Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

blackxacto

macrumors 65816
Original poster
Jun 15, 2009
1,157
124
Middle TN
I have a 17,1 iMac / 3.3GHz Intel Core i5 / USB 2.0 hub / USB 3.0 Bus.

I want to buy a 128 usb flash stick for added backups of music, movies, etc. I daily backup the iMac w CCC and a 1TB drive.

I want to match a new USB stick to the iMac. Should the new stick be USB 2.0, 3.0, 3.1, or does it matter with the above iMac ports?
 

redheeler

macrumors G3
Oct 17, 2014
8,419
8,838
Colorado, USA
The iMac USB ports are standard USB-A 3.0. USB 2.0/3.0/3.1 drives should all work fine as long as they are USB-A, but a 3.1 drive will be restricted to the slower speed.
 

blackxacto

macrumors 65816
Original poster
Jun 15, 2009
1,157
124
Middle TN
A USB 3.0 flash drive will give you the best overall performance.

Also consider the Minidrive
Thank you for responding. After thinking about my situation for several days, I decided to buy the 1TB Buffalo Thunderbolt drive. Yes, it's a bit of a step up, but I have a 1TB Buffalo Thunderbolt already for bootable backup. In addition I needed an incremental backup that Time Machine provides, so that will be enough for awhile.
 

blackxacto

macrumors 65816
Original poster
Jun 15, 2009
1,157
124
Middle TN
If you ever decide to get a USB stick, be advised there are large performance differences even among USB 3.0 versions. The fastest I have tested is the 128GB Sandisk Extreme Pro: http://amzn.com/B00HR7FWUC
jorma2, I have a datastickSport 128gb stick, and it is slow as Xmas, but it gets the job done as a data backup, movies & music mostly. How many brands have you tested this year? I appreciate your recommendation for the Sandisk Extreme Pro. Has anyone seen a published speed/reliability test on the web?
 

joema2

macrumors 68000
Sep 3, 2013
1,644
861
jorma2, I have a datastickSport 128gb stick, and it is slow as Xmas, but it gets the job done as a data backup, movies & music mostly. How many brands have you tested this year? I appreciate your recommendation for the Sandisk Extreme Pro. Has anyone seen a published speed/reliability test on the web?

I've tested the 128GB Lexar JumpDrive P20. It is also very fast: http://amzn.com/B00S6DBGKG

There are many speed tests of USB sticks on the web but they are usually out of date, and you're never sure if the I/O pattern they use for their test is representative of the workload you'll be using.
 

blackxacto

macrumors 65816
Original poster
Jun 15, 2009
1,157
124
Middle TN
"I/O pattern they use"

Would you share w me what you mean? I'm just a user, not any kind of programmer.[/QUOTE]
 

joema2

macrumors 68000
Sep 3, 2013
1,644
861
...Would you share w me what you mean? I'm just a user, not any kind of programmer.
[/QUOTE]

Each I/O operation can be in various sizes, can be either read or write, and it can be sequential or random. Each I/O can be issued synchronously, which means the program waits until the first operation finishes before issuing another one. Or the I/O commands can be issued asynchronously in an overlapped fashion. Different devices have varying capability to handle this. If I/O is semi-random there is a locality of reference for how far from the average address the I/O is, which forms a curve, which is affected by caching behavior. All these parameters form a pattern of I/O. Some USB sticks are better than others at large I/O vs small, some are better at sequential vs random I/O, some are better at read vs write. There is not one single benchmark which accurately characterizes I/O performance.
 

blackxacto

macrumors 65816
Original poster
Jun 15, 2009
1,157
124
Middle TN

Each I/O operation can be in various sizes, can be either read or write, and it can be sequential or random. Each I/O can be issued synchronously, which means the program waits until the first operation finishes before issuing another one. Or the I/O commands can be issued asynchronously in an overlapped fashion. Different devices have varying capability to handle this. If I/O is semi-random there is a locality of reference for how far from the average address the I/O is, which forms a curve, which is affected by caching behavior. All these parameters form a pattern of I/O. Some USB sticks are better than others at large I/O vs small, some are better at sequential vs random I/O, some are better at read vs write. There is not one single benchmark which accurately characterizes I/O performance.[/QUOTE]

If there is no way to compare how sticks read and write, what's the point of using it as some sort of measure?
 

joema2

macrumors 68000
Sep 3, 2013
1,644
861
...If there is no way to compare how sticks read and write, what's the point of using it as some sort of measure?

It is like most performance tests -- whether CPUs, GPUs, I/O or automobiles. You can compare different items but any single test may not be revealing. That is why some benchmark packages like QuickBench (available for both Mac and PC) do multiple tests of varying types: http://www.speedtools.com/QuickBench.html

This article discusses various Windows benchmarking tools for USB drives. A Mac would require Boot Camp to run these properly: http://www.pcworld.com/article/2455205/test-the-speed-of-your-usb-drives.html

Besides synthetic benchmarking it's always good to follow up with your own workflow tests. E.g, if your main activity with USB thumb drives is copying a certain folder tree containing many small files, test that -- for both read and write. Or if your main activity is copying a few large video files, then test that, and keep track of the results as you test different drives.
 

blackxacto

macrumors 65816
Original poster
Jun 15, 2009
1,157
124
Middle TN
It is like most performance tests -- whether CPUs, GPUs, I/O or automobiles. You can compare different items but any single test may not be revealing. That is why some benchmark packages like QuickBench (available for both Mac and PC) do multiple tests of varying types: http://www.speedtools.com/QuickBench.html

This article discusses various Windows benchmarking tools for USB drives. A Mac would require Boot Camp to run these properly: http://www.pcworld.com/article/2455205/test-the-speed-of-your-usb-drives.html

Besides synthetic benchmarking it's always good to follow up with your own workflow tests. E.g, if your main activity with USB thumb drives is copying a certain folder tree containing many small files, test that -- for both read and write. Or if your main activity is copying a few large video files, then test that, and keep track of the results as you test different drives.
I appreciate your time. Thanks.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.