7 Jul
2009
7 Jul
'09
9:14 p.m.
Troy Kisky wrote:
Mark Brown wrote:
Even when using SDRAM, there is still another advantage to double buffering. That is tolerating very long interrupt latency on dma completion. The current code sets up the next periods dma transfer on dma completion. This could easily lead to underruns. So, even in this case, I think double buffering is worth the extra dma engine copy.
I overstated my case, it still would not "easily" happen.