Talk:CPU power up state

Parts of the APU which appear to be affected by reset (determined by tracing the reset signal throughout the chip):

• Square channel volume/envelope timer (not the reload value at $4000/$4004 bits 0-3) and counter (the actual volume output)
• Noise channel volume/envelope timer (not the reload value at $4008 bits 0-3) and counter (the actual volume output)
• Square channel frequency counter, including the 3 duty cycle bits
• Triangle channel frequency counter
• Noise channel frequency counter, sort of - it forces the LFSR input to 0, so a very short reset pulse might only partially clear it
• DPCM channel frequency counter, sort of - it forces the LFSR input to 0, so a very short reset pulse might only partially clear it
• Noise channel frequency ($400E bits 0-3)
• Square channel sweep counter
• All 4 length counters
• All 5 channel enables ($4015)
• Triangle channel linear counter (not the reload value at $4008)
• Triangle channel Position
• DPCM channel Sample Length counter (not the reload value)
• DPCM channel Position ($4011)
• DPCM channel Sample Buffer and Bit Counter
• DPCM channel Sample Address counter (not the reload value)
• DPCM channel state machine
• Sprite DMA address counter and state machine
• Part of the Frame Counter (notably, not the two writable bits of $4017)
• Debug register $401A bit 7

--Quietust 16:40, 8 June 2011 (UTC)

Detecting User Reset

Is there way using common mappers to check if power on or reset button pushed? --Zzo38 18:35, 10 September 2012 (MDT)

Common practice seems to be "put a canary in memory, if it's still there assume warm (re)boot instead of cold boot. e.g., Galaxian puts the programmer's name in $0100-$010F, and if it's there on boot, it doesn't clear high scores or the easter egg counter.—Lidnariq 19:47, 10 September 2012 (MDT)