Sorgelig wrote:Btw, i see you use different MPFE config. I found HPS tends to hang if MPFE config changed between cores. That's why all my cores use the same config. And i couldn't find the way to reset it or safely re-configure without hang either in current core or in the core loaded next.
Yes, in the latest version I recovered your configuration, and it works better. The changes I made were not really needed.
Why so much RAM bandwidth ?
- Raster resolution is 1024x768. Of course it could be lowered, it's just changing a scale factor and the "modeline" configuration.
- Image is generated as series of "dots" calculated by adders and multiplexers (replacing analog integrators, DAC and muxes). To get continuous lines with high slew rate, the sampling rate must be quite high.
- "Dots" are rendered from 4x4 pixel patterns, which are selectable, allowing configurable spot size. 16 pixels are updated for each "dot". Of course it could be reduced.
- Each pixel uses 16bits! 8 bits for intensity, 8 bits for the age of the pixel, actually a timecode. Older pixels are darkened then blanked : this is how phosphor persistence is simulated. It could be reduced, with linear decay 8 bits is enough for both age and intensity.