bpftime

BPFtime Uprobe Benchmark Results

Generated on 2025-10-27 18:42:42

Environment

• OS: Linux 6.11.0-1018-azure
• CPU: AMD EPYC 7763 64-Core Processor (2 cores, 4 threads)
• Memory: 15.62 GB
• Python: 3.12.3

Summary

This benchmark compares three different eBPF execution environments:

• Kernel Uprobe: Traditional kernel-based eBPF uprobes
• Userspace Uprobe: BPFtime’s userspace implementation of uprobes
• Embedded VM: BPFtime’s embedded eBPF VM

Times shown in nanoseconds (ns) - lower is better

Core Uprobe Performance Summary

Operation	Kernel Uprobe	Userspace Uprobe	Speedup
__bench_uprobe	2331.72	366.06	6.37x
__bench_uretprobe	3079.14	366.56	8.40x
__bench_uprobe_uretprobe	3247.72	367.86	8.83x

Kernel vs Userspace eBPF Detailed Comparison

Operation	Environment	Min (ns)	Max (ns)	Avg (ns)	Std Dev
__bench_array_map_delete	Kernel	9524.11	9541.79	9531.93	7.36
__bench_array_map_delete	Userspace	6284.30	6290.33	6287.79	2.55
__bench_array_map_lookup	Kernel	3268.74	3326.11	3295.50	23.58
__bench_array_map_lookup	Userspace	5973.09	6006.62	5984.33	15.76
__bench_array_map_update	Kernel	44338.72	44469.09	44395.19	54.63
__bench_array_map_update	Userspace	9707.71	10057.08	9835.58	157.25
__bench_hash_map_delete	Kernel	56012.71	56301.56	56191.91	127.77
__bench_hash_map_delete	Userspace	11892.68	11903.38	11898.37	4.40
__bench_hash_map_lookup	Kernel	42747.44	42950.95	42856.00	83.64
__bench_hash_map_lookup	Userspace	21539.36	23047.81	22513.33	689.79
__bench_hash_map_update	Kernel	147160.87	147202.38	147176.28	18.55
__bench_hash_map_update	Userspace	25579.23	27232.98	26612.98	735.81
__bench_per_cpu_array_map_delete	Kernel	9527.68	9552.14	9542.15	10.47
__bench_per_cpu_array_map_delete	Userspace	6284.11	6291.83	6287.07	3.40
__bench_per_cpu_array_map_lookup	Kernel	3309.42	3356.96	3339.07	21.11
__bench_per_cpu_array_map_lookup	Userspace	11893.35	12174.82	11988.02	132.09
__bench_per_cpu_array_map_update	Kernel	44744.96	44929.23	44837.98	75.24
__bench_per_cpu_array_map_update	Userspace	28735.08	28762.98	28753.46	12.99
__bench_per_cpu_hash_map_delete	Kernel	55311.36	56280.12	55635.98	455.47
__bench_per_cpu_hash_map_delete	Userspace	114027.23	114565.86	114372.13	244.50
__bench_per_cpu_hash_map_lookup	Kernel	41956.75	43135.44	42551.94	481.27
__bench_per_cpu_hash_map_lookup	Userspace	71786.74	71938.77	71840.77	69.42
__bench_per_cpu_hash_map_update	Kernel	110290.13	111480.62	110772.81	511.42
__bench_per_cpu_hash_map_update	Userspace	114216.49	135335.41	122071.83	9431.82
__bench_read	Kernel	78713.30	79447.12	79098.71	300.72
__bench_read	Userspace	4414.68	4417.54	4416.02	1.18
__bench_uprobe	Kernel	2312.84	2346.00	2331.72	13.92
__bench_uprobe	Userspace	364.64	367.18	366.06	1.06
__bench_uprobe_uretprobe	Kernel	3196.86	3341.00	3247.72	66.05
__bench_uprobe_uretprobe	Userspace	367.25	368.17	367.86	0.43
__bench_uretprobe	Kernel	3041.68	3130.77	3079.14	37.73
__bench_uretprobe	Userspace	364.49	369.35	366.56	2.05
__bench_write	Kernel	79050.26	79937.57	79350.66	415.05
__bench_write	Userspace	4415.28	4416.45	4416.01	0.52

Embedded VM Performance

Operation	Min (ns)	Max (ns)	Avg (ns)	Std Dev
embed	95.80	201.10	130.96	49.60

Benchmark Metadata

• Number of runs: 3
• Timestamp: 2025-10-27 18:42:42
• Total duration: 367.28 seconds

This site is open source. Improve this page.