[0x4243]

Hi everybody,

yes, I know there are already a few threads about UHS II in the K-3 III. But these threads did not answer my question about the real speed of a UHS II card in the camera.

I did some tests and came to the conclusion, that my K-3 III with my SanDisk Extreme Pro UHS II 64GB 300MB/s V90 writes only 50-70 MB/s to the card.

How did I test? I recorded the shutter sound with audacity so I could make relatively precise measurements.

I first burst shoot until the buffer was full, waited until the LED went out and then pressed the shutter once again to have a acoustic mark in the recording.
The burst was 28 pictures which was in sum 932MB of data (DNG only). The LED went out after a total of 19 seconds (beginning to count with the first shutter click). Thats 932MB/19s = 49MB/s.

Clearing the full buffer takes about 15-16 seconds.


Another test: After the burst, when the shutter slowed down I kept on shooting until 30 seconds. I counted the images during these 30 seconds and added the time to clear the buffer (LED stays out). The number of images multiplied with the size (roughly) diveded by the time givs me 56MB/s.

Another test: After the buffer is full and the shutter slowed down, I took a look at the recorded shutter sound and counted the number of pictures during 20 seconds. It was 34 images in 20 seconds. With roughly 35MB per image thats 34*35 / 20 = 59 MB/s.

So, no matter how I count, the result is always way from the UHS II speed of 300MB/s (even much lower than UHS I with 104MB/s).


The card in the PC (with SanDisk Extreme Pro USB C UHS II card reader) on the other side writes between 220 and 250MB/s and reads between 270 and 300MB/s.


camera settings: manual focus, 1/4000, F5.6, DNG, card in slot 1 (the one on the back side of the camera), High continuous shooting


What is your experience with UHS II cards in that camera. Is the speed I get normal or is there something wrong?

BTW: A 170MB/s UHS I card is a bit slower, here I get 40MB/s with a clearing time of 21 seconds. But thats still too low for UHS I

I am curious.....
Christian

[sbc]

I am guessing here but maybe the time includes processing of the RAW files? I believe noise reduction is baked in?

[0x4243]

I am guessing here but maybe the time includes processing of the RAW files? I believe noise reduction is baked in?

Well, thats a good point. Indeed, I had the automatic noise reduction switched on. After switching off, I feel like getting some more performance (31 images in the first burst instead of 28). But thats minimal. Still get less than 60MB/s transferred to the card.

So I wonder why UHS II and expensive cards.


there must be something wrong...

[sbc]

Have you tried repeating the test using slot 2 (UHS1)? If you still get the same speed, it means there is a lot of overhead in preparing the RAW files in the camera and UHSII bus speed is not fully utilised. If it is slower still in slot 2 then there is a possibility of a flaw in the calculation.

[0x4243]

Have you tried repeating the test using slot 2 (UHS1)? If you still get the same speed, it means there is a lot of overhead in preparing the RAW files in the camera and UHSII bus speed is not fully utilised. If it is slower still in slot 2 then there is a possibility of a flaw in the calculation.

With the UHS II card in the UHS I slot I get even a bit more images during 30 seconds. The first burst until the buffer fills is the same but after the buffer is full the rate is a bit higher (but not as constant) as in the Slot 1 (UHS II). At a total its about 65MB/s which I'd say it's ok for UHS I.

Here you can see my measurements. Same settings, same card. upper one is in Slot 1(UHS II, the fast one....), lower one is in Slot 2.

Maybe anyone could repeat this to proof my camera has a problem...?

[Kunzite]

Here you can see my measurements. Same settings, same card. upper one is in Slot 1(UHS II, the fast one....), lower one is in Slot 2.

Just to be sure, you didn't mix the slots? Slot 1 (UHS-II) is the one close the back of the camera, and Slot 2 is the one in the front. It's the other way than in older models.

[0x4243]

Just to be sure, you didn't mix the slots? Slot 1 (UHS-II) is the one close the back of the camera, and Slot 2 is the one in the front. It's the other way than in older models.

Yes, I know. Its different to the K-1 and I triple checked the numbering.

I already took a look into the slots with a flashlight to see that there are really more contacts on the slot at the back of the camera...

[stevebrot]

I first burst shoot until the buffer was full, waited until the LED went out and then pressed the shutter once again to have a acoustic mark in the recording.
The burst was 28 pictures which was in sum 932MB of data (DNG only). The LED went out after a total of 19 seconds (beginning to count with the first shutter click). Thats 932MB/19s = 49MB/s.

Clearing the full buffer takes about 15-16 seconds.

This would be the time to clear the buffer when full - (minus) one shot going out + (plus) one shot incoming.

Is the speed I get normal or is there something wrong?

Is there any chance you angered an evil witch? You may well have been cursed.

More seriously, unless you can put a monitor on the write buffer for the card reader, you will never really know whether the camera is fully utilizing the card's write speed. (There is likely more than one buffer in the process stream.) Historically, tests similar to yours have resulted in throughput lower than the card's rating. The larger question might be whether it is faster than slot 2.


Steve

[0x4243]

This would be the time to clear the buffer when full - (minus) one shot going out + (plus) one shot incoming.



Is there any chance you angered an evil witch? You may well have been cursed.


Steve

I think the theory with the evil witch is the most likely so far.

I guess I'll do some more tests and send them to ricoh support. Maybe they can tell me if this is ok.

I agree that the data rate can not always be reached because of internal processing, etc. But the question then is, why UHS II, why ultra fast expensive cards if the camera only uses less than 100MB/s.
And yes, it's way faster then my K-1. I already was able to get shots that I would have missed otherwise. So I'm still impressed by the K-3 III and I like that camera. But I'd like to know if this is normal or not...

So, if it's only a marketing gag, it worked. At least for me....

[i_trax]

friend sent me this :

Blackmagic

Disk

Speed

​​

Test

If you own a Mac and want to verify your

SDcard

,

BlackmagicDiskSpeed

​​

Test

is a useful tool for you. This free utility is designed to test the speed of larger drives. However,

BlackmagicDiskSpeed

​​

Test

will also work with

SD

and microSD

cards

.


Not sure if this will be useful here?

---------- Post added 06-09-21 at 10:56 ----------

why my last post looks so odd?
was the "Global Formatting" changed ??

[angerdan]

Not sure if this will be useful here?

No, because you need to measure the transfer speed between camera and memory card.

[Kobie]

No, because you need to measure the transfer speed between camera and memory card.

Exactly. If the bus/pipeline speed isn't there, you can write real fast to the card, but it'll take time for the readout before that happens. Nothing to do with the speed of the card or the card slot. It the speed of the path between the camera and card is limited, well, then you get a limited amount of speed. Without measuring the actual pipeline speed and actual card speed, there's too many pieces in the way to really know actual speeds per section being used.

[ulrich]

I guess I'll do some more tests and send them to ricoh support. Maybe they can tell me if this is ok.

I agree that the data rate can not always be reached because of internal processing, etc. But the question then is, why UHS II, why ultra fast expensive cards if the camera only uses less than 100MB/s.

Unfortunately I cannot test - I do not yet have a UHS II card - but the answer would be very interesting! If - in practice(!) - there is no big difference between UHS I and UHS II, this could save some money.

Nobody else here, who can test this?

[Adam]

I'll be performing some tests next week, but even if the actual write speed is around 60-65 Mb/s, that's still up to 20 Mb/s faster than the K-3 II. It'll be interesting to see if the same speeds can be achieved with UHS-I cards in the first slot; I'm thinking probably not.

[kwb]

Hi Christian, your numbers are worse than the tests a Japanese user posted on youtube, what was your ISO? Pentax always says that it depends on ISO.

SanDisk Extreme PRO SDXC II 64GB, R:300MB/sec, Class 10/U3, probably the same card as yours?

You Tube

You Tube

PROGRADE COBALT II 128GB, R:300MB/sec, W:250MB/sec, Class 10/U3/V90

You Tube

You Tube

In the latter, which was somewhat better performing card of the two, the camera slowed down after 40 frames in 3.5 sec, he stopped after 41th frame. From the first shutter to the point where the LED stopped blinking, it took ~12 sec. We don't know the file size, but that's 41 frames in 12 sec.


Interesting caveat (which he notes in his blog PENTAX K-3 Mark III ??????????????? – ???????) is that PROGRADE performed poorly in his initial test (39 frames in 4 sec, 40 frames in ~17sec from start to finish). After publishing the result, PROGRADE contacted him suspecting that the card was working in UHS-I mode, and offered him a replacement, which he took. The video above was after the replacement.