Tiggy, just to be clear on your last post, its not as much on whether you are using EVF vs LCD when shooting but rather after shooting while the buffer is clearing? So if somebody was using the EVF to shoot a substantial sequence of shots and continued to hold it up to their face it would take the longest to clear? Moving the camera away from the face immediately when done shooting would be faster and shooting/clearing buffer purely by LCD would be the fastest?
EverLearning wrote:
OK. Here are the test results from the new Delkin 128GB CFe card I just received. Same test process as I outlined on page 13 of this thread. To summarize key points, freshly charged battery (last night), card low-level reformat before each test. MS, FF, H+, FL 400mm, focused and then set to MF, 30 second continuous shooting. Used a trigger and iphone timer to be as accurate as possible. Two tests set to RAW and two tests set to CRAW. All shot at 1/800 sec. Unfortunately, I didn't reset my aperture so that and ISO are different between the two sets and are specified.
RAW - 5.6, 6400 ISO (also, EC was at +1/3):
Test 1: 316 photos @ 60.9MB. 18.79GB/30s, 641.5MB/sec.
Test 2: 318 photos @ 60.9MB. 18.91GB/30s, 645.5MB/sec.
Battery now at 94%.
Other observations: first stutter at 12 seconds shooting. It took about 2.5 seconds for buffer to clear once I stopped shooting.
CRAW - 4.5, 2500 ISO
Test 1: 339 photos @ 29.5MB. 9.77GB/30s, 333.4MB/sec
Test 2: 335 photos @ 29.5MB. 9.65GB/30s, 329.4MB/sec.
Battery now at 91%.
Other observations: first stutter at 20 and 21 seconds shooting respectively. It took about 2.5 to 3 seconds for buffer to clear once I stopped shooting.
General observations: Not really any different than most of my other tests with the Delkin 128. Really bizarre how the card seems constrained more by the number of files written than the actual amount being written. Apparently whatever is causing this situation has not changed between July and January....Show more →
Just so I'm clear and can try my own test like this, are you in MS or MS/EFCS?
This is an interesting piece of information. When I did my test, my eye/face was not in front of the EVF as I was having to operate the stop watch. Maybe that's why my numbers were higher.
tiggy wrote:
OK, so another interesting update. It turns out that a fellow over on DPReview's R forum (Tazz93) discovered that whether the EVF versus the LCD panel is operating determines how fast the cache gets cleared for CFexpress cards.
I came very, very close to not testing this, as it's just too insane to be true. But, after testing, yeah it is. The difference is pretty huge, too. 4.3 seconds to clear my Angelbird card with the LCD, and just under 20 seconds to clear it using the EVF.
This means that when you're mashing the shutter button and you've run out of cache, the fact that you're face is sitting there in front of the EVF is the primary impediment to the data clearing.
It also might explain why some people got very different results in testing. ...Show more →
I've confirmed this latest twist in both MS and ES.
Additionally: I have my camera set up to switch between EVF and LCD manually, with a button press. If I fill the buffer shooting via the EVF and leave it in EVF, it can take about 20 seconds for the buffer to clear. If, after filling the buffer via EVF, I push the button to switch to LCD, the buffer clears faster - in maybe half the time or less.
I have not tested with auto-switching between EVF and LCD, but I am guessing the the behavior would be similar - faster clearing when the LCD is active.
bobbytan wrote:
This is an interesting piece of information. When I did my test, my eye/face was not in front of the EVF as I was having to operate the stop watch. Maybe that's why my numbers were higher.
My more recent tests were done via the LCD as I had setup on tripod in the backyard to make things easy and consistent. However, some of my earlier tests on the Transcend were done with eye up to the EVF.
For your 400+ test were you in MS or EFCS(MS)? Also was your SS high enough to shoot 12FPS? Finally was the Green Icon present?
Mechanical. Not aware if there was a green icon. Drive mode was Continuos H+. I was getting about 410 frames in 38 seconds ... so it's about 10.8 fps. I think my shutter speed was 1/60s.
arbitrage wrote:
My more recent tests were done via the LCD as I had setup on tripod in the backyard to make things easy and consistent. However, some of my earlier tests on the Transcend were done with eye up to the EVF.
For your 400+ test were you in MS or EFCS(MS)? Also was your SS high enough to shoot 12FPS? Finally was the Green Icon present?
bobbytan wrote:
Mechanical. Not aware if there was a green icon. Drive mode was Continuos H+. I was getting about 410 frames in 38 seconds ... so it's about 10.8 fps. I think my shutter speed was 1/60s.
Maybe it was down in White H+ then which is supposed to be 10.6FPS or so. That would explain the 400+ shots. The R5 can drop to White H+ without any rhyme or reason but usually around 45% charge or if its a full moon etc...
This is great. Thanks for that data. I'm drawing 2 inferences from your test:
1) CRAW might involve some time overhead for squishing down the files that is non-trivial, which kind of makes sense to me. One of two situations is happening:
a) This compressing happens between the sensor and the cache, so the gating factor is how quickly pictures get into the cache.
b) The compression happens between the cache and the CFexpress card, so the cache must wait for the processor to clear a picture before it can fit another one behind.
You can probably determine the answer to A versus B based on the remaining pictures indicator on the LCD. If that counter acts just like it does in normal RAW, then it'd be A. If it is slow to be reduced and slower to return to normal, it'd be B.
2) CRAW will let you use a smaller hard drive, but it's not going to multiply the number of shots you can get through the cache in a given period of time.
I'll also note that if you'd usd 6400 ISO on the CRAW tests, they'd probably slow significantly to be about the same or lower the number of pictures taken in RAW because the higher ISO's noise would cause the compression to be a bit more time intensive (mostly due to file size, not greater detail).
Everlearning, from the speeds you showed, you must have been using the LCD, not the EVF, no?
EverLearning wrote:
Tiggy, just to be clear on your last post, its not as much on whether you are using EVF vs LCD when shooting but rather after shooting while the buffer is clearing? So if somebody was using the EVF to shoot a substantial sequence of shots and continued to hold it up to their face it would take the longest to clear? Moving the camera away from the face immediately when done shooting would be faster and shooting/clearing buffer purely by LCD would be the fastest?
Yes. In fact two of us tested this, and you can actually temporarily slow the onboarding of files on the card by peeking into the EVF, and then speed it up by moving your head away again. Go ahead and test this yourself. It's not subtle. Because it's just two of us now (discovered first by Tazz93), I'd like to get more corroboration.
On another forum, we've been discussing this, and we think this is an unnoticed foible that is likely a software happenstance, not a physical limit. Although an outside possibility is that it involves a limited power draw issue. We might see this go away in a year in some firmware upgrade.
Good point, Andrew. Maybe I'm underestimating the overhead of that. If this were the case, then changing our refresh rates to something lower might mitigate this? Will test that. -tig
UPDATE:
Hoo boy. OK, Andrew was right. It turns out that if you lower your refresh rate (red menu, submenu 8, "display performance") the cache clears just as fast with EVF as with LCD. That's actually useful to know.
So we likely have a processor-limited ability to clear the cache when you up the electronic viewfinder refresh rate to the smoother setting. It will increase the time it takes to put files from cache to memory card by roughly 450 percent. I was mistaken to think that a future firmware update might help this.
When others were doing card testing were you actively AF during the burst? I had AF engaged the entire burst with points dancing on my subject. I wonder if that makes any difference??
For my part, I was never doing active AF while testing. Just a duck decoy on a shelf. I wasn't testing the camera's abilities, but rather the relative performance of cards, so I had to nail down variables as much as possible. I think it unlikely that this would have an effect, but I've been wrong 2 for 2 with the last couple of crazy theories.
rndman wrote:
Amazon has tempting low price for Transcend 256GB CFexpress currently (about $0.79/GB).
Is that a good price to buy (knowing it's limitations)?
Yes. MemoryC has been selling it for $211 since the beginning so $202 is the best price yet. About equal to what I bought mine for on Amazon Canada before the doubled they price to be more in line with B&H price after forex.
arbitrage wrote:
When others were doing card testing were you actively AF during the burst? I had AF engaged the entire burst with points dancing on my subject. I wonder if that makes any difference??
I was always doing AF, and moving the camera around to makes sure it was doing something.
rndman wrote:
Amazon has tempting low price for Transcend 256GB CFexpress currently (about $0.79/GB).
Is that a good price to buy (knowing it's limitations)?
Well, I will be darned. I was out shooting some ducks this afternoon in the continuous mode and the camera stops and I was having to wait well over a minute for the images to write to my Delkin CFE card! I may be back to square one! And yes, I had my eye on the VF while I was shooting. Got to run some tests tomorrow.
