My eventual goal is to burn ISO files of distros to USB 3.0 flash drives (as a possible alternative to Virtual Boxes).
On my mainboard, there are two blue ports that SHOULD be USB 3.0, but System Profiler says only one really is. Okayyyyy.
I bought a Verbatim 3.0, 16 GB, plugged it in, checked again with System Profiler and yes, it reads as Version 3.1.
But when I ran Partition Benchmark in the DISKS utility, it yielded the following results:
Sample Size, 10 MiB
Average Read 95.9 MB/sec (100 samples)
Average Write 10.5 MB/sec (100 samples)
Now this seemed slow, so I ran
lsusb -t
and it indeed said 5000M at the line indicating the flash drive.
Is this slower than it should be?
Is this slower than will work running a distro from such a flash drive?
I know y’all love Virtual Boxes, but I am thinking that, with only 8 GB RAM, I can only devote 4 GB and 2 cores to a Box, so this seemed like a better option.
There is a theoretical speed for USB and a practical one. While it may be capable of reaching it’s designated speed in bench marks in the real world it depends upon the speed of the hard drive and other hard ware. It can also depend on the speed of the operating system you are using. With out plugging it in, it will reach the speeds it says it can, but when you plug it in, it possibly won’t do so. Some times it is the reverse as the hardware and OS are quicker that the USB.
So to answer your question. No it is not, because this might just be what your system is cable of doing. This article kinda explains it : https://www.pcworld.com/article/2360306/usb-3-0-speed-real-and-imagined.html
that is the exact same article that i found when looking into usb 3.0 speeds i will add that i have read a few times on reddit that it isn’t a good idea to run a distro from a usb for an extended period of time. i certainly understand launching one live to get a feel for it.
this is my quandary about virtual boxes (and why i don’t use them myself) as well. i was never sure i was allocating enough resources to get the full feel of a distro. that’s why i partitioned off a bit of my main drive (80 gb) and have a triple boot (still considering adding a couple more) so that i can experience the distro in full without limiting resources or dealing with the speed difference of a usb.
It’s been an interesting journey!
First, what a difference a variable can make! I got the following results from other benchmarks:
10 MiB samples, read 95.9 MB/s. Write 10.5 MB/sec using Ubuntu, Disks utility
50 MiB samples, read 118, write 15.8, access time 1.12, using Ubuntu Disks utility
10 MiB samples, read 101.46, write 12.76, using Windows usbbench.exe
50 MiB samples, read 119, write 21.42, using Windows Crystal Disk Mark
Next, I found the following website: https://usb.userbenchmark.com
They report that my Linux numbers are quite lower than their user database for this drive, and that the drive is in the bottom half of all tested flash drives of that size.
My conclusions are that there are better drives I could use instead (I ordered a SanDisk Extreme from NewEgg), and that running a distro from a flash drive would suit me better than Virtual Box (@01101111). just as long as I do not rely on it for too long (maybe 6 months?)
I’d like to extend cordx’ reply by the following information and questions.
What were the presented reasons?
The point of using a live system and/or VirtualBox is to test out functionality and maybe compatibility. Not performance (I will explain, why.). I have a Laptop with 4GB RAM which allows me to easily run VirtualBox appliances to run with 1GB or 512MB RAM. You don’t need more. There is no need to test out an unknown distribution for playing YouTube videos or anything similarly RAM heavy, as this is something generally achievable and/or solvable.
If a live system is loaded into RAM, it actually could be faster than the installed version of that very live system’s underlying distribution.
Now, you won’t be able to “test out” performance as this is something too abstract to test if you try to reach a general conclusion out of it. Let’s say, you try out a system on by using your triple boot method. You install a fresh new distribution and play a YouTube video. Wow, it’s so fast. Great, it doesn’t lag. Your excited and you install the distribution seriously now. You use it for days, weeks, months. The more you use it the more the wowness and greatness depletes. Why? Because in this example scenario you loaded your distribution with additional software frequently which decreases the overall performance of your system. Is your OS “slow”? No, but your personal OS situation is.
The only performance tests making (slight) sense are when tests are done on a fixed environment with one or more fixed applications. That’s how game performance benchmarks are done, btw. Even those slightly valid benchmarks aren’t that useful in the real world, as your computer is probably differing from the benchmark environment, more or less. Usually a lot.
So back again to testing a YouTube video on a freshly installed OS. It will probably run great. Can you rely on this test? Probably not, as you probably will install more programs, use more programs, maybe even run more stuff in the background, all this while a YouTube video might play.
Therefore, testing performance before seriously installing and using a foreign OS isn’t a concern and can’t be a serious one, from an evaluation perspective.
a quick search didn’t bring up the threads i was thinking about, but from what i recall the idea was that the usb isn’t designed to be written to and read from as consistently as either an hdd or ssd (or hybrid for that matter) so it would shorten the lifespan of the usb considerably. because i wasn’t inclined to use this method, i didn’t read much beyond the first few comments. in my mind i linked this to what i believe was one of your comments here (also unsure of the thread) about how (especially the cheaper) usb’s are generally the chips that aren’t of high enough quality for ssd drives proper. or something to that effect.
i had considered this when writing my reply, but since i don’t run many live systems anymore i wasn’t sure if all/most/or just some distros had this option.
I have seen exactly what @01101111 cites about a limitation on the number of write operations on solid state devices, and have had flash drives fail as a result (some of them were MP3 players). But, when I read up on these, the consensus was that we’re talking about millions of writes, not hundreds. Which I would say dates the lifespan of a USB Flash containing a full distro to a year or two, perhaps more.
When I consider looking into a distro, it is to judge if I want a divorce from Ubuntu GNome and get married to another. So I am with @01101111 in actually doing the things I need a computer for, and not just looking at the cool stuff it can do. Assessing it means actually running the programs I need, in full resource mode. That is why the USB Flash option seemed like a route I could take.
i don’t think any of the above rules out trying the usb option. one thought might be to try your present system on a usb just to see how it fares so you can tell what the relative difference would be with other distros. like Akito said, if you use a new iso or install (especially directly to ram) the performance would probably be snappier than a lived-in daily use system but it might give you an idea.
SSDs are nowadays so cheap (relatively speaking) that I would go for an SSD instead of a USB stick. Here in EU-countries 120 GB cost no more than equivalent US$ 30.-
This size of SSD could accommodate multiple distros , provided one does not want to store hours-long videos.
i will add that i have read a few times on reddit that it isn’t a good idea to run a distro from a usb for an extended period of time. i certainly understand launching one live to get a feel for it.