Hello That was an enlightening story!
If I understand you right, it is possible to do some kind of power management, it is just not manageable? If so here is a naive question: Would it be possible to shut down the entire card completely, totally ignore any dependencies between it's components, maybe leave it in some partly unworkable state, and simply reset / power it up when needed? I could easily live with some reinitialization delay and other quirks if it would just stop drawing 10-15W at all times. Simon On Wednesday 07 June 2006 16:15, Chris Kennedy wrote: > Wilhelm Eger wrote: > > Hi there, > > > > is there any chance to shut down the decoder of a PVR 350 to save power? > > The cx23415 (the main mpeg encoder/decoder chip) actually uses hardware > units from both the encoder/decoder for both encoding and decoding, so > basically no it's impossible (for the mpeg encoder/decoder chip that is) > because the decoder is used for encoding somewhat (like audio, PCM > processing, and even DMA xfers cross over for how they function). It's > a bad design being tangled up like that and one of the big flaws of > these chips is exactly that, power hungry and encoder/decoder were not > ever engineered properly to be separate (the cx23416 on the > pvr250/150/500 had the hardware units switched around some to chop out > the decoder but leave the units needed for the encoder). It originally > was designed under the concept of a full set top box core and so > everything was just thrown in together, they even forgot to have the > Java processors separated enough to safely encode/decode together :) (so > obviously more than just bad design but also major flaws that weren't > able to be fully realized till too late in the game, it took years it > seems to fully test all hardware units enough to know they were flawed > it seems, I'm not sure what they were doing from 99'-01', the chip was > birthed back in the late 90's) hence it's lucky that works so well > actually through mulitple firmware hacks and driver hacks (and the > wonderful DMA errors are partly a still visible side effect of that > problem, 3 non-atomic locking Java processors working side by side > blindly and accessing the same global block of registers is a 'bad thing'). > > Thanks, > Chris > > > Greetings, > > > > Wilhelm > > > > _______________________________________________ > > ivtv-devel mailing list > > [email protected] > > http://ivtvdriver.org/mailman/listinfo/ivtv-devel > > _______________________________________________ > ivtv-devel mailing list > [email protected] > http://ivtvdriver.org/mailman/listinfo/ivtv-devel _______________________________________________ ivtv-devel mailing list [email protected] http://ivtvdriver.org/mailman/listinfo/ivtv-devel
