bpftime

BPFtime Uprobe Benchmark Results

Generated on 2026-01-26 07:31:46

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	2269.42	361.17	6.28x
__bench_uretprobe	2979.27	363.74	8.19x
__bench_uprobe_uretprobe	3153.18	365.43	8.63x

Kernel vs Userspace eBPF Detailed Comparison

Operation	Environment	Min (ns)	Max (ns)	Avg (ns)	Std Dev
__bench_array_map_delete	Kernel	9392.52	9406.43	9400.08	5.74
__bench_array_map_delete	Userspace	6599.98	6601.46	6600.58	0.64
__bench_array_map_lookup	Kernel	3193.12	3206.52	3198.63	5.72
__bench_array_map_lookup	Userspace	5969.05	5974.00	5970.80	2.26
__bench_array_map_update	Kernel	44228.52	44252.01	44241.60	9.77
__bench_array_map_update	Userspace	9759.72	11215.66	10279.49	663.32
__bench_hash_map_delete	Kernel	54673.04	55183.77	55005.63	235.38
__bench_hash_map_delete	Userspace	11897.60	12138.33	11981.29	111.13
__bench_hash_map_lookup	Kernel	41582.05	41588.47	41584.60	2.78
__bench_hash_map_lookup	Userspace	22099.51	23125.15	22578.07	421.51
__bench_hash_map_update	Kernel	147542.03	154888.20	150173.25	3341.46
__bench_hash_map_update	Userspace	25208.95	25498.13	25359.98	118.41
__bench_per_cpu_array_map_delete	Kernel	9396.30	9412.45	9402.25	7.24
__bench_per_cpu_array_map_delete	Userspace	6597.34	6607.82	6601.27	4.66
__bench_per_cpu_array_map_lookup	Kernel	3496.31	3528.85	3509.91	13.81
__bench_per_cpu_array_map_lookup	Userspace	11892.79	13561.81	12455.32	782.45
__bench_per_cpu_array_map_update	Kernel	44600.67	44670.62	44637.57	28.69
__bench_per_cpu_array_map_update	Userspace	28438.89	28659.94	28513.33	103.67
__bench_per_cpu_hash_map_delete	Kernel	54927.53	55219.24	55110.36	130.07
__bench_per_cpu_hash_map_delete	Userspace	103901.54	113846.95	110174.08	4456.94
__bench_per_cpu_hash_map_lookup	Kernel	41582.47	41669.49	41615.93	38.26
__bench_per_cpu_hash_map_lookup	Userspace	70925.04	71280.85	71147.56	158.37
__bench_per_cpu_hash_map_update	Kernel	108842.63	110721.91	109848.01	772.82
__bench_per_cpu_hash_map_update	Userspace	113403.77	114665.33	114009.40	516.26
__bench_read	Kernel	77084.89	78752.96	77721.95	735.76
__bench_read	Userspace	4718.60	4722.06	4720.06	1.46
__bench_uprobe	Kernel	2259.05	2285.08	2269.42	11.26
__bench_uprobe	Userspace	359.68	363.14	361.17	1.45
__bench_uprobe_uretprobe	Kernel	3098.04	3221.24	3153.18	51.12
__bench_uprobe_uretprobe	Userspace	364.95	365.68	365.43	0.34
__bench_uretprobe	Kernel	2956.86	2991.53	2979.27	15.87
__bench_uretprobe	Userspace	363.02	364.53	363.74	0.62
__bench_write	Kernel	78237.37	78259.28	78247.79	8.98
__bench_write	Userspace	4567.45	4728.21	4672.53	74.35

Embedded VM Performance

Operation	Min (ns)	Max (ns)	Avg (ns)	Std Dev
embed	99.71	101.49	100.85	0.81

Benchmark Metadata

• Number of runs: 3
• Timestamp: 2026-01-26 07:31:46
• Total duration: 362.13 seconds

This site is open source. Improve this page.