Hi Viral,
I have changed the frequency and checked it on a socpe at about 50Mhz,
its definately a crappy driver for writes. I have not checked reads yet,
but from the scope data they must be over triple the write rates.
A chap from on another forum got in touch with me, he dosent know the
Davinci platform, but rather the OMAP2/3 and apparently they are getting
significantly faster performances.
Using the following script:-
//*********************************************************************
echo STARTING
umount /media
mount /dev/mmcblk0 /media -o sync
echo COPYING 80MB megabytes
time dd bs=8192 if=/dev/zero of=/media/tstfile count=10240
echo UMOUNTING
time umount /media
echo DONE
//*********************************************************************
I get these results:-
Card
Minutes
Seconds
mbit/s
mbyte/s
512 Integral
MMC
1
58
5.42
0.68
1GB Integral
MMC
2
1
5.29
0.66
1GB Kingston
MMC
3
35
2.98
0.37
1GB Sandisk SD
5
15
2.03
0.25
512 Sandisk SD
9
25
1.13
0.14
I think its about time the people at TI get to work on a proper driver,
there must be loads of people who would like some decent speed for
writing. This causes all sorts of issues when writing video to the
MMC/SD and really eats up system resources.
Can we please have a Ti response regarding this?
Regards,
Andrew
On Sun, 2007-05-13 at 10:20 +0530, Viral Sachde wrote:
> Hi Andrew,
>
> I feel that you are getting this results because of low freq. Can you
> try and increase freq. I think you will get information for how to
> increase freq in manual. (you will need to set couple of reg.s)
>
> Regards,
>
> Viral
>
>
>
>
> On 5/11/07, Andrew Armstrong <[EMAIL PROTECTED]> wrote:
> >
> > I have some more information regarding this issue, I have
> been testing
> > out the driver and discovered that in fact varying of the
> internal
> > settings, dma, 4 bit access etc does not make any
> difference.
> >
> > A scope revealed that the card is only being accessed only
> for a
> > fraction of the time. So the key to getting this upto any
> reasonable
> > speed is increase the rate at which the kernel services this
> device.
> >
> > Any ideas?
> >
> > Andrew
> >
> >
> > On Thu, 2007-05-10 at 10:04 +0100, Andrew Armstrong wrote:
> > > Hi Guys,
> > >
> > > I have been experimenting with the MMC/SD interface. I am
> finding that
> > > transfers are really slow, with figures like 3.96 mbit/s
> (async) and
> > > 2.94 mbit/s (sync).
> > >
> > > Does anyone know anything about the MMC driver and where
> the bottleneck
> > > is? I am considering modifying it accordingly, but I do
> not yet have
> > > enough understanding of driver and hardware involved.
> > >
> > > Is there any way to gauge what the theoretical
> through-puts to this
> > > device should be?
> > >
> > > Regards,
> > >
> > > Andrew
> > >
> > > _______________________________________________
> > > Davinci-linux-open-source mailing list
> > > [email protected]
> > >
> http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
> >
> > _______________________________________________
> > Davinci-linux-open-source mailing list
> > [email protected]
> >
> http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
> >
>
_______________________________________________
Davinci-linux-open-source mailing list
[email protected]
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
