2014 a positive year for the IZO-SGI Computing Service
2014 year report
The Computing Service IZO-SGI SGIker of the UPV/EHU has published the 2014 year report, which can be found in the next link:
IZO-SGI 2014 year report (pdf in spanish)
The Computing Service has maintained or increased its main quality indicators as can be observed in the table below. Among the positive aspects, the increase in the use of the resources and the increase of the HPC resources that where available at the end of the year. In the negative part, we can mention the reduction of the scientific publications . The Service continues having an excellent score in the satisfaction poll between the researches.
Acknowledgments
We would like to acknowledgment to all the researchers that have relied on the service, most of the good results of the Service are in fact their merits and not ours.
2014 Relevant posts in our Blog
The IZO-SGI in numbers
In the table below, the main figures of the Service in the last years is summarized and in the graph at the very bottom the use of Arina is shown.
| 2009 | 2010 | 2011 | 2012 | 2013 | 2014 | |
| Computing cores | 440 | 872 | 1.520 | 1.520 | 1.400 | 1.3008 |
| Consumed millions of hours | 2,11 | 2,26 | 5,76 | 11,27 | 10,4 | 10,1 |
| Active researches | 101 | 98 | 90 | 89 | 99 | 107 |
| Active groups | 53 | 40 | 39 | 42 | 47 | 45 |
| New accounts | 49 | 33 | 33 | 20 | 34 | 23 |
| Satisfaction of the researches¹ | 8.9 | 8.8 | 8.8 | 9,3 | 9,3 | 9,6 |
| Scientific articles² | 44 | 58 | 57 | 74 | 87 | 75 |
| Web visits | 4.925 | 5.957 | 8.022 | 9.899 | 5.924 | 2.340 |
| View pages | 58.731 | 40.987 | 30.042 | 30.738 | 20.323 | 12.057 |
| Posts in the HPC blog | – | 29 | 34 | 27 | 47 | 36 |
| Visits of the HPC blog | – | 1.431 | 3.818 | 4.741 | 24.613 | 23.677 |
| Arina | ||||||
| Computing cores | 320 | 752 | 1.360 | 1.360 | 1.320 | 1.300 |
| Consumed millions of hours | 2,07 | 2,11 | 5,20 | 9,6 | 9,8 | 10,1 |
| Mean occupation | 75% | 47%³ | 74% | 79% | 83% | 87% |
| Submitted jobs | 38.497 | 65.179 | 68.991 | 98.383 | 100.214 | 76.912 |
| More than 2 minutes jobs⁴ | 28.377 | 50.624 | 51.735 | 78.846 | 75.406 | 64.335 |
| Job mean computing time⁵ | 76 hours | 44 hours | 101 hours | 122 hours | 130 hours | 156 hours |
| Job mean waiting time⁶ | 5,7 hours | 3,8 hours | 3,8 hours | 5,2 hours | 5,5 hours | 10 hours |
| Péndulo | ||||||
| Computing cores | 120 | 120 | 120 | 80 | 80 | 70 |
| Consumed millions of hours | 0,02 | 0,15 | 0,09 | 0,07 | 0,04 | 0,006 |
| Ikerbasque | ||||||
| Computing cores | – | – | 208 | 208 | 208 | – |
| Consumed millions of hours | – | – | 0,47 | 1,6 | 0,55 | – |
¹ Satisfaction pool by the SGIker Quality Unit.
² Where the IZO-SGI is acknowledged.
³ The use in 2010 is apparently low because new machines where installed passing form 300 cores to 752 cores in July. In August and September Arina was quite empty and the weigh of these months has had a lot of importance in the mean value. In fact in November the mean use was already in the 76%.
⁴ Jobs with less than 2 minutes of computing time are usually failed jobs. In any case due to the short duration they don’t have relevance in the usage.
⁵ The jobs with less than 2 minutes of computing time are not taken into account.
⁶ The jobs are executed through a queue system that assign proper resources to the jobs. The waiting time is the time the jobs spend in the queue waiting for the resources to be available.
⁷ The machine was switch off in April.
8 At the end of the year we got 652 new cores.
Use of Arina between 2013 December and 2014 November.
Comentarios