SD cards for Pinebook Pro

[IoSer]

information on F2FS and exFAT 

---

[z4v4l]

exFAT FUSE != exFAT normal and efficiently implemented. I was not trying to convince anybody, just noted, you may of course use any esoteric thingy you wish. seeing people trying to use btrfs on SD cards, this F2FS thing doesn't look insane after all.

ps. oh, that's that moronix site, where linux always "wins" in every benchmark. but, surprisingly, even they admitted, that exFAT will be unbeatable (third screenshot).

[IoSer]

exFAT FUSE != exFAT normal and efficiently implemented. I was not trying to convince anybody, just noted, you may of course use any esoteric thingy you wish. seeing people trying to use btrfs on SD cards, this F2FS thing doesn't look insane after all.

ps. oh, that's that moronix site, where linux always "wins" in every benchmark. but, surprisingly, even they admitted, that exFAT will be unbeatable (third screenshot).

[Arglebargle]

exFAT FUSE != exFAT normal and efficiently implemented. I was not trying to convince anybody, just noted, you may of course use any esoteric thingy you wish. seeing people trying to use btrfs on SD cards, this F2FS thing doesn't look insane after all.

ps. oh, that's that moronix site, where linux always "wins" in every benchmark. but, surprisingly, even they admitted, that exFAT will be unbeatable (third screenshot).

There's nothing wrong with btrfs on an SD card at all. You have to be careful not to trip up wrt write amplification but as far as snapshots and transparent compression go it's actually a really good choice.

Keep giving Phoronix shit, any site that just churns out passive benchmarks with zero understanding of what's actually happening deserves all the mockery the internet can throw at it. The data is sort of useful in that it'll show you how default installs of things compare to each other but there's zero understanding of what the benchmarks are actually doing and the conclusions can be really misleading.

If you care about why that's a bad idea read this post about active vs passive benchmarking by Brendan Gregg and take a look at his analysis of a Bonnie++ bench run to see in detail why just stacking up benchmark numbers can give comically wrong information.

This clip is also hilarious: http://www.youtube.com/watch?v=vm1GJMp0QN4#t=17m48s

[jiyong]

exFat can be installed with Package Manager. It's already in the stable repository (as opposed to fonts-noto-color-emojis, which is still in testing).
If we want the Pinebook Pro to be ready for the mainstream, I think exFAT needs to be there by default, especially when the micro SD card slot comes standard.

And it's a shame we need exFAT, as I see it as Microsoft tax.
But it will probably take a long time before we can convince hardware companies and the SD association to go for open solutions.
In the end I really hope Pine64 can accelerate the push for more open source.

[ElektromAn]

So I've considered about the SD card thingy on "device" level

on lifetime

You can use what FS you want. ™
.. unless you avoid any unnecessary writes

Nowadays we have at least "Multi Level Cells" aka MLC which can hold 2 bits in a flash cell.
Newer design uses TLC Tripple Level Cell (3 bits)
QLC Quad Level Cell (4 bits) is currently at the end of design phase, IIRC

I don't know if F2FS and/or ExtFAT knows about the internals of your flash memory.
And I think this will be some kind of intellectual property.
But the most important information is the "erase block"
size which is not equal as the "flash block size"

If you want to set all bits to "1" you must erase the whole
block and this will wear out the flash.
At the beginning we have on to one balance with bit and cell.

Currently we have more, read at the beginning.
And we have some more bits to do our forward error correction.
... with more magic because every cell is *money*
So the magic lies at the vendor/manufacturer of the flash and/or the flash controller.

SD Card vendors optimize their Wear leveling for ExtFat standard.
But unfortunately this is not a native "Linux FS" with uid/gid

And I've not considered to problems we have with two cells as neighbors
This will interfere on quantum level and some cells may leak charge over time.