Quote:
Originally Posted by igorsk
The Renesas chip is the "subcpu" mentioned in the subject of this thread. It is present in all models beginning from PRS-500 (and probably Librie). Sony uses it to handle the simple tasks of controlling keyboard, power, leds and touch screen. I guess this way the big and hungry ARM can spend most of the time sleeping on low power but when it needs some time for number crunching - such as rendering HTML - it doesn't need to bother about all the "small fish". I think it's also for this reason that Sony went back to using a dedicated SD/MS controller (Ricoh R5C807) after a short stint of relying on the built-in CPU ports in PRS600. Such extra hardware might increase the cost a bit but it seems it does help to save power.
|
Hmm.. I thought about that when I first was looking at it; I don't have any power measurements to check with, but what is really odd is that Sony still chose to use that particular ARM processor with all its internal peripherals. I was looking at another arm 9, which does 800MHz, but doesn't have the multimedia extensions -- and the power consumption can be down in the microamps, with the processor running at 1Mips or so -- the same as the Renes... cpu. The thing is, the particular processor they have chosen -- the same as the kindle, BTW, but a mysterious extra letter "D" in the sony version.... and no explanation in the data sheets -- I would think still has to clock all those peripherals even when not in use. It can partially shut them down, but their very existence is going to cost some power. Multiplier units with multiword fp registers are notorious power hogs. Yet they still put this ARM in a machine with a display that is incapable of 30 frames a second video.... My old pentium SSE2 machine with SSE2 turned off renders HTML much faster than the Sony.
Looking at the video driver, too, is a puzzle -- for there are huge timeouts on the order of a second (100's of milleseconds at least); when a complete shut down and restart of the video chip is mentioned in the comments to be on the order of 2ms -- and I know for a fact the screen keeps its image with power off. OUCH! There are reasons for this kind of code -- but it really looks to me like someone had to get it done FAST, and that information was not given to the engineer/software people in a timely fashion or something was misrepresented. Many of the datasheets are not available directly -- eg: samsung has 1G memories on their site, but the 2G memories in the sony have no datasheets. I found the pinouts, but it wasn't direct.
It all speaks of closed shop deals between companies, and a little misrepresentation going on.
Ps. I just re-tried the 1.05a flasher. Copied the autorun and the test-flash image files into the internal /test directory of the Sony. Negative -- nothing bad or good at all, as if it wasn't there. I don't have windows, so I didn't run the msc executable -- but that shouldn't matter for I did everthing the batch files indicated, and I even tried adding an asterisk to the MD5 just for retentiveness on the fourth try.
Getting the data off the ROM is problematic, for its interface to the ARM isn't clear to me. It has an MCU of some kind according to Samsung, and I am not sure if it is emulating a hard drive or what. I need to adsorb the kernel code for the nand memory .... but I don't think I have enough time.
cheers .... comments, suggestions, and questions welcome.