I just tested file system speed on the Pinephone vs the Librem on the builtin eMMC:
PP: Read 8150kB/s Write 2727kB/s
L5: Read 17.1MB/s Write 5708kB/s
Details on https://wiki.mobian-project.org/benchmarks
5 Likes
Minor detail: Change āLibrem5 (Edgewood)ā to āLibrem 5 (Evergreen)ā.
PS: I find it really annoying that the meaning of a MB is being changed from 1,048,576 (220) bytes to 1,000,000 (106) bytes, and Iām never sure which it is. I really think that the marketing genius who came up with the idea that he could sell more hard drives by using 106 instead of 220 should be banished to the deepest bowels of Danteās Inferno for deliberately causing so much confusion.
3 Likes
I can think of a few Big Tech peeps who should consider joining him there.
I really donāt want to be the smartass here (okay, in fact Iām maybe enjyoing it a littleā¦) but it is in fact totally correct that 1 Megabyte (MB) are 1.000.000 (10^6) Bytes. Whereas 1 Mebibyte (MiB) are 2^20 Bytes.
Itās just very unfortunate that computers donāt follow the SI standard. 
5 Likes
Itās quite common for non-physics units not using SI standards. No one measures money in kilodollars (if only ironically). So I donāt se a reason for smartassness in this area really.
I somtimes do write things like 2kā¬.
Oh, one smart ass reply deserves another!
In computers and anything using digital logic, using base 2 for counting makes a lot more sense than using base 10, and at the time when the harddrive industry decided to arbitrarily change the meaning of kilobyte, megabyte, gigabyte, etc., they were overthrowing a convention that had been used for half a century.
Maybe you can argue that the computer pioneers were arrogant to redefine the standard meaning of terms like kilo-, mega-, giga-, tera-, etc., but by the same measure the French government in 1790s that created the metric system and much of what has become the SI standard was arrogant to redefine the meaning of a word like ātonā to mean 1000 kilograms and āweekā to mean 10 days.
Maybe you have an argument that it was wrong to say that a kilobyte equals 1024 bytes, since the word kilo literally means a āthousandā in Greek, but mĆ©gas means āgreatā, gĆgas means āgiantā and teras means āmonsterā in Greek, so you get into problems if you retreat to the original meaning of words.
The fact of the matter is that ākilobyteā, āmegabyteā, āgigabyteā, etc. had an established meaning that was widely accepted when the marketing teams at Western Digital, Seagate, Maxtor, Toshiba, etc. arbitrarily decided to change their meaning, because it made their harddrives appear larger.
Frankly, I donāt care what term is used, as long as its meaning is clear, but currently it is unclear whether āMBā means 106 or 320 bytes whenever I see it being used.
1 Like
Itās all Greek to me, but in all honesty, I only care how many gyrosbites I get.
5 Likes
Maybe you are right on criticising them. But not all metrics are based on the number 2 as powers of two.
And here you mix pure coefficients with a named value and a unit.
Iāve met fio package today for the first time, thanks! Anyway, can you check those two (separate) commands (if you perhaps like them, meaning further modifications to get precise results might be welcomed):
fio --kb_base=1024 --randrepeat=1 --gtod_reduce=1 --fsync=1 --name=test --filename=test --bs=4096k --iodepth=64 --size=128M --readwrite=write:8 --loops=2
fio --kb_base=1024 --randrepeat=1 --gtod_reduce=1 --fsync=1 --name=test --filename=test --bs=4096k --iodepth=64 --size=128M --readwrite=read:8 --loops=2
1 Like
Here you go:
WRITE: bw=62.1MiB/s (65.1MB/s), 62.1MiB/s-62.1MiB/s (65.1MB/s-65.1MB/s), io=248MiB (260MB), run=3993-3993msec
READ: bw=110MiB/s (116MB/s), 110MiB/s-110MiB/s (116MB/s-116MB/s), io=248MiB (260MB), run=2249-2249msec
2 Likes
Thanks! IMHO, those numbers show that the Librem 5 eMMC performance in real life usage is almost two times better (quicker) than running the same software on the PinePhone, as you confirmed already within this thread:
2 Likes
I can only hand out a single heart, so here are some more (for making me laugh out loudly)
2 Likes
and that is āonlyā the disk speed. I guess the much more powerful GPU does its part to more responsiveness too, scrolling that does not feel sluggish, etc. I am truly in love.
Having said all that, there are still serious usability flaws and deficiencies, and we should not forget that we are still a LOOONG way from replacing a recent cheapo Android in terms of functionality. (or just basic usability in some cases)
3 Likes
Itās like when the internet providers market their speed in Megabits/second. Megabytes, Megabits, most non-technical people donāt know the difference. Whoever has the most āMegaāsā must be better (or so they think).
1 Like
if you buy a 1 TB WesternDigital SSD you get a total of 932 GiB (gibiBYTES) so if you have 32 GB RAM total you can partition your SSD to have a 32 GiB swap partition and you are left with EXACTLY 900 GiB for other stuff ā¦ so basically you buy 1TB and you use only 900 GiB 
Guys and girls, I appreciate your interest in standars and units, but as the thread opener, may I politely nudge you towards returning discussing my first impressions of the L5, rather than ISPās marketing tactics (as sleazy as they are). Thank you 
4 Likes
You can get pretty inexpensive docking stations (cradles) on Amazon. I bought one a few years ago for accessing Samsung Dex (Samsungās attempt at convergence). It worked well on my Note 9 except after putting the phone in an Otterbox, it wouldnāt mount to the cradle that way. The L5 may not mount in to a cradle either because like the Otterbox, itās thicker. But I bought some inexpensive USB-C extensions. They work well. Finding a good cradle might be a challenge if the phone is too thick, depending on where the connector is in the thickness of the phone.
MiB is of course unambiguous. So if an underlying value is output both in MiB and in MB then MB is also unambiguous. (In theory therefore MB could have been made unambiguous in isolation by introducing a symbol like MBd, at least as a transitional unit.)
There is ambiguity because not all software has been updated.
In general, it looks to me that Linux is moving (and has been moving) in the direction of embracing the new decimal prefixes - so you might as well learn to love them. 
If you find some old software that outputs MB when it means MiB then you should just fix it.
1000 kilograms is a ātonneā. A ton remains as meaning 2240 lb (1.016 tonne).
For RAM you are definitely correct. Stick with the binary prefixes.
For hard disks it was always somewhat borderline. The underlying unit of storage and/or addressing will definitely be a power of 2 in practice (e.g. 512 byte block / sector or 4096 byte block). Everything above that really may or may not be anything whatsoever to do with powers of two - and disk addressing is generally done by block number.
There was a time, in the dark ages, when hard disks had fixed geometry i.e. cylinder-head-sector, and if the number of cylinders and heads and sectors were all powers of 2 it could make sense to stick with binary - but generally they werenāt all powers of 2.
Then hard disk manufacturers realised that fixed geometry was inherently inefficient and started to produce variable geometry disks, and software be damned. I will claim that it was about this time that disk manufacturers stopped quoting in MiB and started quoting in (decimal) MB.
Network speeds have, to my knowledge, never been quoted using a binary scheme. So if you want to know how long it will take to transfer an X GB file over a Y Gbit/sec link then it makes a whole lot more sense to forget about GiB.
Sorry, @spaetz. No further digressions.
2 Likes