Performance Analysis for Subroutine CNDF ; Time used=8:54

COMPLETED

55% MM

25% TR

12% DIA

Times and Problem

Job Started	0	14:27:16 Thu Apr 27 2017	Equations	100000	Compute Rate (Gflops)	5448
Routine Started	112	14:29:08 Thu Apr 27 2017	Vectors	0	Read Rate (MBytes/Sec.)	0
Current Time	646	14:38:03 Thu Apr 27 2017	Block Size	25088 X 25088	Disk->Memory Reuse	50176
Estimated Completion	646	14:38:03 Thu Apr 27 2017	Data Type	Complex	Disk Speed / Compute Speed	0.00

Total Processor Performance
(64-bit floating point operations per second)

Tera				Giga				Mega				Kilo				Flops
		5		4	4	8		2	4	9		8	5	3		5	7	4

Component Performance (Gflops)

Routine	All	32 CPUs	GPU 1 44 °C	GPU 2 35 °C	GPU 3 52 °C	GPU 4 39 °C	GPU 5 35 °C	GPU 6 49 °C	GPU 7 40 °C	GPU 8 55 °C
Matrix Multiply	5968	0	772	783	761	765	752	748	748	759
CPU(0%)
Triangle Solve	5568	0	722	732	708	711	699	696	701	711
CPU(0%)
Diagonal Factor	2709	10	394	406	398	401	395	397	400	384
GPU model = Tesla K80

Time (Sec.)

Routine	All	CPUs	GPU 1	GPU 2	GPU 3	GPU 4	GPU 5	GPU 6	GPU 7	GPU 8
Matrix Multiply	293	0	283	279	287	286	290	292	292	288
Triangle Solve	135	0	130	129	133	132	135	135	134	132
Diagonal Factor	62	5	44	45	48	51	54	57	60	61
Total Compute	489	5	457	452	468	469	479	484	486	481
Assembly	2
I/O wait	0
HTML	4
Other	41
Overall	534