7th February 2006, 01:59 pm
Last year I did some experiments with the Theora video decoder on a Texas Instruments DM642 DSP. A royalty free video decoder is very attractive for embedded devices, but after some major restructuring for performance, some problems remained.
The main problem is that, unlike MPEG video, Theora video is not packed in the bitstream in the raster order that it is displayed on screen, but instead in Hilbert curve order. This is not a problem in itself, but Theora’s DC prediction and post-processing loop filter are both defined in raster order. The need to go over the data once in Hilbert curve order and once in raster order leaves Theora decode requiring higher memory bandwidth than MPEG decode.
The encoder faces a similar problem. Andrey N. Filippov describes an FPGA implementation of the Theora Encoder, and comments on the high memory bandwidth required. The solution in the article is to implement a custom SDRAM controller with knowledge of the Theora data structures, an option not available on a DSP.
There are other minor problems remaining. The DM642 has instructions to assist video encoding and decoding, but these are optimised for MPEG and may not easily apply to Theora. For example, the avg2 instruction averages two pairs of 16-bit values, but it uses the formula (x + y + 1) >> 1, whereas Theora’s half-pixel predictor uses the formula (x + y) >> 1.
Where does this leave Theora decode on DSP? The DM642 is just capable of decoding NTSC quality video (640×480, 30fps) provided that the bitrate is controlled. The good news is that the newer DaVinci architecture provides extra memory bandwidth through a DDR2 memory controller, plus the possibility of splitting the workload to place bitstream decode on the ARM processor and frame reconstruction on the DSP.
1st February 2006, 04:02 pm
After several years of use, the server for my development machines developed a problem. If I rebooted the machine without an internet connection, it would pause for several minutes while starting the NFS service.
The Linux NFS server keeps track of status in several files. Whenever a client mounts a filesystem, the server records this in /var/lib/nfs/rmtab. If the server reboots, exportfs passes this list to the kernel to ensure that the reboot is invisible to the clients. So far, this is harmless.
If the server is used with clients that do not send clean unmount requests, such as diskless machines in a development lab, then rmtab fills up with entries for machines that aren’t around anymore. This is mostly harmless.
The final part of the problem is that on reboot exportfs performs a reverse DNS lookup for each entry in rmtab. If the DNS server is unavailable, the request waits for a timeout. This can take a very long time.
And the moral of the story? If NFS clients come and go on your network, check rmtab for clutter.
24th August 2004, 12:00 am
As of Linux 2.6.9.1, the driver is in the kernel as i2c-mpc.c. This version of the driver has been tested on Tsi107 (formerly MPC107), and on MPC8240, MPC8540, and MPC5200 PowerPCs. Thanks to all the testers.
18th August 2004, 12:00 am
The Tsi106 (formerly MPC106) and Tsi107 (formerly MPC107) PowerPC host bridges both contain a data cache. This note describes why it may be necessary to mark memory as SMP coherent even in a single processor system using these bridges.
Continue reading ‘MPC107/Tsi107 Cache Coherency’ »
29th March 2003, 12:00 am
The driver is supplied ready to be inserted into the linuxppc-2.4-devel tree.
The driver has so far been tested on a small number of systems and with a small number of I2C parts. Thanks to Gleb Natapov, Matthew S. McClintock, and others for testing. Further test results are welcome.
Continue reading ‘MPC107 I2C Driver’ »
