On 10.04.19 21:26, cagnulein via Xenomai wrote:
After 3 days of test (latency + windows 10 with cpu-z and stress test) I reached rare peaks of 130 us. For me it's OK but I think there is a small "regression" against the 4.9 Do you need any other test?
What is the CPU / SoC here? Are we talking about some Atom or rather some Xeon? If you want us to understand that peak, and if it was just less likely on 4.9, we would need an ipipe trace from that event. The latency test can trigger a freeze. We likely needs around 1000 back_trace_points for that. BTW, as mixing virtualization workload with RT won't make both faster and may have bad side effects, also on both, isolating them on separate cores should be preferred whenever possible. Jan
