[quote name='elessar123']Class had nothing to do with read speeds, and I'm pretty sure never has. Some manufactures make up stuff. Transcend, for example, have class 10s that write at only 10 MB/s, when the spec is supposed to be something like 25 MB/s on sequential writes. They do it obviously to sell more to unsuspecting costumers. No one actually tests them either, so they can put whatever they want. [/QUOTE]
https://www.sdcard.org/downloads/pls/simplified_specs/part1_410.pdf
See section 3.4 for regards to what Class actually means. The 25MB/s number you're getting is the bus speed, which you can find in the same PDF at section 3.9.6. It's a little more complicated though.
And yes, there are companies out there that will abuse the standards, whether it's due to their inability to make anything better or they just want a cash grab at naive people. My original post still stated that people should always research further. Unfortunately however most people only look at the price and that it's a higher number listed on the box. This is an issue with everything and why we have so many thriving shitty companies producing shitty products and there's no way to change that without somehow mass educating the consumer.
[quote name='elessar123']Saves can actually matter a great deal, where some Class 2s and Class 4s can write small blocks like a hundred times faster than most Class 10s. In something save heavy, like Etrian Odyssey... [/QUOTE]
I'm curious where you're getting "hundred times faster" from. That would be an insane amount, which I'm certain isn't true. And your speed on saving will ultimately depend on the cluster size the card is formatted to and the size of the save data. While 4K Random Write/Read is a good number to get an idea on how slow it can possibly go, it's not a
common real world number that you use
unless your card is formatted in 4K cluster size.
And while I can't really see the inner workings of Etrian Odyssey, one thing I do know is it saves four times. It saves the game data, backs up the game data (To help stave off save corruption maybe?), then saves the map data and finally saves street pass data. If they cut it down to two operations to save, I'd be willing to bet that it would significantly increase the save speed alone regardless of the card you use. Of course I can't be certain as I don't have access to the source code.
I would test this as well if I bought it digitally, but unfortunately I didn't. So the only thing that gets saved to my SD card is the street pass data.
[quote name='elessar123']If you're interested, look up SD benchmarks. It pretty clearly demonstrates how useless how the class system is mostly useless.[/QUOTE]
Most of these benchmarks are from 2011 and don't really include Class 4, but instead Class 6 and higher. Newer versions of the same card have since come out with increased speeds. Often making user images on the purchase page outdated or the benchmarks outdated because they don't get a separate store page with the old one discontinued. (I'm more talking about Amazon here as it's the frequent go-to place for most online people)
For example, the card I linked had increased in speed since 2011 benchmarks as it was 30MB/s then and is 40-45MB/s now. This is mostly a sequential increase, but that's the main focus on SD cards as they are mostly used in cameras and video cameras where you want that. And yes, you won't see a big increase between card classes that only write/read 4K worth of data in random spots. This is why you format things into larger cluster sizes to take advantage of having to random read less and increase times because the same amount of data separated in 4K clusters will take longer than the same data separated out in 32K clusters (Less I/O overhead in larger clusters).
(And yes, I wish we had a different system. It's not totally broken though, unlike the contrast ratio you see on televisions and monitors because there is no standard at all)