Cabal
02-22-2007, 01:22 PM
Objective
To test bandwidth capabilities of a stock, unhacked TiVo Series2 DT and determine any contributing factors to throughput.
Background
http://en.wikipedia.org/wiki/TiVo_DVRs
http://en.wikipedia.org/wiki/TiVo_DVRs#Series2
http://www.broadcom.com/products/Satellite/Satellite-Set-Top-Box-Solutions/BCM7318
Test Unit
Tivo Series2 DT 649-series
CPU: Broadcom BCM7138, 266 MHz MIPS
RAM: 64 MB of 133 MHz 16-bit DDR
MPEG Encoder: Two Broadcom BCM7042
Hard Drive: 80 GB
Network: Onboard ethernet, negotiated at 100 mbps full-duplex
Software: 8.1a
Methodology
Testing was done with Curl and the HTTP interface on a mostly-idle, switched, gigabit ethernet network. Each test used a 500 MB file, left running to get an accurate sustained speed, and repeated 5+ times for accuracy. Results were sanity-checked with a browser and TivoToGo. Use of Curl was as follows:
curl -k --digest -u tivo:XXX -c /dev/null -o test.tivo $URL
Results
http://home.comcast.net/~axp696/tivo_speeds.jpg
Observations
- The tests were all extremely consistent. They reached top speed immediately, and rarely deviated by more than 10 KB/s.
- Decoding has a significantly greater effect on throughput than encoding.
- Encoding has a small effect on throughput, but shouldn't be discounted entirely.
Conclusions
- Bandwidth is limited by the processing power of the BCM7318 chip, which covers decoding, general processing, and ethernet.
- The best speed gains, upwards of 25%, can be had by limiting the decoding work of the unit. The blue non-channels (1, 99, etc) work best for this.
- The best transfer rates are achieved with both tuners viewing blue non-channels. This configuration is recommended for GB+ transfers when viewing is not necessary.
Comments, arguments, or further analysis welcome.
To test bandwidth capabilities of a stock, unhacked TiVo Series2 DT and determine any contributing factors to throughput.
Background
http://en.wikipedia.org/wiki/TiVo_DVRs
http://en.wikipedia.org/wiki/TiVo_DVRs#Series2
http://www.broadcom.com/products/Satellite/Satellite-Set-Top-Box-Solutions/BCM7318
Test Unit
Tivo Series2 DT 649-series
CPU: Broadcom BCM7138, 266 MHz MIPS
RAM: 64 MB of 133 MHz 16-bit DDR
MPEG Encoder: Two Broadcom BCM7042
Hard Drive: 80 GB
Network: Onboard ethernet, negotiated at 100 mbps full-duplex
Software: 8.1a
Methodology
Testing was done with Curl and the HTTP interface on a mostly-idle, switched, gigabit ethernet network. Each test used a 500 MB file, left running to get an accurate sustained speed, and repeated 5+ times for accuracy. Results were sanity-checked with a browser and TivoToGo. Use of Curl was as follows:
curl -k --digest -u tivo:XXX -c /dev/null -o test.tivo $URL
Results
http://home.comcast.net/~axp696/tivo_speeds.jpg
Observations
- The tests were all extremely consistent. They reached top speed immediately, and rarely deviated by more than 10 KB/s.
- Decoding has a significantly greater effect on throughput than encoding.
- Encoding has a small effect on throughput, but shouldn't be discounted entirely.
Conclusions
- Bandwidth is limited by the processing power of the BCM7318 chip, which covers decoding, general processing, and ethernet.
- The best speed gains, upwards of 25%, can be had by limiting the decoding work of the unit. The blue non-channels (1, 99, etc) work best for this.
- The best transfer rates are achieved with both tuners viewing blue non-channels. This configuration is recommended for GB+ transfers when viewing is not necessary.
Comments, arguments, or further analysis welcome.