FYI....I
Printable View
FYI....I
Same here guys, Cheers!
I hope you guys realize that the 7D cuts it's frame rate in-half in dark light or with the lens cap on, and the MKIV does not.
If you're shooting with the lens cap on with the 7D then this will slow the frame rate down, therefore, it will take longer for the buffer to fill up, which will make the RAW shot burst data appear higher.
Aren't you really more concerned to see if the CF card can keep up with the maximum burst rate (when you see the busy signal) as opposed to 1/2 the burst rate?
Rich
Rich
One thing I wanted to find out, is if your in a situation where you have a target that you are continually shooting would the faster card help or is it just wasted speed.(lets say big foot just come out in front of you and has been there for 2 minutes, I would want every picture I could get)
The original idea was to find out how fast the camera would write to the card, and how many mbs it was writing. We were trying to find out if it worth spending the extra money on a 90mbs card. Unfortunately I don't think we have fully answered that, however it appears that both bodies will not write faster than the 60mbs. Maybe when John gets his, he will be able to compare both.
One thing I noticed is that the shorter the burst, the faster mbs it writes. So going for 20 seconds it is bottle necking in the camera somewhere. But the fastest rate I got was 56mbs.
But I did find it interesting that the bursts were 35-40 frames rather than the 26 published in the book.
As far as the lens cap goes, if that is the case there is no way to draw equal comparisons between the 7D and 1D IV if what you say is correct. Unless of course you have them set up side by side with equal light.
What puzzled me was why I was getting slower rates than Brant was on the 7D. I think the answer is I had the ISO on Auto on the first round of tests. Cap on it probably was shooting at 6400 and was just chocking down.
Rick
Side Note: I just sent an email to Hoodman and asked why their $400 32GB card wasn't keeping up with the 16GB card.
Quote:
Originally Posted by HDNitehawk
The way I understand it, is that you want the card to write faster than the speed of the buffer, so this way you can clear the buffer faster to keep on shooting.
Lets say we take Canon's conservative numbers, in that the 7D can shoot 15 RAW shots in a burst. At 20MB each shot that would be 300MB of data that we have to get out of the buffer quickly so that we may shoot another burst. So, if the card can be written to at 30MB/sec then it would take 10 seconds to clear the buffer, however lets say it only takes 3 seconds to fill it. So, you will start to see a backlog. That's why it's good not to hold your finger on the shutter continuously, because you wouldn't want Bigfoot to come out while you're getting a busy signal. Now keep in mind that you don't have to clear the entire buffer before you may resume shooting, you only have to clear some space for more shots.
Now, if we have the 60MB/s card then we can clear the 300MB of data in our buffer in 5 seconds.
With the 90MB/s card the hope was that our 300MB of data would be cleared out of the buffer in 3.3 seconds. Now, were talkin'. However, if the data from everyones testing is correct, then this test seems to demonstrate that even though the "cards can be written to" at faster speeds, the camera can only "write to the card" at approx 40MB/s for the 7D and 50MB/s for the 1D MKIV.
So, if the testing above is accurate, then there doesn't seem to be any advantage of the 90MB/s over the 60MB/s.
By the way, I hadn't noticed any difference between these two cards in actually shooting, but I never tested them.
So, to answer your question it does appear that the faster 60MB/sec card (as compared to the 30MB/sec card) does help with the 7D, and perhaps the 90MB/s card will help when we get some faster bodies from canon.
Digic V here we come!
Can you say "Hello to My Litlle Friend" the 1D X?
Rich
Quote:
Originally Posted by Richard Lane
Yep, all of my tests were conducted with the cap off. The "test" where Rick and I had similar set ups, I was shooting at a white piece of paper that was lit by a desk lamp. My burst rates have been consistent at 22-24 frames in ~3 seconds and then it slows down. With the setting we had agreed to, the exposure was still something less than -3EV. In those tests, I had ~40-43 MB/s write rates. Previously I seen ~48 MB/s, but I am wondering if the ~5-8 MB/s difference is just error in my calcs or tests. Anyway, I should get my 32 GB 90 MB/s card in soon, so I'll see if it is an improvement with the 7D.
Quote:
Originally Posted by Richard Lane
Rich
Just when I think I have all the answers I want, a few more pop up.
With the 32mb card I did the 3 second burst, which gave about 26 shots. It cleared in 7.5 to 8 seconds. It wrote at 51KBs compared to the 38KBs test for 20 seconds.
I didn't test the 16mb card in this scenario, so I thought I would give it a go. Here are the results:
- 4 tests for the average
- 3 second burst
- 430mb on the card
- 26 average pics
- 6.7 seconds to recover from the initial start
- 64MBs
So then I decided to try for 1 second
- About 4 tests and it was very hard and not as accurate
- 159 mb on the card
- 9 pics
- 3.1 to recover fully
- 51MBs
So with the 16mb card recovering from a full burst, it was writing 64mbs. This is the only instance so far over the 60mbs number. So 4MBs advantage, that would be a .41 second advantage over being limited to 60MBs.
Maybe on the next Jack Links Beef Jerky commercial I will set up and practice on Sasquatch and see if the .41 second is going to matter.
Congrats, on breaking the 60MB/s barrier!
I
Ok, I just received my new Sandisk Extreme Pro 32 GB 90 MB/s card. And it is the slowest of my three cards.
Test results for the card using the same settings as before:
3 sec test: Shots=23, File size= 418 MB, Write Time = 23 seconds for a write rate = 18 MB/s
20 sec test: Shots = 42, File size 731 MB, Write Time = 45 seconds for a write rate = 16 MB/s
For comparison, my Sandisk Extreme 8 GB, 60 MB/s card was giving write rates under identical conditions of 40 MB/s last week. I tested again today and I got:
20 sec test: Shots = 66; File Size = 1.21 GB; Write Time = 31 secs for a write rate = 39 MB/s
With my Sandisk Extreme III 4 GB, 30 MB/s card I just got:
20 sec test: Shots = 48; File Size = 915 MB; Write Time = 38 seconds for a write rate = 24 MB/s
So, by far, my best performing card on my 7D is the 8 GB, 60 MB/s Sandisk Extreme.
I'm am interested to see what others find.
Thanks,
Brant
Edit:
I wanted to put my results into a Table:
[img]/resized-image.ashx/__size/800x0/__key/CommunityServer-Discussions-Components-Files/24/0488.CF-Card-Write-Rate.jpg[/img]
It is interesting, but the efficiency "Percent of Rated Speed" is nearly linear (R2 = 0.989) with the card size when comparing the 20 second test results. I tend to not get too excited about 3 data points, but if someone has a 16 GB card, it should clock in at ~50% (ok....51.6% according to the formula) of it's rated speed, if this is linear and holds at all.
Brant
I found several blogs and articles on the net, and the consensus is that then maximumwrite speed of the 7D is 48mbs to 53mbs.
Since your 8mb card is 60MBs, what your seeing is what I saw when comparing the 16mb vs the 32mb card.
Since my 16GB card was faster than the 32GB card, could it be that the camera does write slower as the card gets larger? It is starting to appear that way.
Rick
Rick,
That is what I am beginning to think. I had expected the 90 MB/s card to come in similar to the 60 MB/s card, expecting the limitation was the camera and not the card. But that doesn
Hey Brant,
Something must be wrong with the card or testing. Did you try to reformat the card and run the test again?
Rich
Hi Rich,
I formated before testing, reformated between each test, and ran each test twice. I know, something seems off. Initially I wasn't planning on retesting my other two cards, but I did because I wanted to see if they had similar results as before and they did. I'll do the test again tonight to see if it is different. Maybe the new card was cold as it had been sitting outside? Maybe cards need to be "broken in" a little? I am open to suggestions as I was expecting 40-50 MB/s and got 16-18 MB/s.
Thanks,
Brant
EDIT---I was just re-reading this thread and Rick has run a 16 GB card....the linear relationship predicted 51% efficiency at 16 GB...Rick got 51, 56, and 64 MB/s on a 100 MB/s card on the 1DIV. That's four points on the trend line....still, not great, but it is getting there. I'll re-run my new card tonight.
Quote:
Originally Posted by Kayaker72
Brant, Remember, My test was only total shots over 15 seconds--I didn't consider file size, buffer lag time etc. I went back and looked at my CF cards and discovered that on the 16mb card average file size was about 20,480KB and on the 32 the average file size was 20,382 KB, I realize that isn't much, but I have no idea why there is any difference at all since all shooting paremeters were identical. I would be interested in any guesses.
Brant,
How are you calculating your Write Time (sec)? I think your numbers (23sec and 45sec) sounds too high.
Rich
Hi Rich,
I am assuming the camera starts to write the second I start shooting (there is a very slight lag to the red LED/access lamp going on) to when the Red LED/access lamp is off. So those times include the 3 seconds and 20 seconds (respectively) of shooting.
I agree, I am surprised by the performance of the 32 GB 90 MB/s card. As you said, I may have just received a bad card.
Thanks,
Brant
Something is wrong with your data. How can it take 23 seconds to write a 3 second burst? I think you have a bad card.
Quote:
Originally Posted by Kayaker72
Brant, One thing Dawned on me---Maybe we are mixing apples and oranges and expecting the same results:
On my test, I only timed 15 seconds andwas getting some significant buffer lag by the end of the test.
On your test, you did 20 seconds, so I can only assume you exprienced the same the last few seconds of the test.
Now your total calculationsmost likley included those "slow moments" when you are only getting one or two shots off per second
With your test being 20 seconds long you had 5 more seconds of "slow" to throw into your calculations, which would reduce the overall calculated speed.
My suggestion is for two or three of us to use the exact test parameters (or as close as possible) and then see what happens---My guess is they will probably be fairly close.
The other thing I noticed is that I reported an average "per file" size whereas you reported the sum of all the file sizes. If assumptions andreverse math is correct you were getting individual file sizes around 18MB per file and I was getting around 20 MB per filewhich tells me weprobably haddifferent camera setups.
If one of our science based forum members are willing to design a test we can all do, I would be more than willing to participate
Bob
I also received my 32GB 90MB/s SanDisk Extreme Pro CF card.
With the 7D, I just shot 70 raw shots which totaled 1.31GB of data for 20 seconds and the red access light stayed on for an extra 7 seconds after I stopped shooting, for a total of 27 seconds.
I ran the test 3 times with similar results.
1310MB/27sec = 48.5MB/sec.
@Bob , my first tests I was getting larger file sizes. Once I turned off the Auto ISO I didn't have that problem.
If you were going to compare from one person's 7D to the other, the simplest way to match settings would be to go in to the camera and just restore it to the default settings. That way everyone would start with the same setting.
One problem I see, is if you are doing it without the lens cap, each of you will get different results. I noticed this with the 1D IV, I did a series without the cap and I was outside. A bit later I did another series and got different results, the only difference was the lighting. It was outside and the sun had gotten lower in the sky. When I did tests with cap on the results were consistent from one test to the next.
It would be interesting to see if everyone that bought one of the "on sale" cards, get the same results.
@Brant..I talked to Hoodman earlier, explained what I found on the testing (16MB vs 32MB). They said it is possible the camera takes longer to write to the larger card. They took my phone number, camera type info and told me they were going to check it out themselves. Odd though they said they had not heard this scenario / complaint before.
Even if that is the case it sounds like your card is really slow.
7D with 32GB 90MB/sec SanDisk Extreme Pro CF card:
In a 3 second burst; I shot 22 RAW shots totaling 408MB. The red access light stayed on for 5 seconds after I stopped shooting for a total time of 8 seconds. I ran the test 3 times with similar results.
408MB/8sec= 51MB/sec
Quote:
Originally Posted by Richard Lane
Rich
Do you have a smaller card you can compare to, see if there is any difference?
No, but I have a larger card, namely the SanDisk Extreme Pro 64GB 90MB/sec CF card.
I ran the test 3 times with eachand posted the average speeds:
With the 32GB 90MB/sec Card:
3sec burst test= 51MB/sec
20sec test= 48.5MB/sec
With the 64GB 90MB/sec card:
3sec burst test= 47MB/sec
20sec test= 44.5MB/sec
The larger 64GB card was a little slower!
<div></div>
Quote:
Originally Posted by Richard Lane
Hi Rich,
Good, I am glad to hear it. I just re-ran my tests and the new 32 GB 90 MB/s card is still writing at 17-18 MB/s. I must have received a bad card. I'll be contacting B&H, not so much that my shooting style needs 90 MB/s, but more because if one thing is wrong with the card, what else is wrong with it?
Bob,
I'd be happy to run additional tests. I have been conducting mine with the following conditions:
- EF 100 mm L, AF off, IS off, auto review off, manual mode set to 1/100, f/5.6 and ISO 100, RAW only, High-speed continuous mode
- Shooting at a blank white paper
- Recording the number of shots, file size and write time (red LED) after shooting for 3 seconds (typical time to fill buffer) and 20 seconds.
My file sizes are typically 17.5-17,.8 MB per picture.
That said, I buy Rich's results. It makes sense that the 7D is limited to the ~50 MB/s range and that cards can perform up to that range. I still think I will stick with the 90 MB/s even though it is not needed for the 7D just because it may be of use for something down the line.
Thanks for all your help,
Brant
Quote:
Originally Posted by Kayaker72
Brant, I suspect there is nothing wrong with your new card. You just need to update your 7D's firmware. I did notice that the most recent updates for both the 5DII and 7D included, "Improves writing/reading speeds when using UDMA 7-compatible CF cards." Now, I realize that the 90 MB/s cards are rated UDMA 6, but still, it seems reasonable that something stated to affect speeds for faster cards might be the issue.
I've been running older firmware on both my bodies for a while, not seeing a need to update. ButI just ran tests based on your settings, first with the older versions I had been running (1.2.2 on the 7D and 2.0.8 on the 5DII), then after the update (1.2.5 for the 7D, 2.0.9 for the 5DII). I tested 4 cards (all SanDisk, 2 GB 20 MB/s, 8 GB 60 MB/s, 16 GB 60 MB/s, and 32 MB 90 MB/s) on both bodies. But the short version that addresses your issue, Brant, is that with the old firmware, the 32 GB 90 MB/s card performed only slightly faster than the old 2 GB 20 MB/s card, but with the current firmware, performance of the 90 MB/s card on both bodies was slightly better than the 60 MB/s cards (both 8 and 16 GB, which weren't different from each other).
I'll tabulate all the results and post them (probably not until tomorrow, I'm pretty busy for the rest of the night).
Hope that solves a mystery for everyone, and fixes your problem, Brant!
--John
After reading John's post, I went back to check if I had the current firmware update on the 1D IV. Sure enough I didn't
I rechecked my results. After the firmware update I am getting identical results to those I had previously. So on my end no mystery solved yet.
"cough" ....um...yeah....I would like to amend my previous results. The new 32 GB 90 MB/s card gets 48 MB/s transfer rates, 24 shots, 430 MB from a 3 sec burst and and 45 MB/s tranfer rates and 75 shots, and 1.3 GB from a 20 second burst on the 7D....:)
Quote:
Originally Posted by neuroanatomist
Mystery solved......I and an unsuspecting B&H employee that doesn't have to deal with me trying to return the card thank you.
Brant
Can you repost the link you did earlier for the 1D IV test. It was lost in the move.
John
We lost your chart as well :(
Rick,
Sure...
http://www.slashgear.com/high-perfor...rk-iv-1577718/
Unfortunately it seems a lot was lost in the move, but I have to say, I do like this new forum.
Brant
Brant
Thanks, I was talking to Hoodman today, and I was trying to direct him to that website but didn't have the link.
Thought I'd throw my own results in here, even though I've only got 2 cards the same speed and one 7D body.
70-300L, 70mm, f/5.6, iso100, 1/125s, pointed at a white wall. TC-80N3 set to 20s-long 'bulb', but camera set to HSC. Fully charged battery, cards formatted in-camera. Firmware 1.2.3.
ALO off, PIC off, all noise reductions off. I didn't time the length of the "busy" light or the red "write"-light afterwards
8GB Sandisk Extreme 60MB/s card, RAW-only: 68 frames, 1.2GB (1,304,866,608 bytes).
16GB Sandisk Extreme 60MB/s card, RAW-only: 69 frames, 1.2GB (1,324,110,624 bytes).
a) that's a good result, because I bought my 16GB card in hong kong for the same price as the 8 in Holland (ie, it's genuine or at least a good-speed ripoff).
b) Not sure why the 8GB missed its last frame, maybe because I forgot to format it but did format the 16GB?
c) viewfinder says 16 images possible, but I get 22 before the slowdown to bursts-of-2 roughly every second.
d) 18MB per .CR2 file.
8GB card, RAW+LJPG: 56 frames, 1,240,964,468 bytes.
16GB card, RAW+LJPG: 57 frames, 1,262,978,180 bytes.
a) again the 16GB gets one more frame in (both formatted in-camera pre-test this time).
b) viewfinder says 6 possible shots, but I get 8 before the slowdown.
c) I always thought that the jpg-processing overhead wasn't much, seems I lose ~60MB over a 20 second burst, ~3MB/s slower.
d) 18MB CR2, 3MB JPG or so.
8GB card, LJPG-only: 153 frames, 435,550,489 bytes.
16GB card, LJPG-only: 153 frames, 429,442,384 bytes.
a) size difference was about 2 JPGs-worth, but same amount of frames.
b) Viewfinder said 96 shots available, but absolutely no slowdown in 20 seconds. Should i add more to my shutter-count to see how many JPGs I can get before a slowdown?
c) 2.7MB per JPG or so.
For the hell of it, I found an old 512MB "Pretec 80x" card that i've been using as an IDE drive in a firewall (built from an old pentium 1 using ipcop) for a few years and recently upgraded.
RAW: 20 frames or so, then 2 second wait, frame, 2 seconds, frame, 5 seconds, frame, then Full CF (maybe missed out on one frame in the end). 25 frames, 476,331,819 bytes.
RAW+JPG: still get 8 frames before the slowdown, then a few bursts of 1 or 2, a 5-second wait, 1 shot, a 5 second wait, then the 20-seconds is up. The "writing" led doesn't go out for about a minute or two afterwards, I thought I'd broken my camera. 20 frames in total, 440,834,628 bytes.
JPG: 13-seconds before a slowdown, then a few bursts of 1 or 2 frames each. 113 frames, 315,670,763 bytes.
The results for the crap-card are interesting in that you get an idea of the performance of the buffer. ie, I still got nearly the same number of frames before the slowdown in all but LJPG, but once the slowdown started, the camera's nearly unusable for 2 minutes.
Dr. Croubie....I like your results and I think they highlight one of the things I've learned doing the testing. I think the burst number of shots shown in the viewfinder only reflects the images that can be written to the buffer but does not account for images being cleared from the buffer during shooting. In otherwords, the number of shots you can take until the buffer is full is the buffer size/image size + write rate * time/image size, or, for the 7D, the 15 Images the viewfinder says plus the extra 7-8 images that can be written in 3 seconds of shooting (with a fast enough card) equals the 22-24 burst images we've seen in our test results.
Speaking of the test results, I wanted to summarize those that were conducted using the same methodology and I assume John/Neuro will repost his table....
Attachment 16
Rick and Rich, please let me know if you would like me to modify in any way....
Thanks,
Brant
Looks like CF card prices are back up...almost double what they were.