| 11 Jan 2024 |
 tuxinator2009 | FYI: Kevin and I got the question a lot even for something as simple as the Arduboy. | 16:53:37 |
| tuxinator2009 | * FYI: Kevin and I got the question a lot even for something as simple as the Arduboy at the Portland Retro Game Expo. | 16:53:48 |
| tuxinator2009 | To be fair part of the confusion behind the OS topic is because a lot of these emulation station handhelds are more-or-less the same as a raspberry pi with a custom interface (or even just retro arch preloaded). | 16:54:54 |
 mlxxxp | Actually, a desktop consumer printer is probably a better example to compare the processor in the Pokitto to than a microwave. | 17:16:14 |
| mlxxxp | * Actually, a desktop consumer printer is probably a better example to compare the processor in the Pokitto to than a microwave, from a processing power perspective. | 17:16:34 |
| mlxxxp | * Actually, a desktop consumer printer is probably a better example to compare the processor in the Pokitto to than a microwave, from a processing power requirement perspective. | 17:17:09 |
| tuxinator2009 | True, but printers are notorious for not working and breaking down so 🤷♂️ 🤣 | 17:17:28 |
| mlxxxp | Well, the Pokitto hasn't been without issues 😉 | 17:18:49 |
| tuxinator2009 | Haha, fair point | 17:19:17 |
 Jonne | Again, almost first result from Google: "For black and white laser printers, the memory range will be between 32 MB to 64 MB. On the other hand, color laser printer memory requires 128MB." | 17:50:12 |
| Jonne | Call me back when you find a printer with 36kb ram | 17:50:49 |
| Jonne | Any kind of printer that stores a page needs a chip with a static memory controller for external RAM | 17:52:58 |
| Jonne | I don't mind arguing but I don't like to argue for the sake of arguing. Put something on the table, a source or any kind of proof positive for your views | 17:54:37 |
| Jonne | I maintain that the most likely home of a Cortex M0 of any variety is a consumer device with a simple HMI and a need for a bunch of I/O, such as pushbuttons. Hence Yale Doorman | 17:58:11 |
| Jonne | * I maintain that the most likely home of a Cortex M0 of any variety is a consumer device with a simple HMI, a fairly complex state machine and a need for a bunch of I/O, such as pushbuttons. Hence Yale Doorman | 17:58:58 |
| Jonne | The only unorthodox thing w/r to Pokitto is the fact that a display is driven by bitbanging the display controller. | 18:01:37 |
| Jonne | That's the only thing the Cortex M0 was not "designed" to do, since it has no LCD driver of any kind | 18:02:17 |
| tuxinator2009 | It also doesn't have that higher amount of parallel SPI available. | 18:02:47 |
| tuxinator2009 | Some chips will have QSPI so as to support larger SRAM chips, but nothing they're designed for would really need 8-bit parallel I/O let alone 16-bit. | 18:03:34 |
| Jonne | Yes and if you need external ram for an ARM Cortex you use a chip with a FSMC and the ram is mapped into the 32-bit 4GB address space for direct accessing. LPC4096 is an example | 18:07:16 |
| Jonne | You don't even use the QSPI in those cases | 18:07:40 |
| tuxinator2009 | Ah yes I often forget about FSMC because it's not really seen much in hobbyist electronics. | 18:08:11 |
| Jonne | Now I need to make some food | 18:09:05 |
 carbonacat | Jonne the amazing thing is that when when a friend talks to me about the Pokitto she often says "microwave console" (because that's I described the mcu being used in microwave) 😛 somehow it sticked better than Pokitto | 19:14:56 |
| carbonacat | "I make games that toast the CPU" | 22:10:35 |
 spinal_cord | Here's a question to ponder before I go to bed. I've always used 4 audio buffers in my loop so that reading data never crosses the audio that is playing, there is alsway a 1 buffer gapp between the two. But is double buffering better from a speed point of view? Does it make a difference? | 22:27:25 |
| tuxinator2009 | I usually try and go for a double-buffer approach where the interupt function simply increments the pointer until it reaches the buffer end then gets set to a bufferStart value and sets a needsUpdating boolean to true. Then there's an update method called each frame which checks if this flag is set and if so fills the second buffer and changes the bufferStart to point to the start of the other buffer. Not sure if this is actually faster or not though. | 22:29:41 |
| tuxinator2009 | To prevent race conditions the second buffer is filled and then the bufferStart is the last thing to be changed so that in the event the interrupt reaches the end again before the other buffer is filled it would simply repeat the contents of the first buffer (causing the common stuttering effect when the game loop hangs too long). | 22:31:23 |
| carbonacat | I think double buffer is already what you're doing if you're writing to one buffer while the other is being read | 22:31:52 |
| carbonacat | the current audio code let the audio handlers write into a buffer as soon as it's ready to be filled (the interrupt does the swap only if it finished reading the current buffer) | 22:34:42 |
