SD card speed

[Intron]

Marc van Woerkom wrote:

> I did some more tests.
> 
> Using smaller blocksizes (newfs_msdosfs -b 4096) yields even slower write
> transfer rates. It looks like there are only 7 transfers per seconds, thus 
> if
> the blocksize is smaller, less data is shoveled over.
> 
>      tty             da2             cpu
> tin tout  KB/t tps  MB/s  us ni sy in id
>   0   65  4.00   7  0.03  14  0  6  2 78
>   0  181  4.00   7  0.03  11  0  7  2 80
>   0   89  4.00   7  0.03  13  0  8  4 75
>   0   69  4.00   7  0.03   9  0  9  2 80
>   0   76  4.00   6  0.02  16  0  4  4 77
>   0   71  4.00   7  0.03  12  0 10  3 75
>   0  132  4.00   7  0.03   6  0 16  3 75
>   0   76  4.00   7  0.03  34  0  8  4 54
>   0   78  4.00   7  0.03   8  0  5  2 86
>   0  152  4.00   7  0.03  14  0  5  1 80
>   0   87  4.00   6  0.02  12  0  6  1 81
>   0   80  4.00   7  0.03   9  0  7  3 80
> 
> It doesn't seem to matter if it is FAT or FAT32.
> 
> However if I format the SD card as UFS2 system, the card is fast:
> 
> [root at hokage /]# fdisk /dev/da2
> ******* Working on device /dev/da2 *******
> parameters extracted from in-core disklabel are:
> cylinders=472 heads=64 sectors/track=32 (2048 blks/cyl)
> 
> parameters to be used for BIOS calculations are:
> cylinders=472 heads=64 sectors/track=32 (2048 blks/cyl)
> 
> Media sector size is 512
> Warning: BIOS sector numbering starts with sector 1
> Information from DOS bootblock is:
> The data for partition 1 is:
> sysid 165 (0xa5),(FreeBSD/NetBSD/386BSD)
>    start 32, size 966624 (471 Meg), flag 80 (active)
>        beg: cyl 0/ head 1/ sector 1;
>        end: cyl 471/ head 63/ sector 32
> The data for partition 2 is:
> <UNUSED>
> The data for partition 3 is:
> <UNUSED>
> The data for partition 4 is:
> <UNUSED>
> 
>     tty             da2             cpu
> tin tout  KB/t tps  MB/s  us ni sy in id
>  27   55  1.45   1  0.00  12  0  7  2 79
>   0  162 128.00  20  2.50   8  0  9  4 80
>   0   70 128.00  23  2.87   5  0  9  3 83
>   0   66 128.00  21  2.62  12  0  7  1 80
>   0   44 128.00  24  3.00   9  0  8  1 83
>   0   72 128.00  22  2.75   7  0  5  1 87
>   0  127 128.00  23  2.87  12  0 13  3 71
>   0   66 128.00  23  2.87   5  0  7  4 84
>   0   81 128.00  23  2.87  17  0  8  4 71
>   0   66 128.00  23  2.87   7  0  9  4 80
>   0   70 128.00  22  2.75  12  0 11  3 74
>   0   90 128.00  23  2.87  10  0  8  2 80
> 136   81 128.00  20  2.50   9  0  9  2 80
> 
> Thus the slowdown seems to be related to using a FAT filesystem on
> that SD card.
> 
> Last thing I have to test is what happens, if I partition/format that
> SD card under Windows.
> 
> Regards,
> Marc
> 
> _______________________________________________
> freebsd-usb at freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-usb
> To unsubscribe, send any mail to "freebsd-usb-unsubscribe at freebsd.org"

Due to the asynchronous transmission mechanism and somewhat long delay
on line, USB requires device to transfer data in as large block
as possible to gain high transmission rate.

Have you tested your card reader on Microsoft Windows?
FreeBSD's FAT module probably isn't optimized enough.

------------------------------------------------------------------------
                                                From Beijing, China