Home · Register · Join Upload & Sell

Moderated by: Fred Miranda
Username  

  New fredmiranda.com Mobile Site
  New Feature: SMS Notification alert
  New Feature: Buy & Sell Watchlist
  

FM Forums | Sony Forum | Join Upload & Sell

1
       2       end
  

Archive 2021 · A1 Buffer Clearing Rates

  
 
dclark
Offline
• • • • •
Upload & Sell: On
p.1 #1 · p.1 #1 · A1 Buffer Clearing Rates


It was raining a couple of days ago, so I spent some time doing some measurements on the A1.

In a prior thread I showed some data on the buffer clearing rates for the A9 using a UHS-II SD card. I used the same method for the A1 to measure the buffer filling and clearing rates using a CFE-A card and the UHS-II card. The method is to record the HDMI output of the camera, which includes all the information displayed on the EVF and LCD. That HDMI video file is then examined frame-by-frame using DaVinci Resolve to extract data on how quickly the buffer fills and is cleared and how many files are captured. The frame number gives accurate timing information. It's pretty tedious but yields that most accurate data I can get.

One of the surprises for the A9 (at least I was surprised) was that the the rates vary a lot as the camera heats up. One objective was to see if the better thermal management in the A1 improved that situation.

Shown below are three plots of data obtained for the A1 with CFE-A and UHS-II cards. The CFE-A was a Sony 160GB card rated at 700MB/sec write. The UHS-II was a Sony G that is rated at 299 MB/sec write.

This data is for RAW compressed files written at Hi+ 30fps. The camera was in release priority and the lens was set to manual focus.

The data shown below was taken in a random order so that if there was a heating effect the data would not form smooth lines. You can see that any temperature effect is negligible, which is consistent with the A1's greatly improved thermal management.

The first chart shows that the A1 stores files at the full 30fps until the buffer fills. I found that the rate was always about 29-30 fps. Once the buffer fills, the frame rate drops to the rate frames can be cleared from the buffer. For the CFE-A card that is about 11 fps and for the UHS-II card it is about 4.7 fps. The chart shows an abrupt break in the slope for the SD card at 150 frames. Up to 150 frames the camera is running at 30 fps and then abruptly reduces to 4.7 frames/sec. For the CFE-A the transition is not as large so the transition appears to be more gradual.

The second chart shows how many frames need to be captured at 30 fps to fill the buffer and it shows that the buffer holds about 127-128 frames. That means the buffer is about 6.7GB. You see that that the buffer fills when 220 frames have been captured using the CFE-A. That means it takes 7.3 seconds to fill the buffer at 30fps, and (220-127)=93 frames have already been written to the CFE card by the time the buffer is filled. For the SD card, it takes 5 sec and a total of 150 frames to fill the buffer and (150-127)=23 frames have been written to the SD card by the time the buffer fills. 220 is short of Sony's spec of 238. I am not sure why, but it is true that the data from analysis of the HDMI output is dependent on the camera quickly updating the displayed information.

The time to clear the full buffer is very consistent, and is 11.5 sec for the CFE card and 27 sec for the SD card, a factor of 2.35.

The third chart shows the measured write speed. There is some variability of the data for short bursts, which may be due to inaccuracy of the displayed data in the HDMI stream. For short bursts even small errors can create large variations in the measured write speed. I used only the larger bursts to measure an average of 587 MB/sec for the CFE-A and 246 MB/sec for the SD card. That means they are both running at about 82-84% of their rated spec.

Operationally these results can be summarized as follows:
(a) There is no noticeable thermal effect in the A1.
(b) The CFE-A captured about 220 frames before the rate drops from 30fps to about 11fps.
(c) The SD card captured about 150 frames before the rate drops from 30fps to less than 5fps.
(d) The CFE-A clears the full buffer in 11.5 sec.
(e) The SD card clears the full buffer in 27 sec.

If it rains again I may make some measurements for lossless compressed and uncompressed files.

Dave






Total Frames Captured vs Capture Time. Both cards capture at 30FPS until the buffer fills, then slow to the rate at with frames are written to the storage card.







This Chart Shows how many frames are captured before the buffer fills. It also shows how many frames have be written to the card by the time the buffer fills.







Write Rate for the CFE and UHS-II Cards



Edited on Mar 18, 2021 at 12:35 PM · View previous versions



Mar 17, 2021 at 11:51 PM
robert614
Offline
• • •
Upload & Sell: On
p.1 #2 · p.1 #2 · A1 Buffer Clearing Rates


Nice work Dave!

Very informative.

I appreciate all the work you put into it.

Thank you very much!

Robert



Mar 18, 2021 at 12:49 AM
Douglas L
Offline
• • • • •
Upload & Sell: On
p.1 #3 · p.1 #3 · A1 Buffer Clearing Rates


Thank you for the hard work, Dave! Very informative.


Mar 18, 2021 at 04:44 AM
nobody23
Offline
• •
Upload & Sell: Off
p.1 #4 · p.1 #4 · A1 Buffer Clearing Rates


Loseless is the most important one....
Looking forward to a rainy day at your place!



Mar 18, 2021 at 04:58 AM
ilkka_nissila
Offline
• • • •
Upload & Sell: Off
p.1 #5 · p.1 #5 · A1 Buffer Clearing Rates


Hmm. You seem to have changed the labeling between the different charts. For clarity it would be good to use the same labeling and mark the labels in each figure.


Mar 18, 2021 at 04:58 AM
arbitrage
Offline
• • • • • • •
Upload & Sell: On
p.1 #6 · p.1 #6 · A1 Buffer Clearing Rates


Excellent work Dave. Thanks so much for doing this stuff.




Mar 18, 2021 at 06:28 AM
Jemini
Offline
• • •
Upload & Sell: On
p.1 #7 · p.1 #7 · A1 Buffer Clearing Rates


Great work. I was wondering about these numbers especially for the SD card. One thing I noticed with A9, if the camera is writing to both cards simultaneously, it take double the time. If that's true for A1, total time might be around 24 seconds for CFE-A and 55 seconds for UHS-II.

Also curious about the numbers for RAW (lossless)+JPG

Thank u for the effort

dclark wrote:
It was raining a couple of days ago, so I spent some time doing some measurements on the A1.

In a prior thread I showed some data on the buffer clearing rates for the A9 using a UHS-II SD card. I used the same method for the A1 to measure the buffer filling and clearing rates using a CFE-A card and the UHS-II card. The method is to record the HDMI output of the camera, which includes all the information displayed on the EVF and LCD. That HDMI video file is then examined frame-by-frame using DaVinci Resolve to extract data on
...Show more



Mar 18, 2021 at 06:46 AM
arbitrage
Offline
• • • • • • •
Upload & Sell: On
p.1 #8 · p.1 #8 · A1 Buffer Clearing Rates


Not if you want 30 FPS

nobody23 wrote:
Loseless is the most important one....
Looking forward to a rainy day at your place!




Mar 18, 2021 at 07:48 AM
schlotz
Offline
• • • • •
Upload & Sell: On
p.1 #9 · p.1 #9 · A1 Buffer Clearing Rates


Very informative Dave, thanks for all your efforts


Mar 18, 2021 at 08:03 AM
nobody23
Offline
• •
Upload & Sell: Off
p.1 #10 · p.1 #10 · A1 Buffer Clearing Rates


arbitrage wrote:
Not if you want 30 FPS



Then you go CF-E-A anyway.



Mar 18, 2021 at 08:27 AM
arbitrage
Offline
• • • • • • •
Upload & Sell: On
p.1 #11 · p.1 #11 · A1 Buffer Clearing Rates


nobody23 wrote:
Then you go CF-E-A anyway.


And yet you still won't get 30FPS if you set it to Lossless...no matter what memory card you stick in there



Mar 18, 2021 at 08:36 AM
Vetteman
Offline
• •
Upload & Sell: On
p.1 #12 · p.1 #12 · A1 Buffer Clearing Rates


Nice job - great info. Thanks Dave!


Mar 18, 2021 at 10:14 AM
dclark
Offline
• • • • •
Upload & Sell: On
p.1 #13 · p.1 #13 · A1 Buffer Clearing Rates


ilkka_nissila wrote:
Hmm. You seem to have changed the labeling between the different charts. For clarity it would be good to use the same labeling and mark the labels in each figure.


That's embarrassing!
Fixed.



Mar 18, 2021 at 10:24 AM
louie champan
Offline
• • • • • •
Upload & Sell: On
p.1 #14 · p.1 #14 · A1 Buffer Clearing Rates


Very informative and most helpful Dave, thanks for your efforts.


Mar 18, 2021 at 11:30 AM
dclark
Offline
• • • • •
Upload & Sell: On
p.1 #15 · p.1 #15 · A1 Buffer Clearing Rates


Three additional bits of information.

1. I forgot to mention that the buffer indicator that can be displayed on the left side of the EVF (Menu: Setup > Display Option > Remain Shoot Display) is not accurate at all. When shooting it shows the buffer is full at 150 frames when it has quite a bit of space left if shooting with a CFE-A. When clearing the buffer, it shows the buffer cleared when there are still 50 frames to go. Its a minor issue, but don't depend on that display for accurate information.

2. These charts and data points depend on file size and frame rate. The average file size for Sony's compressed files is pretty stable but does vary a bit depending in how much compression the embedded JPEG images have. The same should be true for uncompressed. Lossless compressed file size may vary quite a bit. Shooting JPEG or including JPEG files with also add variability in the file size.

3. This data can be scaled for file size and frame rate and should be a pretty good predictor of the results for other file formats. The Buffer Clearing Rate chart should not change, except for the possibility that it will be a little bit faster to write large files than smaller files.

Scaling the first chart is based on computing the rate at which the buffer is filling. The formula is:

BFR = FR*FS - BCR

Where, BFR is Buffer Filling Rate, FR is Frame Rate, FS is File Size, BCR is Buffer Clearing Rate.
In the case presented, FR=30/sec, FS=54MB, BCR = 587 MB/sec for CFE and 246MB/sec for UHS-II.
You also need the size of the Buffer, BS, which is 127*FS = 127*54MB

Time to fill the Buffer = BS / BFR.

Once the buffer is filled, then BFR = 0, so the FR drops to FR = BCR / FS.

Note that the BFR can go to zero or negative in the formula above. That means there is a FR*FS below which the buffer clears as fast as it is being filled and the buffer will never fill. For example, with the CFE-A, Compressed RAW, at 10fps, the buffer never fills. You can shoot continuously until the card fills.

Since the Buffer Size and the Buffer Clearing Rate do not change with File Size or Frame Rate, the time to clear the buffer does not change, it is 11.5 sec or 27 sec regardless of how you managed to fill the buffer.

This gives you what you need to predict the results for different file sizes and frame rates.


Another point of caution is be careful to include the proper number of factors of 1.024 in file sizes and specs. A kB is 1024 bytes. A MB is 1024*1024 bytes. A GB is 1024*1024*1024 bytes. Lots of errors on internet postings can be traced to this error.

Dave



Mar 18, 2021 at 09:53 PM
Holger
Offline
• • • •
Upload & Sell: Off
p.1 #16 · p.1 #16 · A1 Buffer Clearing Rates


dclark wrote:
It was raining a couple of days ago, so I spent some time doing some measurements on the A1.

In a prior thread I showed some data on the buffer clearing rates for the A9 using a UHS-II SD card. I used the same method for the A1 to measure the buffer filling and clearing rates using a CFE-A card and the UHS-II card. The method is to record the HDMI output of the camera, which includes all the information displayed on the EVF and LCD. That HDMI video file is then examined frame-by-frame using DaVinci Resolve to extract data on
...Show more
Nice.
When looking at the first graph, I get a final rate of 11fps. Initially, after the rate drops you still seem to get about 14-15 fps from 5 seconds to about 12 seconds. At least that is what I get when fitting and doing change point analysis.




Mar 19, 2021 at 04:20 AM
QuietOC
Offline
• • •
Upload & Sell: Off
p.1 #17 · p.1 #17 · A1 Buffer Clearing Rates



dclark wrote:
Another point of caution is be careful to include the proper number of factors of 1.024 in file sizes and specs. A kB is 1024 bytes. A MB is 1024*1024 bytes. A GB is 1024*1024*1024 bytes. Lots of errors on internet postings can be traced to this error.


FWIW: Using kilo for 1024 was a convention in the computer industry that didn't match other standards and has been addressed by creating new prefixes for those binary units. A kilobyte is 1,000 bytes. A kibibyte is 1,024 bytes. The abbreviation for kibibyte is KiB.



Mar 19, 2021 at 04:26 AM
snapsy
Offline
• • • • • •
Upload & Sell: On
p.1 #18 · p.1 #18 · A1 Buffer Clearing Rates


dclark wrote:
It was raining a couple of days ago, so I spent some time doing some measurements on the A1.

Nice work Dave! Just an FYI, you can automate the process of calculating buffer depths, clearing times, and card throughput by utilizing the sub-second timestamp within the EXIF of images. For example, in exiftool you can specify the -SubSecDateTimeOriginal tag to extract only that field, along with -q to suppress the banner. You can run it against a directory of files and ingest the output into a spreadsheet to automatically calculate the intershot times, from which you can calculate the buffer depths and clearing times.

Here is an analysis I did of XQD vs CFE on the Nikon Z cameras which used that technique. That conclusion post also describes the process of how to determine where the bottleneck is, between the image processor (Expeed on Nikon, BIonz Z on Sony) and the card throughput. One technique I used was to use alternate file sizes, in particular TIFF output which the Z's support - you can alter the processing time vs output size ratio to determine if the bottleneck (ie, a TIFF is much larger but shares almost the same imaging pipeline, so if the total throughout of TIFFs is higher (sustained frame rate x average file size) than raw/jpg then that indicates the bottleneck is in the image processor. You can view my post here on automating the collection of file sizes for the total throughput calculations.



Mar 19, 2021 at 11:02 AM
dclark
Offline
• • • • •
Upload & Sell: On
p.1 #19 · p.1 #19 · A1 Buffer Clearing Rates


Holger wrote:
Nice.
When looking at the first graph, I get a final rate of 11fps. Initially, after the rate drops you still seem to get about 14-15 fps from 5 seconds to about 12 seconds. At least that is what I get when fitting and doing change point analysis.



I agree that the CFE curve does not have an abrupt change in slope like the UHS-II does. The change in slope should be at ~220/30=7.3 sec, not at 5 sec like it does for the UHS-II, The CFE curve seems to roll off rather than abruptly change slope. I don't know why. You can see a bit of the same in the second chart.



Mar 19, 2021 at 05:31 PM
dclark
Offline
• • • • •
Upload & Sell: On
p.1 #20 · p.1 #20 · A1 Buffer Clearing Rates


QuietOC wrote:
FWIW: Using kilo for 1024 was a convention in the computer industry that didn't match other standards and has been addressed by creating new prefixes for those binary units. A kilobyte is 1,000 bytes. A kibibyte is 1,024 bytes. The abbreviation for kibibyte is KiB.


Windows 10 shows file sizes in KB, MB and GB. They are all in terms of 1024 factors. KiB is not used. You can see the actual number of bytes by using "Properties", and then it gives the number of bytes. It might be even more confusing if some were given in terms of factors of 1000 and others in terms of factors of 1024, but if distinct symbols were used that would alert people to the fact there is a difference.

In this case I stated the write rate in MB/sec, where MB means 1024*1024 bytes. The spec on the CFE and UHS-II cards are given in MB/sec and I assume they also are given in terms of factors of 1024.


Edited on Mar 19, 2021 at 05:57 PM · View previous versions



Mar 19, 2021 at 05:53 PM
1
       2       end




FM Forums | Sony Forum | Join Upload & Sell

1
       2       end
    
 

You are not logged in. Login or Register

Username       Or Reset password



This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.