Read and Write on USB and SD

[tinpanalley]

I'm in the midst of doing some research about flash media and benchmarks for different types, I won't get into that because it's not related to this post but it's here if you're interested.

I'm constantly looking at USB 3.0 flash drives, SD cards, and microSD cards and trying to compare different media and formats by analysing read/write and now I'm looking at sequential and random for the first time which I didn't know much about before.

Is there any way at all to know definitively, or at least within a range, what speeds USB2.0, 3.0, and microSD ports are all supposed to be able to operate at? Some site that lists official benchmarks? I feel like if I could know what those numbers are it would go a long way towards helping me buy good media. Yes, I know there are other factors (CPU, device specs, etc) but there must be some kind of standard for these. Otherwise what exactly would we measure the speed of one card vs another with?

Does my question make sense? Simply: If the media themselves offer the variables in terms of what they can do, what are the average read/write speeds expected (sequential and random) for USB2.0, 3.0, SD, and microSD in flash storage. Or can my question not be asked at all?

 

[Cerb]

Is there any way at all to know definitively, or at least within a range, what speeds USB2.0, 3.0, and microSD ports are all supposed to be able to operate at?

Realistically, USB 2.0 will not go over 40MBps, and USB 3.0 about 400MBps. I don't recall non-UHS SD limits, as the bridge chips usually get you first, on USB 2.0.

Some site that lists official benchmarks?

You can dream. Nobody wants that, because even makes of fast SD cards are makers of slow SD cards. Sadly, though, most web reviews are garbage, as well, when it comes to USB and SD. Even StorageReview, a site that should know better, makes it common practice to offer nothing but bullshit advertisement reviews of USB and SD (hint: Iometer can be run on SD and USB device just as well as SATA, should be just as telling, and yes, we do use SD and USB for operating systems and the like).

Userbenchmark.com is actually among the better sources for USB drive data, but no SD stuff.

 

[tinpanalley]

Wow, hate to be a pain but there's a lot of things you mentioned here that I think you assumed I knew about. Which is nice of you to not treat me like a child, but I'm a little lost.

the bridge chips usually get you first, on USB 2.0.

No idea what that means...

Even StorageReview, a site that should know better, makes it common practice to offer nothing but bullshit advertisement reviews of USB and SD

I do understand this, I just wanted to say that I find it completely stupid that that's true.

(hint: Iometer can be run on SD and USB device just as well as SATA, should be just as telling, and yes, we do use SD and USB for operating systems and the like).

Again, sorry totally lost me on what iometer is and what you mean when you say "we do use SD and USB for operating systems". We who? And how so?

It's staggering the degree to which regular consumers must get completely shafted on this storage media because of their massive appetite for cards for their mobile phones, tablets, cameras, etc and their complete lack of looking into the specs. Without a formal poll, I'd say about 90% of regular consumers probably go, "Umm.. yeah, this looks good". Plus, I get the feeling specs are often overblown by manufacturers anyway or worded in ways that don't actually tell you anything about their speed. Or worse, the old "Music CD-R", "Data CD-R" bullshit trick of advertising them as "for your camera" or "for tablets and phones". And that's not even counting the people who get screwed by buying knock-off clone media on Amazon, etc.

So, I guess in the end, since you sound like you have really good knowledge of this, I'm curious what you would do when faced with a bunch of SD/microSD and USB flash drives and want to rate them for r/w benchmarks. You must do SOMEthing, some site, some tool, especially since media specs can't really be trusted. I'm sure you don't know the benchmarks for EVERY card in production. So what do you do?

 

[Cerb]

I do understand this, I just wanted to say that I find it completely stupid that that's true.

most card readers are USB 2.0, unless otherwise specified, and of those, I have never seen one capable of saturating USB 2.0, even with fast cards. USB 3.0 is the way to go, for UHS cards and the like.

Again, sorry totally lost me on what iometer is and what you mean when you say "we do use SD and USB for operating systems". We who? And how so?

Me, for one . I have side-loaded programs and data on SD for my phone, and then the RPi and BBB necessitate SD storage for OS/programs (though the BBB rev C does have 4GB, which allows many uses with just the integrated). USB is good for cost-effective storage for x86 appliances, like firewalls, as well. Even those slow SBCs can quickly bet bottlenecked by such storage.

Iometer is a disk system benchmarking tool, that does a good job of simulating the effects of various real-world workloads on drives. It is commonly used in HDD and SSD benchmarks, but only basic file transfers and such are usually used for USB or SD reviews.

So, I guess in the end, since you sound like you have really good knowledge of this, I'm curious what you would do when faced with a bunch of SD/microSD and USB flash drives and want to rate them for r/w benchmarks. You must do SOMEthing, some site, some tool, especially since media specs can't really be trusted. I'm sure you don't know the benchmarks for EVERY card in production. So what do you do?

Read, mostly :\. Amazon and Newegg user reviews are helpful. FI, right now there are some pretty fast Samsung and Sony cards available. Also, with a few occasional exceptions, I stick to makers of flash (Samsung, Sony, Sandisk, Toshiba--technically Lexar, too, but I haven't had look luck with theirs, performance-wise).

 

[tweakboy]

It might not be configured right for your needs. gl