That appears like all my threads are bound to one core on the Pi 3. Value |= (0x01 Write(GpioPinValue::High) įuthermore, how does one get a Win IoT app to use multiple cores? When I run my app (debug or optimized) I can never get more than 25% CPU usage. I expected some overhead for exercising the GPIO pins on Win IoT - but a 20x penalty? Some example code below:įor (int i = 0 i Read() = GpioPinValue::High) The Vortex86DX-based board can do it in less than 1 µs. Simply dropping the !IOR (read) line and then reading 8 data lines is taking on the order of 20 µs. However, I am finding the performance is quite poor. I had to write the GPIO functions in Windows 10 IoT Core using the Windows::Devices::Gpio library. The embedded hardware only uses the simplest 8-bit PC/104 transfers (ALE, !IOR, !IOW, 4 address, and 8 data lines) and even has a CPLD to translate the bus states into chip selects for the hardware. I have the original code (C/C++) ported with very few changes other than I had to use the GPIO to simulate the PC/104 bus.
I thought the Pi 3 should be able to keep up with, if not surpass, this board's performance. The original app is designed as an embedded app for a PC/104-based CPU card containing a Vortex86DX-800MHz processor. I have a Windows CE 5 application that I have ported to the Pi 3 with Windows 10 IoT Core Build 14342.
0 kommentar(er)
