Efficient Parallel Video Processing Techniques on GPU: From Framework to Implementation

Table 5

Performance comparison between the proposed parallel H.264 encoder and other implementations.

Platform	Reference code	Target resolution	Optimized module	Speedup ratio	Performance (fps)
CPU (i7-2600) original	x264	720 p	NA	1	1.05 (for application)
CPU (i7-2600) optimized	x264	720 p	Key function	3∼5	3∼5.5 (for application)
GTX280 [10]	x264	720 p	ME	NA	15.5 (for ME)
Geforce 8800 [23]	x264	720 p	Intracoding	2∼3	NA
AsAP [25]	x264	720 p	CAVLC	4.86	36∼41.3 (for CAVLC)
GTX 240MFP [24]	x264	1080 p	Deblocking filter	10.2	1309 (for deblocking filter)
GeForce 9800 [3]	JSVM	CIF	ME + Intra	6.7	1.02 (for application)
GTX260 The proposed MRMW	x264	720 p	ME	12∼14	50 (for ME)
GTX260 The proposed Intra Coding	x264	720 p	Intracoding	4∼6.8	21 (for Intracoding)
GTX260 Component-based CAVLC	x264	720 p	CAVLC	8	105 (for CAVLC)
GTX260 Direction-priority DB.	x264	720 p	Deblocking filter	9	1050 (for deblocking filter)
C2050 The proposed H.264	x264	720 p	Application	13∼17	32.3 (for application)