OV - - Loops

MAQAO

options

Loops Index

28 loops have been discarded from the report because their ratio ((Max Inclusive Time Over Threads * 100) / Max Thread Active Time) is lower than the threshold set by object_coverage_threshold (0.1%). It represents about 0.02% of the application. To include them, change the value of object_coverage_threshold in the experiment directory configuration file, then rerun the command with the additionnal parameter --force-static-analysis

▶Filters

Loop id	Source Location	Source Function	Level	Max Thread Time / Walltime gcc_3 (%)	Exclusive Coverage gcc_3 (%)	Inclusive Coverage gcc_3 (%)	Max Exclusive Time Over Threads gcc_3 (s)	Max Inclusive Time Over Threads gcc_3 (s)	Exclusive Time w.r.t. Wall Time gcc_3 (s)	Inclusive Time w.r.t. Wall Time gcc_3 (s)	Nb Threads gcc_3	GFLOPS gcc_3	Vectorization Ratio (%)	Vector Length Use (%)	Speedup If No Scalar Integer	Speedup If FP Vectorized	Speedup If Fully Vectorized	Speedup If Perfect Load Balancing gcc_3	Stride 0	Stride 1	Stride n	Stride Unknown	Stride Indirect	Array Access Efficiency
1832	libggml-cpu.so - sgemm.cpp:399-399 [...]	void (anonymous namespace)::tinyBLAS<16, float __vector(16), float __vector(16), unsigned short, unsigned short, float>::gemm_bloc<4, 6>(long, long)	Single	0.11	1.19	1.19	0.16	0.16	0.09	0.09	192	2406.29	100	85.29	1	1	1.29	1.78	0	10	0	0	0	100.00
1390	libggml-cpu.so - vec.h:89-89	ggml_compute_forward_soft_max	Innermost	0.06	0.48	0.48	0.09	0.09	0.04	0.04	192	12.03	100	50	1	1	2	2.33	0	2	0	0	0	100.00
1370	libggml-cpu.so - ops.cpp:5552-5563	ggml_compute_forward_set_rows	Innermost	0.04	0.35	0.35	0.06	0.06	0.03	0.03	192	2.02	0	12.5	1	1	8	2.3	2	0	0	4	2	50.00
983	libggml-cpu.so - ops.cpp:6210-6245 [...]	ggml_compute_forward_rope_f32(ggml_compute_params const, ggml_tensor, bool)	Innermost	0.04	0.27	0.27	0.06	0.06	0.02	0.02	192	786.09	7.55	7.84	1.17	1.56	8.17	2.93	NA	NA	NA	NA	NA	0.00
66	libggml-cpu.so - ggml-cpu.c:1291-1297	ggml_compute_forward_mul_mat	Innermost	0.04	0.24	0.24	0.05	0.05	0.02	0.02	191	20.09	0	11.51	1	1	2	2.99	2	0	0	0	0	100.00
1303	libggml-cpu.so - ops.cpp:4325-4326	ggml_compute_forward_rms_norm	Innermost	0.04	0.23	0.23	0.05	0.05	0.02	0.02	192	48.62	100	39.29	1	1	2.59	3.13	0	0	2	0	0	75.00
1389	libggml-cpu.so - vec.h:677-677 [...]	ggml_compute_forward_soft_max	Innermost	0.04	0.23	0.23	0.05	0.05	0.02	0.02	184	36.66	100	100	1	1	1	3.1	1	1	0	0	0	100.00
951	libggml-cpu.so - vec.cpp:372-372 [...]	ggml_vec_swiglu_f32	Single	0.03	0.18	0.18	0.05	0.05	0.01	0.01	171	1406.99	100	100	1.02	1	1	3.33	0	3	0	0	0	100.00
703	libggml-cpu.so - binary-ops.cpp:18-32 [...]	ggml_compute_forward_mul	Innermost	0.03	0.15	0.15	0.04	0.04	0.01	0.01	165	41.52	100	50	1	1.06	2	3.03	1	3	0	0	0	100.00
619	libggml-cpu.so - binary-ops.cpp:10-32 [...]	ggml_compute_forward_add_non_quantized	Innermost	0.03	0.15	0.15	0.04	0.04	0.01	0.01	164	38.53	100	50	1	1.06	2	3.04	1	3	0	0	0	100.00
1831	libggml-cpu.so - sgemm.cpp:425-427 [...]	void (anonymous namespace)::tinyBLAS<16, float __vector(16), float __vector(16), unsigned short, unsigned short, float>::gemm_bloc<4, 6>(long, long)	Single	0.02	0.15	0.15	0.04	0.04	0.01	0.01	192	2769.41	90.48	34.23	1	1.24	3.33	3.18	0	1	0	1	0	75.00
952	libggml-cpu.so - vec.cpp:411-415 [...]	ggml_vec_soft_max_f32	Single	0.02	0.12	0.12	0.04	0.04	0.01	0.01	192	2305.39	90.47	73.78	1.05	1	1.17	3.98	0	0	0	2	0	50.00
446	libggml-cpu.so - mmq.cpp:822-1392 [...]	void parallel_for<(anonymous namespace)::convert_B_packed_format<block_q8_0, 32>(void, block_q8_0 const, int, int)::{lambda(int, int)#1}>(int, (anonymous namespace)::convert_B_packed_format<block_q8_0, 32>(void, block_q8_0 const, int,...	Innermost	0.03	0.11	0.11	0.04	0.04	0.01	0.01	139	0.00	52.38	47.77	1	1	1.11	3.87	2	0	0	20	0	54.55
2203	libggml-cpu.so - quants.c:298-321 [...]	quantize_row_q8_0	Single	1.04	0.11	0.11	1.57	1.57	0.01	0.01	1	643.58	59.65	29.28	1.03	1.4	2.86	1	NA	NA	NA	NA	NA	0.00
1379	libggml-cpu.so - vec.h:1444-1445	ggml_compute_forward_soft_max	Innermost	0.02	0.11	0.11	0.03	0.03	0.01	0.01	192	290.58	0	6.25	1	1	16	3.07	1	0	2	0	0	83.33

×