On Tue, Feb 3, 2009 at 5:17 PM, Koen Kooi <k.k...@student.utwente.nl> wrote: > > Op 3 feb 2009, om 12:37 heeft Gadiyar, Anand het volgende geschreven: > >> >> >>> -----Original Message----- >>> From: linux-omap-ow...@vger.kernel.org >>> [mailto:linux-omap-ow...@vger.kernel.org] On Behalf Of Koen Kooi >>> Sent: Tuesday, February 03, 2009 5:05 PM >>> To: Remith Ravi >>> Cc: linux-omap@vger.kernel.org >>> Subject: Re: USB and SD performance issue >>> >>> >>> Op 3 feb 2009, om 12:16 heeft Remith Ravi het volgende geschreven: >>> >>>> Hello, >>>> >>>> Recently I have done a performance comparison of OMAP3 'USB >>> >>> OTG Host' >>>> >>>> and 'SD' drivers on different Linux Kernels. >>>> The performance of latest git kernel is less comparing to >>> >>> 2.6.22 TI >>>> >>>> kernel. >>>> Could some one please explain me what cause this performance >>>> difference? >>>> >>>> The performance is calculated by doing data transfer between RAM and >>>> the storage device (using 'time' command). >>>> >>>> The performance details are given below. >>>> >>>> 1) 2.6.29-rc2 kernel >>>> USB OTG (Host) Read (4.2 MB/s) Write >>> >>> (2.6 MB/s) >>>> >>>> SD Card Read (3.88 MB/s) >>> >>> Write (3.1 MB/s) >>>> >>>> 2) 2.6.28 kernel >>>> USB OTG (Host) error error >>>> SD Card Read (4.2 MB/s) >>> >>> Write (3.27 MB/s) >>>> >>>> 3) 2.6.22 TI kernel >>>> USB OTG (Host) Read (5.4MB/s) Write >>> >>> (6.14MB/s) >>>> >>>> SD Card Read (6.7MB/s) Write (6.43MB/s) >>> >>> I get this with 2.6.28: >>> >>> r...@omap3evm:~# hdparm -t /dev/mmcblk0 >>> /dev/mmcblk0: >>> Timing buffered disk reads: 42 MB in 3.10 seconds = 13.53 MB/sec >>> >>> r...@beagleboard:~# hdparm -t /dev/mmcblk0 >>> /dev/mmcblk0: >>> Timing buffered disk reads: 44 MB in 3.13 seconds = 14.04 MB/sec >>> >>> regards, >>> >>> Koen >>> >> >> What does `hdparm -tT /dev/mmcblk0` give? > > r...@beagleboard:~# hdparm -tT /dev/mmcblk0 > /dev/mmcblk0: > Timing cached reads: 240 MB in 2.01 seconds = 119.32 MB/sec > Timing buffered disk reads: 42 MB in 3.01 seconds = 13.97 MB/sec > > r...@omap3evm:~# hdparm -tT /dev/mmcblk0 > /dev/mmcblk0: > Timing cached reads: 300 MB in 2.00 seconds = 149.83 MB/sec > Timing buffered disk reads: 42 MB in 3.17 seconds = 13.23 MB/sec > > Which is misleading, since -T tests the bandwidth or your RAM, which in my > case is limited since both machines are displaying a stream from a webcam > attached to USB in X11, so the overlays and framebuffers are stealing a > large part of the bw. Killing the screensaver on the EVM gave me 35MB/s > 'back' :) > > regards, > > Koen > > >> >> >> - Anand >> > > With "hdparm" kernel 2.6.29 gives better result.
What I tried was booting with ramdisk and transfer a 30MB file between the file system and storage device. [r...@omap3530evm ~]# time cp data1 /mnt/mmc/ [r...@omap3530evm ~]# time cp /mnt/mmc/data2 . Any guess why the contradiction? Regards, Remith -- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
