bpftime

BPFtime Uprobe Benchmark Results

Generated on 2025-10-07 10:59:19

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	2286.48	362.64	6.31x
__bench_uretprobe	3009.91	364.84	8.25x
__bench_uprobe_uretprobe	3156.01	364.80	8.65x

Kernel vs Userspace eBPF Detailed Comparison

Operation	Environment	Min (ns)	Max (ns)	Avg (ns)	Std Dev
__bench_array_map_delete	Kernel	9453.36	9458.29	9455.67	2.02
__bench_array_map_delete	Userspace	6287.58	6305.47	6296.98	7.33
__bench_array_map_lookup	Kernel	3522.24	3531.54	3526.22	3.91
__bench_array_map_lookup	Userspace	5962.79	5972.15	5968.29	3.99
__bench_array_map_update	Kernel	44349.74	44650.82	44456.14	137.86
__bench_array_map_update	Userspace	9701.83	9894.10	9767.74	89.37
__bench_hash_map_delete	Kernel	55273.78	55615.37	55416.78	144.89
__bench_hash_map_delete	Userspace	11584.44	11590.04	11586.45	2.55
__bench_hash_map_lookup	Kernel	42016.74	42516.27	42261.81	204.04
__bench_hash_map_lookup	Userspace	21124.03	22857.07	22151.26	743.12
__bench_hash_map_update	Kernel	145729.38	146200.22	145974.79	192.74
__bench_hash_map_update	Userspace	25114.36	27035.94	26236.22	816.88
__bench_per_cpu_array_map_delete	Kernel	9439.46	9518.86	9472.44	33.78
__bench_per_cpu_array_map_delete	Userspace	6276.62	6289.35	6282.80	5.20
__bench_per_cpu_array_map_lookup	Kernel	3229.38	3276.57	3248.14	20.44
__bench_per_cpu_array_map_lookup	Userspace	11581.24	12090.51	11755.67	236.84
__bench_per_cpu_array_map_update	Kernel	44674.09	44730.55	44710.23	25.62
__bench_per_cpu_array_map_update	Userspace	28122.67	28169.00	28152.94	21.41
__bench_per_cpu_hash_map_delete	Kernel	55265.41	55568.05	55367.33	141.94
__bench_per_cpu_hash_map_delete	Userspace	85455.09	115904.44	98611.97	12770.23
__bench_per_cpu_hash_map_lookup	Kernel	41954.46	42418.45	42117.57	213.00
__bench_per_cpu_hash_map_lookup	Userspace	72942.13	73790.98	73281.01	367.05
__bench_per_cpu_hash_map_update	Kernel	110038.28	110066.17	110054.63	11.88
__bench_per_cpu_hash_map_update	Userspace	113727.13	116894.95	114879.59	1429.97
__bench_read	Kernel	78651.05	78713.95	78681.12	25.75
__bench_read	Userspace	4409.65	4569.31	4464.31	74.27
__bench_uprobe	Kernel	2260.73	2311.39	2286.48	20.69
__bench_uprobe	Userspace	361.66	363.52	362.64	0.76
__bench_uprobe_uretprobe	Kernel	3123.74	3183.66	3156.01	24.68
__bench_uprobe_uretprobe	Userspace	362.24	367.13	364.80	2.01
__bench_uretprobe	Kernel	2983.17	3035.68	3009.91	21.45
__bench_uretprobe	Userspace	362.95	365.93	364.84	1.34
__bench_write	Kernel	78328.18	78601.34	78424.11	125.46
__bench_write	Userspace	4409.60	4412.31	4411.14	1.14

Embedded VM Performance

Operation	Min (ns)	Max (ns)	Avg (ns)	Std Dev
embed	96.58	101.61	98.92	2.07

Benchmark Metadata

• Number of runs: 3
• Timestamp: 2025-10-07 10:59:19
• Total duration: 358.43 seconds

This site is open source. Improve this page.