http://www.raspberrypi.org/phpBB3/viewt ... =29&t=6180Robert_M wrote:Is there a version of these testing programs that will run directly from the Pi?
Don't see why not, the read performance of the two cards seems fairly comparable - it's the write performance that's a disaster on the Kingston, which suggests it would take a lot longer to populate the media library if you were running something like XBMC. Copying your 1080p movie to the Kingston card will also take a fair bit longer than it would on the San Disk... but playing it back, probably very little difference.ech0 wrote: Do you guys think I can still run 1080p movies from the Kingston card or maybe that's based on the seq result since it's a larger file?
+1ech0 wrote:Yeah, I think it would be great if we could sum this up and try finding the ultimate card for the Raspberry Pi.FlangeMonkey wrote:Can anyone please give me some clarify on which is the main runner in performance?
Yup - started out with my Pi using a no-name brand class 10 micro-SDcard in an adapter in May. Quickly found it was much faster in a USB converter, then went and bought an 8GB USB stick and ran with that for a while.ren41 wrote:Has anyone tried booting and running from SD card, and compared it with running from USB drive after boot? I'm not expecting a definitive answer as to which is quickest but some informed opinions would be good - I can only afford to go down one route and I don't know if USB drives are significantly quicker or if the new SD card drivers that are being written will radically speed things up, to the point where there's not much difference. Has anyone been testing both?
In terms of benchmarks figure this thread (testing on Windows) is missing the point. I have been using f3write/read (as a basic sanity test of the device), hdparm -t gives a fair read guess, bonnie++ tries hard but I find the results inconsistent and quite frankly figure Terje's dd script to be pretty cheap, cheerful & worthwhile.danh wrote:Have we a set of benchmarks on the pi that we can use as a standard way to measure sdcard speed. I am just about to buy some sd cards (probably 2 upto 32G) and i am not too bothered about speed but I will post my results if there is a standard benchmark
Code: Select all
sync; rm -rf testing*; sync; time ( for item in `seq 1 1000`; do dd if=/dev/zero of=testing.$item bs=16k count=10; sync; done; )
Colclusion: The only way to benchmark sd is in the pi itself!
Yes. The most valuable benchmarks are done on the Pi itself. The program to use is called iozone. It's installable on my Ubuntu box using apt-getv via the iozone3 package. I'm not sure if it's available in Debian on the Pi, but I assume it is. iozone can test almost anything. Just cd to a folder that resides on the sd card and run iozone. Here are some example switches that will resemble the CrystalDiskMark tests (50 MB file size, sequential read/write, 512k+4kb random read/write):Robert_M wrote:Is there a version of these testing programs that will run directly from the Pi?
Code: Select all
iozone -e -I -a -s 50M -r 4k -r 512k -i 0 -i 1 -i 2
Code: Select all
Command line used: iozone -e -I -a -s 50M -r 4k -r 512k -i 0 -i 1 -i 2 Output is in Kbytes/sec Time Resolution = 0.000001 seconds. Processor cache size set to 1024 Kbytes. Processor cache line size set to 32 bytes. File stride size set to 17 * record size. random random KB reclen write rewrite read reread read write 51200 4 24685 26740 76735 76388 22259 27553 51200 512 37419 43406 91667 217910 244109 39846