p.2 #1 · Sony A9III CFE-a Memory Card Performance Data
dclark wrote:
@Daran@ has provided a partial answer to your comment.
I did not compute the time to clear the buffer or the number of frames in the buffer when clearing is started, I measured it. I record the video from the EVF and then observe the number of frames displayed to determine how many frames are in the buffer when it starts to clear (it was always 236) and then measure the time it takes to clear the buffer by noting the time and frame number when the clearing starts and then moving forward to the frame when the buffer has completed clearing and note the time and frame number (the video analysis is done using Davinci Resolve). Consequently time difference can be measured accurate to 1/30 sec (the frame rate of the video). That time was 11.1 sec for all the cards.
The graph of frame vs time can be used to determine the buffer size by projecting the line back to t=0, as is shown on the chart. The number of frames that have already been written to the card is the difference in frames at the time the buffer fills and the projection back to time zero....Show more →
Thank you for the explanation. There must be something more complicated going on than a simple write buffer for the memory card at play if the shooting rate drops to 5fps after hitting 236 frames since it can write to the card faster than 125MB/sec. Like they're mitigating heat buildup in one of the components. As I mentioned in my other post the A7IV can shoot indefinitely at 6fps writing ~422MB/sec (6*70.3MB) to a sufficiently fast CFe-A card. It seems hard to believe that A9III can only manage 125MB/sec especially since the write speed while shooting changes based on the burst shooting frame rate prior.
p.2 #2 · Sony A9III CFE-a Memory Card Performance Data
Stereodude wrote:
Thank you for the explanation. There must be something more complicated going on than a simple write buffer for the memory card at play if the shooting rate drops to 5fps after hitting 236 frames since it can write to the card faster than 125MB/sec. Like they're mitigating heat buildup in one of the components. As I mentioned in my other post the A7IV can shoot indefinitely at 6fps writing ~422MB/sec (6*70.3MB) to a sufficiently fast CFe-A card. It seems hard to believe that A9III can only manage 125MB/sec especially since the write speed while shooting changes based on the burst shooting frame rate prior....Show more →
There is only one processor. If it's busy responding to interrupts from data coming in, it does not have much time to transfer frames from the buffer to the card. It seems that when the buffer is full the sensor electronics is still trying to send files at the full rate of 120/60fps and consequently occupying the processor with frames that get dropped. That may also be part of the explanation why the frame intervals are so non-uniform.
p.2 #3 · Sony A9III CFE-a Memory Card Performance Data
Stereodude wrote:
Thank you for the explanation. There must be something more complicated going on than a simple write buffer for the memory card at play if the shooting rate drops to 5fps after hitting 236 frames since it can write to the card faster than 125MB/sec. Like they're mitigating heat buildup in one of the components. As I mentioned in my other post the A7IV can shoot indefinitely at 6fps writing ~422MB/sec (6*70.3MB) to a sufficiently fast CFe-A card. It seems hard to believe that A9III can only manage 125MB/sec especially since the write speed while shooting changes based on the burst shooting frame rate prior....Show more →
The complicated part is converting the raw sensor data into a fully-formed image that's ready to write to media. Reading the data off the sensor and dumping it into a frame buffer is only the first step, and not even the most data or computationally-intensive step. That data has to undergo many operations before it becomes a fully-formed raw and jpg. Once those operations are in progress they have to compete with bandwidth and computation demands from ongoing operations in the camera, including new incoming frames. There are ways to tease out how it all balances but it requires running lots of additional burst experments.
p.2 #4 · Sony A9III CFE-a Memory Card Performance Data
dclark wrote:
I did not make measurements for anything other than compressed RAW. I think that photographers that are concerned about performance are shooting in that mode.
That was true for prior Sony cameras, but the A9III is the first camera that doesn't suffer from any performance penalty in any RAW mode, meaning that uncompressed and lossless RAW are also at 120fps 14bit.
The buffer will however be more limited since those files are going to be larger, but it would be interesting to see the impact of changing the RAW format.
p.2 #5 · Sony A9III CFE-a Memory Card Performance Data
randomguy wrote:
Good point, and why would the A9 III require more processing? The new built in raw NR probably.
So type B cards are actually pointless until Sony increases processing power again.
Same for Nikon Z8/Z9, the writing speed range from 600-700MB/s with the best cards.
However they're using type B, but technically they would run at the same speed with type A.
p.2 #6 · Sony A9III CFE-a Memory Card Performance Data
Stereodude wrote:
Thank you for the explanation. There must be something more complicated going on than a simple write buffer for the memory card at play if the shooting rate drops to 5fps after hitting 236 frames since it can write to the card faster than 125MB/sec. Like they're mitigating heat buildup in one of the components. As I mentioned in my other post the A7IV can shoot indefinitely at 6fps writing ~422MB/sec (6*70.3MB) to a sufficiently fast CFe-A card. It seems hard to believe that A9III can only manage 125MB/sec especially since the write speed while shooting changes based on the burst shooting frame rate prior....Show more →
I have an idea that is almost "random" but that at least would explain this phenomenon: if the GS gets very hot at high fps 14bit, then it would require a cooldown period, and the higher the FPS the hotter it would get, which means the longer the cooldown period would have to be, and that would explain why 60+fps require the camera to then operate at 5fps, whereas if you run it at 30fps, then there is no real need for a cooldown at all.
Again, it's just an idea but at least now there's some logical explanation there. Feel free to find other theories.
p.2 #7 · Sony A9III CFE-a Memory Card Performance Data
octo wrote:
Same for Nikon Z8/Z9, the writing speed range from 600-700MB/s with the best cards.
However they're using type B, but technically they would run at the same speed with type A.
Really?!
I thought I read somewhere that with correct settings you get basically infinite buffer on Z9...
p.2 #8 · Sony A9III CFE-a Memory Card Performance Data
j4nu wrote:
Really?!
I thought I read somewhere that with correct settings you get basically infinite buffer on Z9...
If you run in the new HE or HE* modes you can get unlimited. For sure in HE and I think sometimes you can get a slowdown in HE* but it takes some serious effort. But in Lossless Compressed you are buffer limited. Only around 80 shots.
I always shot the Z9/Z8 in HE* as I saw no benefit in Lossless Compressed RAWs.
p.2 #9 · Sony A9III CFE-a Memory Card Performance Data
octo wrote:
Same for Nikon Z8/Z9, the writing speed range from 600-700MB/s with the best cards.
However they're using type B, but technically they would run at the same speed with type A.
I have seen lots of claims about performance of the Nikon Z8/Z9 but I have never seen any data for actual measurements. Can you let me know where I can find such measurements as shown here for the A9III or in the linked post for the A1?
p.2 #10 · Sony A9III CFE-a Memory Card Performance Data
arbitrage wrote:
If you run in the new HE or HE* modes you can get unlimited. For sure in HE and I think sometimes you can get a slowdown in HE* but it takes some serious effort. But in Lossless Compressed you are buffer limited. Only around 80 shots.
I always shot the Z9/Z8 in HE* as I saw no benefit in Lossless Compressed RAWs.
It seems clear that at 20fps the Nikon Z8/Z9 run for a long time in compressed mode. The A9III at 20fps runs for 13.5 seconds and captures 407 frames before slowing to 13fps. As a practical operational limitation (at least the way I operate) that is pretty close to unlimited. Of course that is for 24MP files not 45MP. If we compare that with the A1, which at 20fps runs for 12 sec and captures 242 frames before slowing to 9.8fps, that too is pretty close to unlimited so far as I am concerned. It's also interesting that at 20fps RAW the Nikon Z8/Z9 in lossless compressed and the Sony A1 in uncompressed both get ~80 frames before the buffer fills.
I am surprised that cameras that use cards that claim ~2X the write speed due to having two PCIe-3 lanes instead of one, see in-camera performance that is only marginally better. I would like to see some measurements similar to what we have here for those cameras to see more clearly how long they run at full speed, how much they slow down when the buffer fills, and do they have more than one speed break. How large are the buffers and what are the clearing rates at different capture frame rates, and how fast do the buffers clear once capture stops?
p.2 #11 · Sony A9III CFE-a Memory Card Performance Data
dclark wrote:
I have seen lots of claims about performance of the Nikon Z8/Z9 but I have never seen any data for actual measurements. Can you let me know where I can find such measurements as shown here for the A9III or in the linked post for the A1?
Nikon includes detailed buffer specifications in the manual, including the card used to measure it. Here's the page for the Z8:
The physical buffer size in the camera is 21 frames for all raw types and 24 frames for jpg, which can be determined by half-pressing the shutter on an idle camera and seeing the "rxx" indication on the status display, which shows available buffer capacity. Those values are based on an average image size, which the manual page link I provided above also specifies. Between the physical buffer capacity available via the "rxx" indication and the specs in Nikon's manual, one can calculate the buffer clearing performance without actually having to perform a test.
p.2 #12 · Sony A9III CFE-a Memory Card Performance Data
dclark wrote:
...
It seems clear that at 20fps the Nikon Z8/Z9 run for a long time in compressed mode. The A9III at 20fps runs for 13.5 seconds and captures 407 frames before slowing to 13fps. As a practical operational limitation (at least the way I operate) that is pretty close to unlimited...
That is the 30fps numbers for the A9III you use there, I get about 45-50s in 20 FPS and 900-1000 shots depending on card.
As for practical unlimited, I've heard about golden eagles fights lasting 2min.
p.2 #13 · Sony A9III CFE-a Memory Card Performance Data
Since the Z8 buffer size is the same, 21 Frames, for all RAW types, that seems to indicate the Nikon stores unprocessed RAW data in the buffer. Sony seems to compress the data before storing in the buffer for the lossy compressed and stores unprocessed RAW data for lossless compressed and uncompressed. Nikon Z8 lossless compressed is about the same file size Sony A1 uncompressed and they both hit the buffer at ~80 frames. Sony lossless compressed is smaller and hits the buffer at ~105 frames.
It is not documented by Nikon how fast it captures frames once the buffer is filled, so one would need to make some assumptions to get an estimate. Also there is no data I see for how fast the buffer is cleared once capture stops. It's a small buffer so I assume it should be very quick.
If I had a Z8 I would make the measurements (it's easy), although I think it's buffer limits are probably not operationally important to the way I operate a camera. At 20fps the same is true for the A1 or the A9III. At 30fps, which the Z8 cannot do, the Sony A1 is somewhat limiting and the A9III limit is beyond anything I require. I believe the main operational lesson of this data for the A9III is that I would normally use 120fps or 60fps only in boost. There may be a few situations where operating at 120 or 60 fps in conjunction with the pre-capture may be useful. Others may have different methods and will reach other conclusions, but I believe all the information needed for such decisions is presented.
Anyone using any of these cameras does not have much of a basis to blame the camera if you're not getting the shot. IMO, the weak link is the human attached to the finger that is pushing the shutter button.
p.2 #15 · Sony A9III CFE-a Memory Card Performance Data
dclark wrote:
It is not documented by Nikon how fast it captures frames once the buffer is filled, so one would need to make some assumptions to get an estimate. Also there is no data I see for how fast the buffer is cleared once capture stops. It's a small buffer so I assume it should be very quick.
Here are my rough estimates/calcs based on the specs in Nikon manual. Note HE is infinite per Nikon manual so I just entered a large value instead.
p.2 #16 · Sony A9III CFE-a Memory Card Performance Data
snapsy wrote:
Here are my rough estimates/calcs based on the specs in Nikon manual. Note HE is infinite per Nikon manual so I just entered a large value instead.
The estimated sustained buffer clearing rate would be the values in rows 7 and 8, or some derivative of it.
Thanks, that's quite eye-opening for me .
It looks like the camera can't keep up even with 652 MB/s, which, given it uses CFe-B, is surprising to me.
I'll have to revise me stance on CFe-A vs CFe-B cards I guess, as it seems the only benefit of the latter is price and transfer speed/time to storage...
p.2 #17 · Sony A9III CFE-a Memory Card Performance Data
j4nu wrote:
Thanks, that's quite eye-opening for me .
It looks like the camera can't keep up even with 652 MB/s, which, given it uses CFe-B, is surprising to me.
I'll have to revise me stance on CFe-A vs CFe-B cards I guess, as it seems the only benefit of the latter is price and transfer speed/time to storage...
It's not the CFE interface that's solely limiting HE* throughput to 652 MB/s but a combination of the image generation throughput (ie, ASIC performance in converting raw data to fully-formed images) and CFE interface throughput. You'll note that Lossless yields a higher effective MB/s rate of 764 MB/s. It varies by camera but some are completely ASIC-limited. For example, the original Z6/Z7 which I describe in detail here. More recently in the S5 IIx results I published a few months ago here, which is an interesting test because it was done using the camera's very fast USB-C storge interface rather than internal media.
Based on the estimates I posted above the Z8/Z9 limit appears to be a balance between the throughput and computational demands of image generation and the throughput demands of the back-end media interface that collectively contribute to the overall fps limit. Both place demands on internal buses and memory, which is a finite resource in terms of total bandwidth. Complicating the measurements even more is the fact that SDRAM efficiency drops when you have multiple concurrent consumers of its bandwidth, due to SDRAM's burst-oriented design.
p.2 #18 · Sony A9III CFE-a Memory Card Performance Data
snapsy wrote:
It's not the CFE interface that's solely limiting HE* throughput to 652 MB/s but a combination of the image generation throughput (ie, ASIC performance in converting raw data to fully-formed images) and CFE interface throughput. You'll note that Lossless yields a higher effective MB/s rate of 764 MB/s. It varies by camera but some are completely ASIC-limited. For example, the original Z6/Z7 which I describe in detail here. More recently in the S5 IIx results I published a few months ago here, which is an interesting test because it was done using the camera's very fast USB-C storge interface rather than internal media.
Based on the estimates I posted above the Z8/Z9 limit appears to be a balance between the throughput and computational demands of image generation and the throughput demands of the back-end media interface that collectively contribute to the overall fps limit. Both place demands on internal buses and memory, which is a finite resource in terms of total bandwidth. Complicating the measurements even more is the fact that SDRAM efficiency drops when you have multiple concurrent consumers of its bandwidth, due to SDRAM's burst-oriented design....Show more →
Yes, it's the whole package that counts in the end...
It's just surprising to me that Z8/Z9 does not really reach write speeds that much faster than what CFe-A can do...
p.2 #19 · Sony A9III CFE-a Memory Card Performance Data
j4nu wrote:
Yes, it's the whole package that counts in the end...
It's just surprising to me that Z8/Z9 does not really reach write speeds that much faster than what CFe-A can do...
It's not unlike the limits you might see on a computer doing image processing. It may have an NVMe SSD that can sustain 3 GB/s for long bursts but Lightroom can only generate a fraction of that amount of data when exporting so the utilization of the storage throughout is low.
p.2 #20 · Sony A9III CFE-a Memory Card Performance Data
snapsy wrote:
It's not unlike the limits you might see on a computer doing image processing. It may have an NVMe SSD that can sustain 3 GB/s for long bursts but Lightroom can only generate a fraction of that amount of data when exporting so the utilization of the storage throughout is low.
Sure, but a computer is a lot more general purpose so I can find use for that 3000 MB/s bursts .
I just feel I've fallen for another marketing ploy, as I genuinely thought CFe-B cards are a lot faster in actual usage in-cam...
.
.