madias wrote:I've done a little benchmark with different buffer sizes (with low level pin manipulation):
On a buffer 256 I have
2459us IDLE time
= 8.034% CPU time
On a buffer 16:
= 8.079% CPU time
So the match winner is: Buffer 16 with better audio quality
So in this calculation it's not important how the CPU time for pin manipulation is, because it's about the relationship between the results. So the main CPU time is calculating the real time audio itself.
Strange thing: (I suspect it, must check it again): On a PIC32MX250 @68MHZ nearly the same code (but without DMA for I2s!) is faster.
So in my final code I suspect to need 10-12% CPU cycle time for each oscillator, so all in all about 40-50% of CPU time, but I need TFT, encoders-knobs-buttons, (USB) MIDI, and more permanent calculations (not on a high frequency as the osc code) like 8x ADSR's , 8-12x LFO's and DAC's (12x) outs to control the external filter board....
madias wrote:Yes, the LED pin is on on the IRQ start and off on the IRQ end
madias wrote:Ok, answer to my hw timer questionFreeRTOS uses systick with 1khz
madias wrote:Next question answered to myself:
In the current STM32F1 folder there are two libraries:
"FreeRTOS" this is the old V7.0.1 and
"FreeRTOS821" this is the newest version 8.21
Victor: I found your wav player example at https://github.com/victorpv/Arduino_STM ... odule2.ino
In this example you use coos instead of RTOS - why?
edit: Ok, on the first look I feel Coos has the better user manual and looks easier to understand than RTOS
Users browsing this forum: No registered users and 1 guest