The Saga of Installing 15.04 on a USB Flash Drive

[SteveRiley]

I'm on trusty and /sys/block/sd*/queue/scheduler shows "noop [deadline] cfq" for all the sd* devices. Do I have to "enable backports" to get this fix? It affects me if I do large copies to a USB backup disc.

It's actually in trusty-updates. What's the output of

apt-cache policy kubuntu-settings-desktop

[NoWorries]

Now Installed on Sandisk Extreme

I am pleased to report that all is now well after installing Vivid on a Sandisk Flash Drive in my USB 3.0 port.

The only problem encountered in the installation process was the previous problem I where I was not able to advance to Disk Setup. I used top to monitor what was happening and apt-get was using about 2% of processing and just staying there.

I abandoned the installation process using the X and restarted it and all proceeded as it should.

I now have Vivid setup, all my favourite packages installed and files copied. I have found Vivid very fast and so decided to do a stop-watch measurement which is not very accurate but gives an idea. The approximate times are:

Vivid power on to login screen 19s
Vivid login screen to Desktop 9s

By comparison my Utopic speeds that I measured are:

Utopic power on to login screen 13s
Utopic login screen to Desktop 12s

I was able to measure read access speeds on my disks using

sudo hdparm -t /dev/hdb
   
sudo hdparm -t /dev/hdc

For the Vivid USB 3.0 disk I got 160 MB/s and for the Utopic SSD disk I got 517 MB/s.

I could only do these measurements on Utopic as the hdparm command does not know devices such as /dev/hdb. These are not in the /dev directory

[SteveRiley]

I could only do these measurements on Utopic as the hdparm command does not know devices such as /dev/hdb. These are not in the /dev directory

See the first paragraph of man hdparm. Not all USB drives can work with this utility.

[NoWorries]

See the first paragraph of man hdparm. Not all USB drives can work with this utility.

Thanks Steve, in my haste I was using the wrong parameters on Vivid and I now get similar speeds using Vivid, ie USB 170 MB/s and SSD 512 MB/s.

I would be interested to know if there is any similar write speed test command as my impression is that the Sandisk USB 3.0 has much better write speed than my previous Datatraveler 111 USB 3.0

[SteveRiley]

I would be interested to know if there is any similar write speed test command as my impression is that the Sandisk USB 3.0 has much better write speed than my previous Datatraveler 111 USB 3.0

I have read in multiple places that the manufacturers of USB drives will use faster memory cells near the beginning of the drive and slower cells near the end. Why? Most benchmarks don't exercise the full capacity of the drive -- they may write only a few megabytes or tens of megabytes. So the drive makers put fast memory at the beginning to make benchmarks look good, then use slow memory to save money.

[SecretCode]

Any recommendations on drive performance test tools we can run ourselves?

[NoWorries]

I have read in multiple places that the manufacturers of USB drives will use faster memory cells near the beginning of the drive and slower cells near the end. Why? Most benchmarks don't exercise the full capacity of the drive -- they may write only a few megabytes or tens of megabytes. So the drive makers put fast memory at the beginning to make benchmarks look good, then use slow memory to save money.

When I was determining what USB 3.0 Flash memory to buy, I relied on a comparison of USB drives that reviewers had conducted. A comprehensive test I found here. I based my decision on these tests as it seems as though they did a thorough job.

[xennex81]

In the past at least for USB 2 it was said to also be a thing of having slow controllers, not necessarily the slow chips. My expensive no-brand (the idiocy...) USB2 drive of 32GB that I am regularly using these days has only a write speed of 7MB/s, and read of 11MB/s. Which is identical to the SD card slot in this laptop.

But USB2 supports up to 35MB/s. Sometimes higher, but few devices make it beyond that number. If USB 3 can suddenly jump it to 170MB/s or whatever, something fishy is going on. I had written a long post before but I did not post it, about the strangeness of seeing these technologies that often seem designed to create a bottleneck for the user that would influence buying decisions, such as with the SMS text message limit of 160 characters, which was obviously (to my eyes at least) designed to bring in more money for the telco's.

My no-brand (it becomes /dev/sdb, or /dev/sdc, depending on what system, ...) USB2 stick works fine with hdparm -t. This SD card reader (which gets a very different device name) also functions with it.

[NoWorries]

I have found that USB drives vary in performance as shown in the report that I referred to for USB 3.0 flash memory drives. I had a number of USB 2.0 flash memories which I tested and the results that I got using hdparm are as follows:

USB 2.0 Lexar 8GB > 31MB/sec
USB 2.0 Toshiba 8GB > 32.8MB/sec
USB 2.0 Sony 16GB > 26.7MB/sec
USB 2.0 Lexar 16GB > 6.5GB/sec

As far as the reasons for USB 3.0 being much faster that USB 2, you can find a wikipedia article here which gives details on how the faster speed is achieved.