8.3. Runtime Instrumentation Testing - N1SDP

For this test we used the N1 System Development Platform (N1SDP), which contains an SoC consisting of two dual-core Arm N1 clusters.

The following source trees were used:

• TF-A v2.13-rc0

• TFTF v2.13-rc0

Please see the Runtime Instrumentation Testing Methodology page for more details. The tests were ran using the tf-psci-lava-instr/n1sdp-runtime-instrumentation,n1sdp-runtime-instrumentation:n1sdp-fip.tftf-firmware configuration in CI.

8.3.1. Results

8.3.1.1. CPU_SUSPEND to deepest power level

CPU_SUSPEND latencies (µs) to deepest power level in parallel (v2.13)

Cluster	Core	Powerdown	Wakeup	Cache Flush
0	0	3.32	13.42 (-45.40%)	0.28 (-69.57%)
0	0	4.02 (-35.78%)	18.14 (-52.88%)	0.28
1	0	2.7 (-27.42%)	17.38 (-49.36%)	0.26
1	0	2.96	10.86 (-73.64%)	0.26

CPU_SUSPEND latencies (µs) to deepest power level in parallel (v2.12)

Cluster	Core	Powerdown	Wakeup	Cache Flush
0	0	2.58	24.14	0.28 (-69.57%)
0	0	4.24 (-32.27%)	40.1	0.3
1	0	3.58	35.54	0.28
1	0	3.28	42.36	0.3

CPU_SUSPEND latencies (µs) to deepest power level in serial (v2.13)

Cluster	Core	Powerdown	Wakeup	Cache Flush
0	0	1.62	10.14 (-58.10%)	0.3
0	0	1.86	10.62 (-56.44%)	0.28 (-26.32%)
1	0	1.8	10.16 (-57.84%)	0.32
1	0	2.16	10.6 (-56.84%)	0.5 (+56.25%)

CPU_SUSPEND latencies (µs) to deepest power level in serial (v2.12)

Cluster	Core	Powerdown	Wakeup	Cache Flush
0	0	1.9	23.8	0.36
0	0	2.26	23.86	0.34
1	0	2.02	23.4	0.36
1	0	2.24	23.84	0.36

8.3.1.2. CPU_SUSPEND to power level 0

CPU_SUSPEND latencies (µs) to power level 0 in parallel (v2.13)

Cluster	Core	Powerdown	Wakeup	Cache Flush
0	0	0.8 (-57.45%)	11.98 (-61.75%)	0.26
0	0	1.54 (-30.63%)	16.44 (-53.74%)	0.26
1	0	1.62 (-30.77%)	16.1 (-53.92%)	0.3
1	0	1.8	10.54 (-55.30%)	0.2 (-33.33%)

CPU_SUSPEND latencies (µs) to power level 0 in parallel (v2.12)

Cluster	Core	Powerdown	Wakeup	Cache Flush
0	0	1.46	31.7	0.32
0	0	2.06	35.5	0.48 (+60.00%)
1	0	1.96	35.7	0.32
1	0	2.08	23.38	0.28

CPU_SUSPEND latencies (µs) to power level 0 in serial (v2.13)

Cluster	Core	Powerdown	Wakeup	Cache Flush
0	0	1.44	9.9 (-58.05%)	0.3
0	0	1.74 (-25.64%)	10.4 (-56.23%)	0.28 (-33.33%)
1	0	1.8	10.04 (-57.71%)	0.34
1	0	1.96	10.46 (-56.23%)	0.44

CPU_SUSPEND latencies (µs) to power level 0 in serial (v2.12)

Cluster	Core	Powerdown	Wakeup	Cache Flush
0	0	1.66	23.22	0.36
0	0	2.58	23.72	0.78 (+85.71%)
1	0	2.02	23.84	0.38
1	0	2.16	23.92	0.34

8.3.1.3. CPU_OFF on all non-lead CPUs

CPU_OFF on all non-lead CPUs in sequence then, CPU_SUSPEND on the lead core to the deepest power level.

CPU_OFF latencies (µs) on all non-lead CPUs (v2.13)

Cluster	Core	Powerdown	Wakeup	Cache Flush
0	0	1.64	10.24 (-57.72%)	0.3
0	0	13.92	17.7 (-43.74%)	0.3
1	0	13.54	16.74 (-44.90%)	0.3 (-37.50%)
1	0	14.12	18.28 (-41.93%)	0.3 (-44.44%)

CPU_OFF latencies (µs) on all non-lead CPUs (v2.12)

Cluster	Core	Powerdown	Wakeup	Cache Flush
0	0	1.84	23.82	0.36
0	0	14.18	31.78	0.56 (+86.67%)
1	0	13.64	30.54	0.36
1	0	14.18	31.82	0.68

8.3.1.4. CPU_VERSION in parallel

CPU_VERSION latency (µs) in parallel on all cores (v2.13)

Cluster	Core	Latency
0	0	0.12
0	0	0.2 (-28.57%)
1	0	0.2
1	0	0.24 (-25.00%)

CPU_VERSION latency (µs) in parallel on all cores (v2.12)

Cluster	Core	Latency
0	0	0.14
0	0	0.2 (-28.57%)
1	0	0.2
1	0	0.26

---

Copyright (c) 2023-2025, Arm Limited. All rights reserved.