5238KBsec!!!!

So in the 1D if you pay for the fastest card you can nearly triple your speed (well more than double anyway)

In the D2h you can get a 500% speed boost (or a TRUE 5X)

In the 10D all the extra money will get you at the MOST a 35% increase.

So my conclusion is that in the 10D it is not money well spent.
In a camera that gives you results for the extra cash spent on fast cards,. such as the 1D or D2H,. THEN it is really worht the extra.

But that is of course just my opinion.

Again,. I apologize for speaking ill of Rab Galbraith's tests.
There is no question that it is a great resource for all of us. A more complete set of data on CF card speed can not be found anywhere else.

[AJSJones]

CDS, Good post!

I have a 10D and had only really looked at the data for cards in that camera. It would be interesting to take a couple of cards (e.g. the fastest so far and a middle of the road such as a 1 GB Microdrive) and compare all the cameras in one chart. I had not realized, until this thread, how widely the cameras vary in their write speeds. However, I don't tend to need real fast bursts or frame rates so it's not a particularly high priority for a camera purchase decision for me - if I were a sports shooter, though, it would be a biggy!

I agree with you about the conclusion you cite about getting the fastest card possible for the 10D - going to a card capable of lightning speed won't really help a 10D shooter take pix more quickly than a card that is just a little faster than the camera's highest write speed. However, unless there's a high price premium for the extra speed, it does make sense to get the fastest available - these cards are similar to lenses in the sense that you keep them and re-use them when you buy a faster, newer, shinier, more MP camera. I'm happy with my 1GB microdrives on my 10D but if I upgrade to a newer camera (bring on PMA!) I'll be re-thinking my card choices where I want fast shooting speeds.

Cheers!
Andy

[CyberDyneSystems]

Andy,

A good point you make regarding upgrade path (I too think I will be in line for either a "new" 1D version,. or maybe a cheap used old version when the price drops)

Thanks for hitting me on the head back a few posts :)
I needed it!

By disregarding other data and generally being a jerk, I was invalidating my own point.

Thanks for bringing me round to my senses :)

[PacAce]

However, unless there's a high price premium for the extra speed, it does make sense to get the fastest available - these cards are similar to lenses in the sense that you keep them and re-use them when you buy a faster, newer, shinier, more MP camera. I'm happy with my 1GB microdrives on my 10D but if I upgrade to a newer camera (bring on PMA!) I'll be re-thinking my card choices where I want fast shooting speeds.


Just to be the devil's advocate, wouldn't it make more financial sense to get the fastest card that your current camera can take advantage of at the cheapest price possible? After all, CF card prices are dropping as faster and faster cards are built. If you need a faster card in the future for that new camera you're going to get, wouldn't the same card in the future cost less than it would cost you now?

I guess what I'm trying to say is that CDS does make a good argument (in my eyes at least) regarding getting a faster card using today's money for tomorrows camera. :)

And, BTW, I don't think CF cards are like lenses. Lenses will usually outlast serveral camera bodies. I'll be lucky if my CF card lasts a year or two of constant usage.

[CyberDyneSystems]

True.

I said further up the thread that I was unsure how long a CF cards use would last. The memory crds I got with my first camera were Smart Media,. a discontinued type pf memory,. but more to the point,. even if my first digital was Nikon,. (and therefore used CF cards) the cards I spent a pretty penny for then would be of no use now,. they would be 4 and 8 MB cards!!!!

In 4 or five years,. I truly doubt will be using CF.