Card Speed and Video Performance - V30 V60 V90

[chadandreo]

So, from my research, it is commonly accepted that you need to use a V60 or V90 card when recording at 400Mbps internally to a camera, but I was wondering what are the minimum specs required for 100 and 200Mbps 4K All-i? V30?

[kye]

Be careful, there are other variables.

I recently bought a V30 Sandisk Extreme Pro 95MB/s SD card for my GH5 but it wouldn't record 400Mbps (around 50MB/s) video because it is only a UHS-I card, and the GH5 doesn't write fast enough in that mode, but will if you get a UHS-II card (which are frightfully expensive).

In your case though I don't think you need UHS-II, but you should be careful to oversimplify things.

What camera are you buying it for?

[Shirozina]

2 minutes ago, kye said:

Be careful, there are other variables.

I recently bought a V30 Sandisk Extreme Pro 95MB/s SD card for my GH5 but it wouldn't record 400Mbps (around 50MB/s) video because it is only a UHS-I card, and the GH5 doesn't write fast enough in that mode, but will if you get a UHS-II card (which are frightfully expensive).

In your case though I don't think you need UHS-II, but you should be careful to oversimplify things.

What camera are you buying it for?

The GH5 needs V60 or V90 for 400Mbps codec so it's no wonder a V30 didn't work. For the OP a V30 (30MB/s sustained write) card should work for 200Mbps as it's 25MB/s but brands do seem to differ as I have an ADATA V90 card that won't write 400Mbp/s on my GH5 without external formatting. Read the reviews before you buy is my advice and don't go on even the V rating specs alone.

[androidlad]

Surprisingly V30 is enough for 400Mbps on X-T3.

[kye]

56 minutes ago, Shirozina said:

The GH5 needs V60 or V90 for 400Mbps codec so it's no wonder a V30 didn't work. For the OP a V30 (30MB/s sustained write) card should work for 200Mbps as it's 25MB/s but brands do seem to differ as I have an ADATA V90 card that won't write 400Mbp/s on my GH5 without external formatting. Read the reviews before you buy is my advice and don't go on even the V rating specs alone.

I just did a bit of poking around and it seems that V30 is only a reference to 30MB/s performance (V60 being 60MB/s, and V90 being 90MB/s).  So you could have a 59MB/s card that is only rated V30 because it doesn't quite get to V60.

However, all that said, my V30 card wrote 80MB/s continuously on my computer, but couldn't manage more than about 10s of recording on the 400Mbps mode, despite the fact that 400Mbps is 50MB/s and 50MB/s is definitely less than 80MB/s.

What external formatting did you do to your ADATA card that made it work on your GH5?

55 minutes ago, androidlad said:

Surprisingly V30 is enough for 400Mbps on X-T3.

It'll depend on the card..  as I said above, V30 could be up to 59MB/s, which is enough.

I suspect that it might involve data stream variations too.  If the 400Mbps is always 400 then that's one thing but if it goes up to 600 for a couple of seconds because of crazy movement in the scene then it might overrun the buffer and it wouldn't matter that it went down to 200 for the two seconds after that.

[Shirozina]

I use this app

https://www.sdcard.org/downloads/formatter_4/

No idea why it works better than in-camera formating though?

[Kisaha]

41 minutes ago, Shirozina said:

I use this app

https://www.sdcard.org/downloads/formatter_4/

No idea why it works better than in-camera formating though?

Yes, it works better. Do not know why either.

[chadandreo]

2 hours ago, androidlad said:

Surprisingly V30 is enough for 400Mbps on X-T3.

Thats great news since I own one. Which model card are you using?

[androidlad]

56 minutes ago, chadandreo said:

Thats great news since I own one. Which model card are you using?

Sandisk Extreme Pro 100MB/s V30

[thephoenix]

2 hours ago, androidlad said:

Sandisk Extreme Pro 100MB/s V30

micro ?

[thebrothersthre3]

I used sandisk 95MB/s V30 class 10 for 400mbps on the XT3, maybe its the H265 compression vs H264 on the GH5

[kye]

11 hours ago, Shirozina said:

I use this app

https://www.sdcard.org/downloads/formatter_4/

No idea why it works better than in-camera formating though?

Thanks - I'll check that out.

I've just done a bit of googling about speed performance of SD cards, and the short answer is that I can't find a good answer.

The longer answer is that there are a number of things that might be going on here, but I can't verify if they are or not.

There are two candidates that I can think of.  One is the type of file system that is used on the card, and the other is the pattern of where each chunk of data is written to the card (interleaving).

This is interesting:

Quote

Reading and writing to an SD Card and MultiMediaCard is generally done in 512 byte blocks, however, erasing often occurs in much larger blocks. The NAND architecture used by SanDisk and other card vendors currently has Erase Block sizes of (32) or (64) 512 byte blocks, depending on card capacity. In order to re-write a single 512 byte block, all other blocks belonging to the same Erase Block will be simultaneously erased and need to be rewritten.

For example—writing a file to a design using a FAT file system takes three writes/updates of the system area of FAT and one write/update of the data area to complete the file write. First, the directory has to be updated with the new file name. Second, the actual file is written to the data area. Third, the FAT table is updated with the file data location. Finally, the directory is updated with the start location, length, date and time the file was modified. Therefore, when selecting the file size to write into a design, the size should be as large as possible and a multiple of the Erase Block size. This takes advantage of the architecture.

Some designs update the FAT table for every cluster of the data file written. This can slow the write performance, because the FAT table is constantly being erased and re- written. The best approach is to write all the file clusters then update the FAT table once to avoid the performance hit of erasing and re-writing all the blocks within the Erase Block multiple times.

(Source)

This gives you an idea about the complicated housekeeping going on - you don't just write a block of data to an empty slot - you have to update the table of contents as well.  This all references the FAT file system (of which there are multiple versions), and I'm not sure but it might be that there are different housekeeping requirements between different file systems.

The second potential source of performance could be timing and interleave issues.

Interleave is something that happens on physical disk drives.  Imagine you have a disk spinning, and you write a block of data to it that is a quarter turn, and the drive then needs to ask for the next block of data, but by the time that block of data arrives the disk has spun a little bit more.  If you organised the disk so that the location of the next block was straight after the previous block then you would have to wait for the disk to do almost a complete 360 degree rotation before that next block came around again.  To solve this you would format a disk so that the blocks were arranged to minimise that waiting time.
With an SD card there is obviously no physical movement, but there might be timing issues.  For example it might take time to change between different blocks on the disk, or there might be different modes where if you miss a cycle then you have to wait for the next cycle to begin, or whatever.  The above quote indicates that for every chunk of data written it also has to update other areas, and that the largest block size would be useful because it means you write more data for each set of housekeeping, however if every time you wanted to write some data it took a whole cycle then it might be useful to have smaller cycles so that when you do the housekeeping writes they have less time waiting for the cycle to end.

I'm really just guessing with this stuff, but I know from older technologies like physical HDDs and memory timings that these things exist for those technologies and might be at play here.

EDIT:

I should add that I'm not even sure of the terminology at use here.  There are even three ways to format a disk.  IIRC Back in the day there were two, normal and low-level.  The normal one was where you overwrote the whole disk but kept the interleaving the same, and low-level was where you could change the interleaving pattern and it also overwrote the whole disk.  Overwriting the whole disk has the advantage of detecting bad sectors and marking them bad in the index of the disk so they wouldn't be used.  Then along came the quick format which just overwrote the file allocation table but left the data all in-tact on the drive.  Somewhere along the way the quick format turned into a normal format and the low-level format became a mystery.  I think that manufacturers got better at optimising interleave patterns and consumers were getting less technically educated so they just stopped letting people play with them.  I know that now if you choose low-level format in a camera it definitely doesn't ask you for interleave options, so who knows if the camera just leaves it alone, has a one-interleave-fits-all approach, or if it runs a bunch of tests and chooses the fastest one like the old software used to (I very much doubt that!).

[Shirozina]

Interesting and BTW I always use the 'overwrite' option in the formatting app as with the quick one  I get write errors. It does take a long time to do as well so not ideal if you are in the middle of a shoot and need to rotate cards quickly and obviously no good if you don't have a laptop with you. 

[leeys]

@kye, in your case, it's because for UHS-I cards, Panasonic did not implement SDR104 bus speeds in the GH5, which limits UHS-I cards to the slower SD50 implementation which gives, as you can guess, 50 MB/s. As transfer rates are never 100% efficient my guess that even in the high 40s it eventually catches up and stops recording.

More here: https://www.sdcard.org/developers/overview/bus_speed/

[kye]

5 hours ago, leeys said:

@kye, in your case, it's because for UHS-I cards, Panasonic did not implement SDR104 bus speeds in the GH5, which limits UHS-I cards to the slower SD50 implementation which gives, as you can guess, 50 MB/s. As transfer rates are never 100% efficient my guess that even in the high 40s it eventually catches up and stops recording.

More here: https://www.sdcard.org/developers/overview/bus_speed/

I'm actually not that concerned.  Considering that 150Mbps 10-bit is so nice and the UHS-II cards are so expensive, I probably would have ended up with the same card even if I knew about it before buying.  Plus I shoot a lot of footage and so I have large capacity cards as well.