>>53472236
ask microsoft or apple

>>53472236
Because 32GB doesn't mean 32 * 10^9 bytes.

You can use 32GB. Windows is miscounting the size.

It's a 32 gigabytes, which is 29.8 gibibytes, and the filesystem formatted on jews the final 200MB. I'm only guessing that's the case, because I thought I read that flash memory always has true power-of-2 capacity (unlike circular platters which are weird).

>>53472364
>I thought I read that flash memory always has true power-of-2 capacity
The chips themselves do, but you NEVER get a perfect chip. A USB stick or SD card will have a flash chip inside that has loads of bad sectors from the factory, and a microcontroller that keeps track of them and maps around them. It holds some good blocks in reserve so it can map them in if some of the good sectors start to go bad. It also does wear leveling so you don't wear out sectors you write to repeatedly.

This is what lets them make these so cheap. A 32 GB device might contain a 256 GB chip that's 75% bad. They just sold you what would be a defective waste chip in any other part of the industry, as a perfectly good and usable drive. They can get nearly 100% yield on their chips if they can take a 99% bad chip and sell it as a shitty 2 GB card.

>>53472236
Flash has a limited number of write/erase cycles before the sectors die and become useless. There are techniques for dealing with the limited erase cycles, but they require some overhead.

Wear leveling: When the FAT filesystem on your PC tells the USB stick to "write a bunch of data to sector number X" the flash drive will actually translate that into some other unused sector number Y. Then the flash drive will maintain a table somewhere that translates Y back into X whenever the PC tries to read from flash. This translation table takes up some space on the chip, so you loose a bit of capacity to it.

The other technique is to have some keep some sectors in reserve for when sectors do start to wear out and fail (similar to how bad blocks are managed in spinning disks).

1gb = 1024 mb
1gb BY producent = 1000mb
iirc of course

Storage companies use a scam naming system. Basically they say 1GB = 1,000 MB...

Ram companies don't do this and they all say 1GB = 1,024 MB

>>53472344
Windows is using the IEC system properly, they just don't use GiB because that would confuse people.

>>53472641
Micron makes SDRAM and NAND chips and they use different units when selling them

>>53472672

They use the same units, different standard.

If AMD/Nvidia/Corsair/HyperX etc all did the same thing as disk companies, there would a big shitstorm.

Your 16 GB of ram? Well, technically it's only 14.9 "GiB" but we're just using a newer standard.

>>53472757
The weirdest thing, is when they sell RAM, they assume 1 GB = 2^30 bytes and when they want to sell storage media, they use 1 GB = 10^9 bytes

1 GB = 1000^3 B
1 GiB = 1024^3 B

>>53472797
It all comes from tradition.

Address spaces are easier to make work if they start and end at even powers of 2, so RAM was traditionally either a power of 2 or a sum of 2-3 powers of 2.

Disks have never made sense in powers of 2. If you could fit 48 512 byte sectors around a disk platter, 1000 rings of that from the center to outer rim, and put 3 platters each with 2 sides like that, you had 512*48*1000*2*3 = 147,456,000 bytes, to use random numbers. It never came out to an even power of 2, and there was no reason to force it to. If they made a disk with those specs, they would have called it 147 million bytes.

Of course these days any quantity of RAM could be mapped equally well so long as it's an even number of 4096 byte pages. And CHS disk geometries are no longer relevant since HDD makers use more complicated layouts and flash doesn't have physical geometries like that at all. Still, the standard remains -- the first company to sell a "14.8 GB" flash drive or a "17.18 GB" stick of RAM would sell very little of it.

>>53472571
Thanks for that anon