-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
NX-1 15FPS?
- Thread starter ROTCWoody
- Start date
In practice it is. Without knowing their exact methodology
Though we know from other testing that there is a delay in the time the camera starts shooting (which causes the processing light to turn on), to the time the camera starts writting to the buffer.
They claim the are getting 31MB RAW files and 12.2MB JPG files in their test. Yet the number of shots varies. Indicating the camera wrote 2.2GB Data. in 30 seconds but also 1.9GB of data in 30seconds which shows wide variations. On those cards they claim the camera is writing at 59MB/s but by their own data it is at least 74.8MB/S If there is a 2 second Delay in the start to writing then that jumps past 80MB/s More so we know that the Processing light is not a write to card light per say. As such the light may stay on after the files are written. (we know images can be accessed before that light stops blinking including being zoomed in on, it might be keeping it from the buffer, but someone would have to go look at the code to find that out) So if it stays on 2 seconds after it finishes writing, now the actual write times are 26 seconds of their test not 30 seconds. This can change numbers by significant margins.
Secondarily they were shooting at a none recommended speed for the NX1. Samsung recommend shooting at 15FPS at 1/500 of a second not 1/60 of a second.
This shows the methodology they are using has flaws. The numbers they indicate don't match the data they are providing. Even from Test to test card to card. A Card that is supposed to write slower takes more shots in RAW+JPG but less in just RAW. There are many inconsistencies. These flaws exist in the 7DII tests too. Where the files being written suggest a higher difference (though by a lower margin) Where they say 59.3 MB/s and the files written suggest closer to 74.8, a 15MB/s difference, the Canon variance is 12MB/s.
I also believe there was a Improvement during a firmware upgrade. That site may very well may be accurate. But In real world application I have heard that the 7DMkII writes a touch slower. It may not be accurate but that site is hard to verify either has there is widely varying information.
The problem with that "study" is that they were recording and counting the flashes from the write LED and assuming that reflected when the camera was doing a write operation (when perhaps all it was doing was flashing the light).
As a result of that assumption their numbers likely don't mean anything.
Also, the camera may be doing things inbetween write operations and not sending a constant stream of data to the card. The dead time between write operations should not be included in data transfer calculations. So the assumption that whatever you measure in terms of the number of files written per second reflects the bus data rate is probably also incorrect.
I think I can shed some light on this as I conducted the test. Your calculation is coming out different because you are using 30 seconds, but that is only is the shooting interval. The buffer continues to clear after the 30 seconds, which is what is used to determine the write interval and write speed. Given how big the buffer is on the NX1 this can be a significant amount of time.
I agree, the access indicator is only as reported by the camera and introduces an unknown. The bigger question is if the write delay is consistent, and how it also is affected by the light remaining on after the camera is done writing. This is something I spent a fair deal of time trying to figure out using different shooting test lengths and running the test multiple times. While it does introduce some margin of error the results are using an average writing 30 second plus buffer clearing time make the effect is actually very small. The difference between cards is also not affected by this if you consider the delay to be consistent between cards or consistent at all. The write speed is an average over the duration of the test. It is not a peak write speed at a given point in actually writing a single file. I actually have some other methods I've experimented with to calculate write speed, but have largely stuck with the current method as it is simple and repeatable if someone wanted to go ahead and try to replicate the test with their own cards. The equipment required and logistics of trying to probe the bus communication directly are not looking like a viable option at this point.
The camera is able to process images faster than clear its buffer (which pretty much is given as the buffer fills at some point), so the shutter speed does not affect the write speed except perhaps for the initial write delay and that only by the shutter speed for the first frame, in other words it is not significant. The the slower shutter speeds affect fps to the sum of the total time, eg. 60 shots at 1/60 take about 1 second longer than 60 shots at 1/500. Having tested the multiple test scenes using multiple shutter speeds, the write speed was nearly same regardless of scene or shutter speed. The range was ~0.5 MB/s different given many tests and shutter speeds up to 1/8000, which was well within the variance you see when testing or benchmarking a given card. SD cards do not operate at a single rate, they can be very inconsistent.
The write speeds are given only for RAW. The RAW+JPEG number of shots on some tests in some cameras are either the results of a specific card with an inconsistent write speed, or an indication of a memory card compatibility problem for a specific card with a specific camera. If you are seeing something different please let me know and I will double check the result.
The firmware of the NX1 tested was the initial one. I have not tested again with newer firmware. I'm open to re-testing the camera, when I looked into it a couple months ago I was not seeing people report that the firmware affected write speed.
I hope my first point about the 30 second interval explains this "innacuracy" you think you are finding.
Last edited:
First I should never engage tecno. I normally have him on my ignore list, but I was already down the rabbit hole in this thread and I made the mistake of opening it.
Second this is true of all your numbers. RAW, JPG, RAW+JPG based on the data provide the NX1 has to write at 15MB/s faster than your test and the 7DmkII 12.5MB/s (based on the fastest SD card). Or inconsistent write speeds indicate that not enough tests are being done to get enough data points to provide accurate conclusions. Which is why I called him out on using it. As all he was doing is what he always does, tries to hate on Samsung.
If you are writing 11MB JPG and 30MB RAW on the 7DMkII and getting 64 files from your test the camera has to write faster than 74.9MB/s for the top Sandisk. It is writing according to your test 64 30MB raw and 11MB Jpgs that means each image is writing two files containing 41MB of data total and doing that 64 times in 30 seconds. That is 2,624MB in 30 seconds (there abouts) That is 87.47MB/s That is a massive difference in write time average. If the files are varying in size drastically enough for that kind of variance, then how are your other tests accurate?
If Samsung has a 2 Second write delay and the Canon has no write delay than the Samsung is actually potentially faster (by your testing). If either camera does something that has the processing light blinking after wards (like deleting the buffer) than that would also impact. Particularly if one camera is multi tasking in between.
If not for the wild variation in data points that are inconsistent with your posted findings. (for example with the Samung just using RAW SanDisk Extreme Pro 74 RAW files, 31MB each. That is 76.44MB/s in order to do that. If the Write time were 59MB/s the average file size would have to 23.92MB or fewer RAW files would have to be take) Than I'd have more faith in the numbers, and would be more incline to ignore the other variable's not taken into account.
At this point though this is off topic and doesn't help the OP at all.
Now if the shot tests are just a shot test, and not actual data about write speeds. Than again why not make that more clear, and why not do more research to make sure you are providing actual results.
When the NX1 first came out several member of this forum had figured that at least 2 second write delay must occur. I seem to recall Samsung during press events saying it could handle in the 70-80 MB/s range. Which if true the processing light is a totally inaccurate measure with that camera, but there is no way to now that based on your testing.
Last edited by a moderator:
Sorry if I did not explain this better, but you can not compare by using 30 seconds. That is only the shooting time. The camera continues to write after 30 seconds. So it is 30 seconds + time after to clear the buffer, which changes depending on the card speed. In the NX1 was closer to 40-45 seconds total for most cards.
Another strange thing about the NX1, RAW+JPEG mode actually created more data, about 49MB total R+J. I don't have the files in front of me, but it was an odd result. Either the RAW or JPEG was larger, vs shooting each one individually.
Another strange thing about the NX1, RAW+JPEG mode actually created more data, about 49MB total R+J. I don't have the files in front of me, but it was an odd result. Either the RAW or JPEG was larger, vs shooting each one individually.
Last edited:
This doesn't matter as you aren't testing the Write speed. you are testing the total time from the point the shutter is pressed to the time buffer is cleared and the camera stops processing.
If the camera has a delay in when it starts clearing the buffer, or if the camera handles the buffer different for different file times, if the camera clears the buffer after each file is written, in batches. Is the Camera Copying and deleting or moving. The actual write speed will certainly effect this, but this doesn't determine the actual write speed.
tecnoworld
Veteran Member
- Messages
- 7,232
- Solutions
- 4
- Reaction score
- 2,123
So I am a troll for reporting a test which has already been reported several times and is considered a reliable source? You should stop offending. I complained about your post and your offense and I hope that some fair moderator will ban you for that. You personally offend people and are very rude.
Keyboard shortcuts
- f
- Forum