bpftime

BPFtime Uprobe Benchmark Results

Generated on 2025-11-13 06:37:28

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	2312.60	365.00	6.34x
__bench_uretprobe	3014.36	366.24	8.23x
__bench_uprobe_uretprobe	3140.24	369.54	8.50x

Kernel vs Userspace eBPF Detailed Comparison

Operation	Environment	Min (ns)	Max (ns)	Avg (ns)	Std Dev
__bench_array_map_delete	Kernel	9494.08	9509.74	9500.39	6.75
__bench_array_map_delete	Userspace	6283.72	6294.01	6287.48	4.63
__bench_array_map_lookup	Kernel	3251.78	3273.53	3262.76	8.88
__bench_array_map_lookup	Userspace	5971.02	5977.85	5975.51	3.17
__bench_array_map_update	Kernel	44390.99	44446.37	44413.07	23.96
__bench_array_map_update	Userspace	9703.63	9826.67	9748.66	55.38
__bench_hash_map_delete	Kernel	55269.78	55534.86	55384.64	111.07
__bench_hash_map_delete	Userspace	11894.09	11910.10	11901.39	6.61
__bench_hash_map_lookup	Kernel	42036.86	43080.07	42549.68	426.07
__bench_hash_map_lookup	Userspace	21344.20	22728.06	22239.85	634.18
__bench_hash_map_update	Kernel	145013.89	145174.74	145116.81	72.97
__bench_hash_map_update	Userspace	25324.22	26827.54	26307.81	695.88
__bench_per_cpu_array_map_delete	Kernel	9483.65	9538.97	9507.48	23.22
__bench_per_cpu_array_map_delete	Userspace	6284.81	6292.30	6288.33	3.07
__bench_per_cpu_array_map_lookup	Kernel	3276.85	3537.54	3366.28	121.14
__bench_per_cpu_array_map_lookup	Userspace	11895.87	12289.69	12029.90	183.73
__bench_per_cpu_array_map_update	Kernel	44578.45	44595.52	44588.23	7.18
__bench_per_cpu_array_map_update	Userspace	28756.10	28816.77	28788.92	25.01
__bench_per_cpu_hash_map_delete	Kernel	55316.23	55870.35	55503.64	259.32
__bench_per_cpu_hash_map_delete	Userspace	78242.26	104157.80	88175.45	11411.92
__bench_per_cpu_hash_map_lookup	Kernel	41851.57	43312.60	42464.30	619.28
__bench_per_cpu_hash_map_lookup	Userspace	71722.37	71882.07	71826.82	73.90
__bench_per_cpu_hash_map_update	Kernel	110226.71	110743.18	110426.71	226.36
__bench_per_cpu_hash_map_update	Userspace	112692.72	114231.12	113348.65	648.15
__bench_read	Kernel	77243.49	78788.30	78242.88	707.66
__bench_read	Userspace	4409.68	4421.25	4415.07	4.75
__bench_uprobe	Kernel	2295.39	2333.23	2312.60	15.63
__bench_uprobe	Userspace	362.21	370.29	365.00	3.75
__bench_uprobe_uretprobe	Kernel	3128.49	3148.33	3140.24	8.51
__bench_uprobe_uretprobe	Userspace	366.74	374.97	369.54	3.84
__bench_uretprobe	Kernel	2999.21	3028.43	3014.36	11.95
__bench_uretprobe	Userspace	363.36	371.96	366.24	4.04
__bench_write	Kernel	78354.07	78869.33	78527.77	241.54
__bench_write	Userspace	4411.88	4427.25	4418.91	6.34

Embedded VM Performance

Operation	Min (ns)	Max (ns)	Avg (ns)	Std Dev
embed	97.10	101.07	98.67	1.72

Benchmark Metadata

• Number of runs: 3
• Timestamp: 2025-11-13 06:37:28
• Total duration: 354.78 seconds

This site is open source. Improve this page.