Performance

The performance benchmark data and corelative commits are listed here.

See also: Performance for x86/x64 Test Guide.
See also: Performance for RaspberryPi.
For multiple processes performance, read #775: REUSEPORT or OriginCluster(CN/EN).
For RTC benchmark, please use srs-bench.

Play RTMP benchmark

The data for playing RTMP was benchmarked by SB:

Update	SRS	Clients	Type	CPU	Memory	Commit
2014-12-07	2.0.67	10k(10000)	players	95%	656MB	code
2014-12-05	2.0.57	9.0k(9000)	players	90%	468MB	code
2014-12-05	2.0.55	8.0k(8000)	players	89%	360MB	code
2014-11-22	2.0.30	7.5k(7500)	players	87%	320MB	code
2014-11-13	2.0.15	6.0k(6000)	players	82%	203MB	code
2014-11-12	2.0.14	3.5k(3500)	players	95%	78MB	code
2014-11-12	2.0.14	2.7k(2700)	players	69%	59MB	-
2014-11-11	2.0.12	2.7k(2700)	players	85%	66MB	-
2014-11-11	1.0.5	2.7k(2700)	players	85%	66MB	-
2014-07-12	0.9.156	2.7k(2700)	players	89%	61MB	code
2014-07-12	0.9.156	1.8k(1800)	players	68%	38MB	-
2013-11-28	0.5.0	1.8k(1800)	players	90%	41M	-

Update	SFU	Clients	Type	CPU	Memory	Threads	VM
2021-05-11	SRS/v4.0.105	4000	players	~94% x1	419MB	1	G5 8CPU
2021-05-11	NginxRTMP/v1.2.1	2400	players	~92% x1	173MB	1	G5 8CPU

Note: CentOS7, 600Kbps, ECS/G5-2.5GHZ(SkyLake), SRS/v4.0.105, NginxRTMP/v1.2.1.

Publish RTMP benchmark

The data for publishing RTMP was benchmarked by SB:

Update	SRS	Clients	Type	CPU	Memory	Commit
2014-12-04	2.0.52	4.0k(4000)	publishers	80%	331MB	code
2014-12-04	2.0.51	2.5k(2500)	publishers	91%	259MB	code
2014-12-04	2.0.49	2.5k(2500)	publishers	95%	404MB	code
2014-12-04	2.0.49	1.4k(1400)	publishers	68%	144MB	-
2014-12-03	2.0.48	1.4k(1400)	publishers	95%	140MB	code
2014-12-03	2.0.47	1.4k(1400)	publishers	95%	140MB	-
2014-12-03	2.0.47	1.2k(1200)	publishers	84%	76MB	code
2014-12-03	2.0.12	1.2k(1200)	publishers	96%	43MB	-
2014-12-03	1.0.10	1.2k(1200)	publishers	96%	43MB	-

Update	SFU	Clients	Type	CPU	Memory	Threads	VM
2021-05-11	SRS/v4.0.105	2300	publishers	~89% x1	1.1GB	1	G5 8CPU
2021-05-11	NginxRTMP/v1.2.1	1300	publishers	~84% x1	198MB	1	G5 8CPU

Note: CentOS7, 600Kbps, ECS/G5-2.5GHZ(SkyLake), SRS/v4.0.105, NginxRTMP/v1.2.1.

Play HTTP FLV benchmark

The data for playing HTTP FLV was benchmarked by SB:

Update	SRS	Clients	Type	CPU	Memory	Commit
2014-05-25	2.0.171	6.0k(6000)	players	84%	297MB	code
2014-05-24	2.0.170	3.0k(3000)	players	89%	96MB	code
2014-05-24	2.0.169	3.0k(3000)	players	94%	188MB	code
2014-05-24	2.0.168	2.3k(2300)	players	92%	276MB	code
2014-05-24	2.0.167	1.0k(1000)	players	82%	86MB	-

RTC benchmark

The RTC benchmark data, by srs-bench:

Update	SFU	Clients	Type	CPU	Memory	Threads	VM
2021-05-10	SRS/v4.0.105	2000	players	~94% x1	462MB	1	G7 2CPU
2021-05-10	SRS/v4.0.105	1000	players	~90% x1	180MB	1	G5 2CPU
2021-03-31	SRS/v4.0.87	800	players	~94% x1	444MB	1	G5 2CPU
2021-05-10	Janus/v0.11.1	700	players	~93% x2	430MB	24	G5 2CPU
2021-05-10	SRS/v4.0.105	1700	publishers	~92% x1	334MB	1	G7 2CPU
2021-05-10	SRS/v4.0.105	950	publishers	~92% x1	132MB	1	G5 2CPU
2021-03-31	SRS/v4.0.87	550	publishers	~86% x1	1.3GB	1	G5 2CPU
2021-05-10	Janus/v0.11.1	350	publishers	~93% x2	405MB	23	G5 2CPU

Note: CentOS7, 600Kbps, ECS/G5-2.5GHZ(SkyLake)/G7-2.7GHZ(IceLake), SRS/v4.0.87, SRS/v4.0.105, Janus/v0.11.1.

Latency benchmark

The latency between encoder and player with realtime config(CN, EN):

Update	SRS	Protocol	VP6	H.264	VP6+MP3	H.264+MP3
2014-12-16	2.0.72	RTMP	0.1s	0.4s	0.8s	0.6s
2014-12-12	2.0.70	RTMP	0.1s	0.4s	1.0s	0.9s
2014-12-03	1.0.10	RTMP	0.4s	0.4s	0.9s	1.2s
2021-04-02	4.0.87	WebRTC	x	80ms	x	x

2018-08-05, c45f72e, Refine HTTP-FLV latency, support realtime mode. 2.0.252

We used FMLE as encoder for benchmark. The latency of server was 0.1s+, and the bottleneck was the encoder. For more information, read bug #257.

Winlin 2021