recommended micro sd card ?

[jl_678]

There is a thread that talks about performance testing. I believe that it talked about using iozone3 to test performance. As @tlim mentioned, the evo+ is a good one and you can get a 64GB card from Best Buy for under $20.

Sent from my XT1254 using Tapatalk

[Bob123456789]

Yes I realize that.

I have already purchased two cards from Pine. I don't see any testing of the Pine cards yet that allows me to compare the Pine cards to the others.

[xalius]

Hi I just tested my sdcard that came with the Pine (16GB) and a Samsung EVO 64GB using a fresh Ubuntu base image, the lastest longsleep kernel (3.10.102-0-pine64-longsleep) , cpufreq governor set to performance and no modifications to the default ext4 fs options:  iozone -e -I -a -s 100M -r 4k -r 16k -r 512k -r 1024k -r 16384k -i 0 -i 1 -i 2

Pine64 Card:

Code:

       Iozone: Performance Test of File I/O
               Version $Revision: 3.429 $
               Compiled for 64 bit mode.
               Build: linux

       Run began: Sat Jun 18 10:57:35 2016

       Include fsync in write timing
       O_DIRECT feature enabled
       Auto Mode
       File size set to 102400 kB
       Record Size 4 kB
       Record Size 16 kB
       Record Size 512 kB
       Record Size 1024 kB
       Record Size 16384 kB
       Command line used: iozone -e -I -a -s 100M -r 4k -r 16k -r 512k -r 1024k -r 16384k -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     bkwd    record    stride
             kB  reclen    write  rewrite    read    reread    read     write     read   rewrite      read   fwrite frewrite    fread  freread
         102400       4     1089     1639     6205     6200     6084      861
         102400      16     4487     5281    16132    16142    15479     3042
         102400     512    11815    13298    23110    23115    23096     9730
         102400    1024    12785    12821    23101    23133    23127     9972
         102400   16384    12869    12833    23200    23204    23203    11373

iozone test complete.

Samsung EVO 64GB:

Code:

       Iozone: Performance Test of File I/O
               Version $Revision: 3.429 $
               Compiled for 64 bit mode.
               Build: linux

       Run began: Sat Jun 18 11:16:53 2016

       Include fsync in write timing
       O_DIRECT feature enabled
       Auto Mode
       File size set to 102400 kB
       Record Size 4 kB
       Record Size 16 kB
       Record Size 512 kB
       Record Size 1024 kB
       Record Size 16384 kB
       Command line used: iozone -e -I -a -s 100M -r 4k -r 16k -r 512k -r 1024k -r 16384k -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     bkwd    record    stride
             kB  reclen    write  rewrite    read    reread    read     write     read   rewrite      read   fwrite frewrite    fread  freread
         102400       4     2340     3036     6793     6788     6827     3094
         102400      16     8928    10921    13699    13680    13701    11189
         102400     512    15670    15867    22157    21623    21658    17665
         102400    1024    14191    17536    22134    21658    21944    18878
         102400   16384    16842    18374    21611    21916    21188    17114

iozone test complete.

As you can see the Samsung card performs better for sequential reads/writes especially for small record sizes, and more importantly for random writes. I will test the EVO64 some more...

Samsung EVO 64GB with data=writeback option set: sudo tune2fs -o journal_data_writeback /dev/mmcblk0p2

Code:

ubuntu@pine64:~$ iozone -e -I -a -s 100M -r 4k -r 16k -r 512k -r 1024k -r 16384k -i 0 -i 1 -i 2
       Iozone: Performance Test of File I/O
               Version $Revision: 3.429 $
               Compiled for 64 bit mode.
               Build: linux

       Run began: Sat Jun 18 11:29:59 2016

       Include fsync in write timing
       O_DIRECT feature enabled
       Auto Mode
       File size set to 102400 kB
       Record Size 4 kB
       Record Size 16 kB
       Record Size 512 kB
       Record Size 1024 kB
       Record Size 16384 kB
       Command line used: iozone -e -I -a -s 100M -r 4k -r 16k -r 512k -r 1024k -r 16384k -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     bkwd    record    stride
             kB  reclen    write  rewrite    read    reread    read     write     read   rewrite      read   fwrite frewrite    fread  freread
         102400       4     2319     3321     7347     7363     7357     3526
         102400      16    10166    12550    14464    14432    14471    12570
         102400     512    21591    21566    23143    23140    23139    21616
         102400    1024    21653    21565    23097    23100    23098    21645
         102400   16384    21678    21662    23084    23082    23082    21690

iozone test complete.

[Bob123456789]

Thank you so much xalius. It so helpful.

For random writes:
Pine 52%
Samsung 79%
EVO 100%

I had ordered three of the Pine cards thinking it would provide me with the most flexibility among operating systems and that the Pine team would have selected the best ones for their boards. I also felt that it would get me up and running quickest without having to chase down the best, copy each operating system and set them up, each with their own quirks.

One of the key things I wanted to do with the Pine was heavily read/write dependent.

I am now wondering if for the primary card for that use if it makes sense to replace it with an EVO.

[xalius]

Thank you so much xalius.  It so helpful.

For random writes:
             Pine 52%
             Samsung 79%
             EVO 100%

I had ordered three of the Pine cards thinking it would provide me with the most flexibility among operating systems and that the Pine team would have selected the best ones for their boards.  I also felt that it would get me up and running quickest without having to chase down the best, copy each operating system and set them up, each with their own quirks.

One of the key things I wanted to do with the Pine was heavily read/write dependent.

I am now wondering if for the primary card for that use if it makes sense to replace it with an EVO.

If it's Linux you are talking about you could also try to run the root-fs from a USB stick and just boot from the sdcard, put /tmp and other temporary folders on a ramdrive...

[Bob123456789]

What speeds would we get using a USB 2.0 stick?

Would it be faster or slower than the Pine SD card?

[tllim]

What speeds would we get using a USB 2.0 stick?  

Would it be faster or slower than the Pine SD card?

Good one can be faster due to USB transfer speed faster than SD bus.

[xalius]

What speeds would we get using a USB 2.0 stick?  

Would it be faster or slower than the Pine SD card?

I tested a random stick the same way I tested my sdcards and while it was a little faster at sequential reads and writes, the random I/O performance was abysmal compared to the Samsung sdcard I use... needs more testing I guess, there are a lot of different flash controllers in those sticks... let me find the test results....

[tllim]

What speeds would we get using a USB 2.0 stick?  

Would it be faster or slower than the Pine SD card?

I tested a random stick the same way I tested my sdcards and while it was a little faster at sequential reads and writes, the random I/O performance was abysmal compared to the Samsung sdcard I use... needs more testing I guess, there are a lot of different flash controllers in those sticks... let me find the test results....

Random IO is very important factor. Looking forward on your finding.

[xalius]

Just tested my USB Stick again, it's a 64GB Sony USB3.0 (dont know the exact model, PID is 0x054c, VID is 0x05b8) here are the results...

Code:

ubuntu@pine64:/mnt$ sudo  iozone -e -I -a -s 100M -r 4k -r 16k -r 512k -r 1024k -r 16384k -i 0 -i 1 -i 2
       Iozone: Performance Test of File I/O
               Version $Revision: 3.429 $
               Compiled for 64 bit mode.
               Build: linux

       Run began: Sun Jun 19 01:08:52 2016

       Include fsync in write timing
       O_DIRECT feature enabled
       Auto Mode
       File size set to 102400 kB
       Record Size 4 kB
       Record Size 16 kB
       Record Size 512 kB
       Record Size 1024 kB
       Record Size 16384 kB
       Command line used: iozone -e -I -a -s 100M -r 4k -r 16k -r 512k -r 1024k -r 16384k -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     bkwd    record    stride
             kB  reclen    write  rewrite    read    reread    read     write     read   rewrite      read   fwrite frewrite    fread  freread
         102400       4     6827     5851    10475    10383     5126      196
         102400      16     7675     7773    21236    21252    14183      795
         102400     512    15876    11976    38357    38526    37429     5403
         102400    1024    11973    11278    39063    39085    38639     5687
         102400   16384    13439    10519    41124    41095    41127    12157

iozone test complete.

As you can see, read performance reaches twice the speed of my Samsung card (see above) for large record sizes but the random writes for small block sizes show really bad performance. I wonder if that is due to the ext4 filesystem I put on the stick...