Annual Report performance data collection

advertisement
TeraGrid Data Reporting by RP
Table 1.2. HPC Resource Utilization
DataStar p655
% Uptime
DataStar p690
Blue Gene Data
TeraGrid Cluster
96.6%
96.6%
97.3%
97.9%
291,764,386
19,929,848
80,520,141
89,738,313
Total NUs Available
Total NUs Delivered
DATASTAR P655
NU
Usage by Category (% of NUs Delivered)
250,479,076
Academic
Allocated
238,890,524
Discretionary
%
DATASTAR P690
NU
%
BLUEGENE DATA
TERAGRID CLUSTER
NU
%
NU
%
85.8
81.9
18,173,948
17,271,766
91.2
86.7
75,288,116
69,889,694
93.5
86.8
85,974,309
84,569,000
95.8
94.2
ASCI
11,588,552
23,312,407
4.0
8.0
902,182
11,817
4.5
0.1
5,398,422
135,949
6.7
0.2
1,405,309
0
1.6
0.0
AAP
SDSC Staff
4,827,391
12,817,311
1.7
4.4
1,667,373
61,768
8.4
0.3
551,037
4,489,241
0.7
5.6
940,097
2,117,973
1.0
2.4
0
328,201
0.0
0.1
11,650
3,292
0.1
0.0
55,317
481
0.1
0.0
700,367
5,567
0.8
0.0
TeraGrid Staff
Industry
RP-only data.
TeraGrid Data Reporting by RP
Table 1.3. Data Resource Utilization
GPFSWAN
SAMQFS
/SRB
% Uptime
96.0
96.5
Users
(Institutions)
38
(16)
~150
(~30)
33,510,4
21
11,404,8
60
14,50
6,137
8,609,
669
12,71
5,117
2,816,
524
0.34
0.60
0.22
RP-only data.
File Writes
Files Read
Read/Write Ratio
HPSS
98.6
>4,00
0
(>250
)
TeraGrid Data Reporting by RP
Table 1.1.4. SDSC Long-Term Storage Resource Summary for FY06
SAM-QFS
DISK
TB
TAPE
%
TB Available
188
TB Requested
405
TB Allocated
TB Stored
TB
GPFS-WAN
HPSS
DISK
TAPE
TOTAL
%
TB
TB
%
TB
SRB
%
TB
%
18,000
18,188
197
18,000
n/a
215.4
2,497
2,902
124
62.9
n/a
92
96
51.1
642
738
124
62.9
n/a
92
64
34.0
530
594
132
67.0
3,440
617
63.5
99.2
593.9
124.0
93.9
1,919
55.8
3.6
5.6
1,040
30.2
772
22.4
93
2.7
AAP
33
1.0
ASCI
18
0.5
Usage by Category (% of TB Stored)
Academic
User Data Archive / Other
530.4 100.0
29.3
5.5
33
Home Dirs / System Backups
Data Collections
Parking
SDSC PIs/Staff
60
93.8
501.1
0
0.0
0
< 0.01
0.0
< 0.01
94.5
0.0
561.1
104.9
79.4
0
19.1
14.5
426
12.4
Recharge/Industry
0
1
0.0
Dual-copy Backup
1,056
30.7
616.8 100.0
30.1
586.7 95.1
0
Definitions. SRB: Numbers reflect amount of data reported for both SAM-QFS and HPSS that is under SRB management. SAM-QFS
and HPSS share the same 18-PB tape archive. TB Requested, Allocated: TB requested via xRAC process or pre-xRAC allocation
process at SDSC. User Archive/Other: Traditional archiving associated with HPC users. Home Dirs/System Backups: SDSC-initiated
backups in support of HPC users. Data Collection: Collections stored with intent of sharing or for single-project use. Parking: Mediumterm storage for data-intensive compute projects that need data resident on disk longer than the standard purge window.
RP-only data.
4.9
0.0
TeraGrid Data Reporting by RP
Table 1.2.2. SDSC Long-Term Storage Resources, Academic Use by Discipline
SAM-QFS (SRB & UNIX)
USERS
TB
GPFS-WAN
%
USERS
Astrophysics
11
142.18 24.0
12
Earth Science
45
149.54 25.2
1
TB
HPSS (INCL. SRB)
%
USERS
90.21 73.4
0
0
SRB (SAM-QFS & HPSS)
TB
%
403
492.0
43.4
USERS
13
146.67 22.1
TB
%
316
99.9
8.8
48
165.61
25
Integ. Bio., Neurosci.
5
54.61
9.2
4
29 23.6
64
51.7
4.6
8
106.23
16
Physics
1
34.77
5.9
2
3.2
2.6
395
93.3
8.2
1
34.77
5.2
182
40.69
6.9
2
0
0
87
4.16
0.4
184
44.7
6.7
28
0
0.00
22
88.73 13.4
Cross-Disciplinary
Info., Intelligent Sys.
23
94.35 15.9
Chem., Thermal Sys.
4
0.02
0
625
89.8
7.9
Molec., Cellular Bio.
10
53.48
9
5
0.03
0
899
34.2
3.0
14
57.59
8.7
Atmospheric Science
3
13.18
2.2
1
0
0
523
60.1
5.3
3
6.63
1
3
0.35
0.3
819
150.8
13.3
275
23.6
2.1
92
6.51
0.6
6
4.93
0.7
776
7.8
0.7
502
7.0
0.6
100
5.1
0.5
469
5.0
0.4
110
0.1
0.0
1
4.37
0.7
5
2.41
0.4
Chemistry
Mech., Struct. Systems
Ocean Science
5
4.24
0.7
Materials Research
1
0
0
Computer Science
Bioeng., Environ. Sys.
2
0.08
0.1
Adv. Sci. Comp.
Mathematical Science
1
4.37
0.7
Education
4
1.39
0.2
1
0
0
19
1.3
0.1
Social, Econ. Science
22
0.1
0.0
Elec., Comm. Systems
159
0.02
0.0
Environ. Biology
34
0.02
0.0
Biol. Instrumentation
12
0.02
0.0
2
0.00
0.0
Behavioral Science
RP-only data.
TeraGrid Data Reporting by RP
Table 2.1.1. DataStar Queues
QUEUE
PURPOSE
MAX
HOURS
MAX % FY06
NODES
USE
p655 nodes
express
debugging, timing,
sole access
high
standard jobs,
higher priority
normal
standard jobs
special
jobs >18 hours
interactive
debugging, timing,
shared access
2
4
0.8%
18
265
6.4%
18
265
86.3%
100
265
5.3%
2
3
1.2%
p690 nodes
high32
standard jobs,
higher priority
18
5
18.6%
normal32
standard jobs
18
5
71.9%
special32
jobs >18 hours
100
5
1.6%
1
7.9%
serial, parallel jobs
interactive32
from command
n/a
/dsdirect
line
All nodes
diagnosis
System admin use n/a
n/a
<0.01%
RP-only data.
TeraGrid Data Reporting by RP
Table 2.1.6. SAM-QFS Monthly Statistics
UPTIME
PM
TERABYTES
OUTAGE
# HRS #
HRS
AVAILABLE
% UP
DISK
TAPE
WRITES
READS
USED
DISK+TAPE
FILES
TB
FILES
TB
Oct-05
0
0 2
8
98.9
120
6,144
398.4
1,690,728
12.02
130,362
12.83
Nov-05
1
16 0
0
97.8
370
6,144
404.1
746,196
5.70
270,791
10.34
Dec-05
0
0 0
0
100.0
370
6,144
409.6
375,160
5.48
36,119
4.00
Jan-06
0
0 1
91
87.8
370
6,144
441.3
3,458,887
31.76
128,292
7.53
Feb-06
2
10 0
0
98.5
316 18,432
456.0
1,539,283
14.67
230,193
6.01
Mar-06
0
0 1
1
99.9
316 18,432
474.2
1,363,008
18.24
5,200,776
5.82
Apr-06
1
9 1
77
88.1
316 18,432
481.8
1,402,958
7.61
280,585
3.77
May-06
1
5 3
78
88.8
188 18,432
491.5
1,660,534
9.70
1,219,292
19.93
Jun-06
1
9 0
0
98.8
188 18,432
500.1
258,862
8.59
287,565
8.36
Jul-06
0
0 1
2
99.7
188 18,432
517.5
892,883
17.33
46,150
4.44
Aug-06
0
0 0
0
100.0
188 18,432
522.5
302,105
5.03
65,759
10.00
Sep-06
0
0 0
0
100.0
188 18,432
528.3
815,533
5.82
713,785
11.79
Totals
6
14,506,137
141.97
8,609,669
104.83
49
9
257
Repeat for each system.
RP-only data.
96.5
TeraGrid Data Reporting by RP
Table 2.1.9. Storage Resource Broker Monthly Statistics
UPTIME
PM
#
HRS
% UP
Oct-05
1
5 0
0
99.3
Nov-05
1
4 0
0
99.4
Dec-05
3
16 1
2.5
97.5
Jan-06
0
0 1
91
87.8
Feb-06
2
20 0
0
97.0
Mar-06
3
13 0
0
98.3
Apr-06
1
9 1
77
88.1
May-06
1
5 3
78
88.8
Jun-06
0
0 0
0
100.0
Jul-06
0
0 1
4
99.5
Aug-06
1
3 0
0
99.6
Sep-06
0
0 2
16
97.8
Totals
RP-only data.
OUTAGE
HRS #
13
75
9
268.5
96.1
TeraGrid Data Reporting by RP
E. Distribution of Wait Times, by Job Duration and Size:
Shown most clearly for DataStar p655 and TeraGrid Cluster, wait times are typically longer for
jobs of larger node counts and of longer duration.
RP-only data.
TeraGrid Data Reporting by RP
G. Distribution of time since last accessed by file count:
A significant fraction of files in GPFS-WAN have typically been read within the last six-months, consistent with its
use in concert with compute allocations. For SAM-QFS, more older-files have been read more recently, consistent
with activity based on shared collections.
GPFS-WAN
SAM-QFS
12,000,000
12,000,000
10,000,000
10,000,000
8,000,000
File Count
8,000,000
6,000,000
File Count
4,000,000
6,000,000
4,000,000
2,000,000
25+
13-24
0
7-12
Never
Read
0-6
7-12
Months Since Last Access
RP-only data.
0-6
13-24
Age of File in Months
2,000,000
25+
13-24
0
7-12
Never
Read
0-6
7-12
25+
Months Since Last Access
0-6
13-24
25+
Age of File in Months
TeraGrid Data Reporting by RP
H. Distribution of time since last accessed by data volume (in TB):
For GPFS-WAN, a large fraction of the data by volume has been read in the past six months, consistent with
computation-related use. SAM-QFS shows noticeable activity within the past six months for files of all age ranges.
GPFS-WAN
SAM-QFS
45
100
40
90
80
35
70
30
60
25
Terabytes
Terabytes
50
20
40
15
30
10
20
25+
5
25+
10
13-24
13-24
0
7-12
Never
Read
0-6
7-12
Months Since Last Read
RP-only data.
0-6
13-24
Age of File in Months
0
7-12
Never
Read
0-6
7-12
25+
Months Since Last Access
0-6
13-24
25+
Age of File in Months
TeraGrid Data Reporting by RP
DataStar GPFS Usage
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
/06
9/1
/06
8/1
/06
7/1
/06
6/1
/06
5/1
/06
4/1
/06
3/1
/06
2/1
05
/06
1/1
/1/
12
05
05
/1/
11
/1/
10
Figure 1.1. DataStar GPFS Daily Utilization
RP-only data.
TeraGrid Data Reporting by GIG
Table 1.2.1. HPC System Academic Use by Discipline
DATASTAR P655
USERS
DATASTAR P690
%
NUS USERS
NUS
NUS
BLUEGENE DATA
%
NUS USERS
NUS
%
NUS USERS
35 52,924,798 21.1
30 5,346,298 29.4
2
34 23,728,931
9.5
15 1,577,639
8.7
6 16,823,296 22.3
36 23,825,128 27.7
8 22,601,069
9.0
66,554
0.4
6 32,522,951 43.2
12
7,396,077
8.6
35 29,751,893 11.9
16 1,612,233
8.9
11
Chemistry
29 29,819,350 11.9
18 5,273,680 29.0
8
9,655,380 12.8
15
2,736,531
3.2
Materials Research
62 27,203,073 10.9
19 2,137,764 11.8
5
6,197,089
8.2
19
4,547,837
5.3
Atmospheric Science
23 22,873,403
9.1
9
93,203
0.5
2
1,575,178
2.1
13
2,750,383
3.2
Earth Science
13 13,558,267
5.4
2
135,670
0.7
19
5,971,935
6.9
Bioeng., Environ. Sys.
17 13,750,215
5.5
5
8,241
0.0
10
2,831,433
3.3
Adv. Sci. Comp.
28
7,343,897
2.9
7
55,881
0.3
7
905,233
1.1
Integ. Bio., Neurosci.
5
2,767,769
1.1
2
11,039
0.1
9
1,494,062
1.7
Ocean Science
5
3,429,940
1.4
1
69,974
0.4
Mathematical Science
2
12,497
0.0
1
197,032
0.2
Mech., Struct. Systems
6
540,999
0.2
5
815,071
4.5
1
25,716
0.0
36
172,758
0.1
4
292,305
1.6
2
211,077
0.2
Social, Econ. Science
2
121,017
0.7
2
946,652
1.1
Elec., Comm. Systems
2
555,959
3.1
7
304,145
0.4
1
17,971
0.0
Computer Science
1
Environ. Biology
1
1
Design, Indus. Eng.
453,657
458,404
2.3
0.6
0.6
1
431,064
0.5
3
351,736
0.4
Biol. Instrumentation
1
13,570
0.0
1
4,250
0.0
1
217
1,420
1,713,223
0.4
0.0
23 22,194,999 25.8
Info., Intelligent Sys.
Behavioral Science
1
269,947
6.2
8,817,478 10.3
Chem., Thermal Sys.
5
4,692,869
13
%
NUS
Molec., Cellular Bio.
5
1.2
NUS
Astrophysics
Physics
926,122
TERAGRID CLUSTER
0.0
This data will be pulled and provided centrally by the GIG. (Dave Hart will pull from POPS)
Page 12 of 23
TeraGrid Data Reporting by GIG
Table 1.1.1. SDSC HPC Allocation Requests and Awards for FY06
DATASTAR P655
MEETING
# REQS
REQUEST (SU)
DATASTAR P690
# AWDS
AWARD (SU)
# REQS
REQUEST (SU)
# AWDS
AWARD (SU)
LRAC Sep 05
13
6,921,928
13
4,005,928
3
180,000
3
180,000
MRAC Sep 05
11
1,345,828
10
931,728
3
220,000
3
163,000
MRAC Dec 05
16
1,698,765
10
765,550
7
433,337
6
256,000
LRAC Mar 06
21
14,638,220
17
6,051,500
6
1,205,000
6
340,000
MRAC Mar 06
6
919,677
6
500,000
0
0
0
0
MRAC Jun 06
15
2,632,800
11
834,000
3
200,000
3
125,000
DAC FY06
57
491,724
52
402,000
38
207,150
29
144,500
139
28,648,942
119
13,490,706
60
2,445,487
50
1,208,500
TOTALS
TOTAL NUS
564,699,296
265,915,306
48,202,994
BLUEGENE DATA
MEETING
# REQS
REQUEST (SU)
# AWDS
23,820,744
TERAGRID CLUSTER
AWARD (SU)
# REQS
REQUEST (SU)
# AWDS
AWARD (SU)
LRAC Sep 05
4
2,750,000
4
3,050,000
14
9,998,185
14
6,163,785
MRAC Sep 05
1
200,000
1
100,000
4
447,670
4
377,670
MRAC Dec 05
6
160,000
6
705,000
9
360,886
6
392,240
LRAC Mar 06
6
2,708,689
6
3,383,689
17
7,791,542
16
5,226,563
MRAC Mar 06
2
15,000
2
15,000
9
646,646
9
602,600
MRAC Jun 06
3
80,000
2
60,000
14
2,448,939
7
955,000
25
280,996
20
156,500
6,194,685
41
DAC FY06
TOTALS
TOTAL NUS
47
74,639,760
7,470,189
90,008,307
n/a
67
21,693,868
533,148,500
56
13,717,858
337,130,078
Note: TeraGrid Cluster data reflect totals for the multi-site resources. TeraGrid DACs are awarded as TeraGrid Roaming.
This data will be pulled and provided centrally by the GIG. (Dave Hart will pull from POPS)
Page 13 of 23
TeraGrid Data Reporting by GIG
Table 1.1.2. SDSC System Utilization Summary for FY06.
DATASTAR P655
DATASTAR P690
BLUEGENE DATA
TERAGRID CLUSTER
NU
%
NU
%
NU
%
NU
%
NUs Available to TeraGrid
NUs Delivered to TeraGrid
375,726,351
291,764,386
77.7
33,152,325
19,929,848
60.1
216,164,844
80,520,141
37.2
113,406,259
89,738,313
79.1
NUs Requested
NUs Allocated
564,699,296
265,915,306
150.3
70.8
48,202,994
23,820,744
145.4
71.9
74,639,760
90,008,307
34.5
41.6
106,629,700
67,426,016
94.0
59.5
85.8
18,173,948
91.2
75,288,116
93.5
85,974,309
95.8
81.9
4.0
17,271,766
902,182
86.7
4.5
69,889,694
5,398,422
86.8
6.7
84,569,000
1,405,309
94.2
1.6
8.0
1.7
11,817
1,667,373
0.1
8.4
135,949
551,037
0.2
0.7
0
940,097
0.0
1.0
% of Resource Available to TeraGrid
Usage by Category (% of NUs Delivered)
250,479,076
TeraGrid Staff Usage
238,890,524
TeraGrid Allocated Usage
11,588,552
TeraGrid Discretionary Usage
TeraGrid Specific Usage
TeraGrid Roaming Usage
23,312,407
4,827,391
Definitions. Total NUs: Processors x Hours in Year x Weighting Factor. NUs Requested, Allocated: Time requested,
allocated via DAC, MRAC, LRAC. TeraGrid Roaming requests are excluded. For TeraGrid Cluster, NUs are 20% of
total requested/allocated for this three-site resource. Academic-Allocated: Time used by projects awarded via allocation
committees (DAC, MRAC, LRAC), including use by TeraGrid Roaming users. Academic-Discretionary: Time made
available by SDSC. ASCI: Usage funded by interagency transfer from DOE/NNSA. AAP: Academic Associates
Program, funded by U California. TeraGrid Staff: Non-SDSC RP or GIG staff.
This data will be pulled and provided centrally by the GIG. (Dave Hart will pull from POPS)
Page 14 of 23
TeraGrid Data Reporting by GIG
Table 1.1.3. SDSC Data Resource Allocation Requests and Awards for FY06
SDSC COLLECTIONS DISK SPACE
MEETING
# REQS REQUESTS (TB) # AWDS
TERAGRID GPFS-WAN
AWARD (TB)
# REQS
REQUESTS (TB)
# AWDS
AWARDS (TB)
MRAC Jun 06
1
20
1
1
LRAC Sep 06
2
41
0
0
2
50
1
40
MRAC Sep 06
0
0
0
0
2
31
1
30
DAC FY06
6
6.2
5
5.1
Totals
9
67.2
6
6.1
4
81
2
70
This data will be pulled and provided centrally by the GIG. (Dave Hart will pull from POPS)
Page 15 of 23
TeraGrid Data Reporting by GIG
Table 1.3.1. DataStar p655 Academic Use by Institution
INSTITUTION USERS
NUS
% NUS
UC Berkeley
35 37,395,871
14.9
UC San Diego
33 31,885,134
12.7
8 21,178,611
8.5
12 16,816,607
6.7
U Colorado
Georgia Tech
Repeated for each system.
This data will be pulled and provided centrally by the GIG. (Dave Hart will pull from POPS)
Page 16 of 23
TeraGrid Data Reporting by GIG
Table 1.4.1. DataStar p655 Top 20 Projects
PI
CONTACT
DISC
Juri Toomre,
U Colorado
303-492-7854
jtoomre@solarz.colorado.edu
AST
Richard I. Klein,
UC Berkeley
925-422-3548
rklein@astron.berkeley.edu
AST
Michael Norman,
UC San Diego
Gregory A. Voth,
U Utah
J.A. McCammon,
UC San Diego
Pui-Kuen Yeung,
Georgia Tech
Tony M. Keaveny,
UC Berkeley
858-822-4194
mlnorman@ucsd.edu
801-581-7272
voth@chem.utah.edu
858-534-2905
jmccammon@ucsd.edu
404-894-9341
yeung@peach.ae.gatech.edu
510-643-8017
tmk@me.berkeley.edu
AST
CHE
PROJECT TITLE
Coupling of Turbulent Compressible Solar
Convection with Rotation, Shear and Magnetic
Fields
Progress Towards a Comprehensive Theory of Star
Formation -- From Low to High Mass and on to
Giant Molecular Clouds
Projects in Astrophysical and Cosmological Structure
Formation
Fundamental Dynamical Processes in Condensed
Matter
MCB Theory of Biomolecular Structure and Dynamics
CTS
BCS
20,205,422
17,010,036
16,657,742
16,541,662
15,800,978
Large Numerical Simulations of Turbulence:
11,660,305
Resolution, Scaling, Shear and Rotation
Failure Mechanisms of Trabecular Bone and the
11,603,234
Whole Vertebral Body
Repeated for each system.
This data will be pulled and provided centrally by the GIG. (Dave Hart will pull from POPS)
Page 17 of 23
FY06 NUS
TeraGrid Data Reporting by GIG
Table 2.1.2. DataStar p655 Monthly Statistics (Centrally, from GIG)
JOBS
TG CPU-HRS
TG
UTILIZATION
MONTH
TOTA
L
200510
1,671 1,618,944
1,269,578
78.4
1,532
170
200511
3,034 1,566,720
1,265,085
80.7
1,584
177
200512
3,816 1,618,944
1,384,344
85.5
1,618
149
2006-1
5,752 1,618,944
1,369,031
84.6
1,674
165
2006-2
5,798 1,462,272
1,126,868
77.1
1,708
182
2006-3
4,320 1,618,944
1,179,118
72.8
1,866
176
2006-4
3,006 1,566,720
1,134,440
72.4
2,135
154
2006-5
2,185 1,618,944
1,076,593
66.5
2,324
159
2006-6
1,847 1,566,720
1,201,941
76.7
2,445
161
2006-7
1,746 1,618,944
1,339,296
82.7
2,566
171
2006-8
1,595 1,618,944
1,297,926
80.2
2,673
144
2006-9
1,441 1,566,720
1,195,704
76.3
2,803
159
FY06
36,21 19,061,76
1
0 14,839,924
77.9
2,906
475
TOTAL
USED
%
TOTAL
TG USERS
LOG
INS
CHARGED
Repeat for each system.
This data will be pulled and provided centrally by the GIG. (Dave Hart will pull from POPS)
Page 18 of 23
TeraGrid Data Reporting by GIG
A. Distribution of CPU-hours Delivered, by Job Size and Duration:
SDSC’s capability HPC resources deliver most of their CPU-hours to larger and longer jobs.
This data will be pulled and provided centrally by the GIG. (Dave Hart will pull from POPS)
Page 19 of 23
TeraGrid Data Reporting by GIG
B. Monthly distribution of CPU-hours delivered, by job size:
Except for DataStar p690, each SDSC system delivers a significant percentage of its CPU-hours to jobs that use
one-quarter or more of the total CPUs.
This data will be pulled and provided centrally by the GIG. (Dave Hart will pull from POPS)
Page 20 of 23
TeraGrid Data Reporting by GIG
C. Distribution of Job Counts and CPU-Hours Delivered, by Job Size (in Nodes):
Small jobs dominate the raw job counts, but larger jobs dominate the total CPU-hours delivered for DataStar
p655 and TeraGrid Cluster.
DATASTAR P655
NODES
JOBS
CPU-HRS
DATASTAR P690
JOBS
CPU-HRS
BLUEGENE DATA
JOBS
TERAGRID CLUSTER
CPU-HRS
JOBS
CPU-HRS
1
48,545
574,064
7,268
709,804
19,682
111,701
2–3
14,276
473,544
307
52,771
5,453
19,696
4–7
8,883
904,706
184
14,334
2,174
40,342
8–15
7,330
1,427,325
5,609
146,693
16–31
3,209
2,328,731
1,896
264,989
32–63
2,076
2,559,980
11
474
1,697
406,423
64–127
1,388
3,828,684
1,646
634,712
1,231
1,044,770
128–255
659
2,020,110
754
690,280
790
1,496,666
256–511
180
752,254
1,530
1,758,452
81
76,128
1,200
3,242,490
388
450,385
512–1023
1024
This data will be pulled and provided centrally by the GIG. (Dave Hart will pull from POPS)
Page 21 of 23
TeraGrid Data Reporting by GIG
D. Distribution of Job Counts and CPU-Hours Delivered, by Job Duration (Nearest Hour):
Short jobs (less than 1 hour) dominate the raw job counts, but larger jobs dominate the total CPU-hours
delivered.
DATASTAR P655
NEAREST HR
JOBS
CPU-HRS
DATASTAR P690
JOBS
CPU-HRS
BLUEGENE DATA
JOBS
CPU-HRS
TERAGRID CLUSTER
JOBS
CPU-HRS
0
50,039
374,589
1,448
5,230
3,573
254,821
14,950
61,359
1
12,428
718,784
864
7,709
549
250,387
11,005
88,666
2
10,223
665,686
464
8,271
144
134,867
2,497
75,155
3
1,759
349,283
366
10,640
74
103,788
1,518
61,216
4
1,614
423,813
636
26,080
88
156,081
1,435
85,363
5
1,148
460,695
320
17,533
96
360,100
1,861
108,919
6
996
413,580
255
15,723
95
336,000
1,264
91,989
7
420
218,664
194
28,009
83
487,809
796
104,624
8
978
580,399
227
23,805
64
425,867
513
155,448
9
708
425,465
238
18,854
135
859,173
334
129,632
10
621
699,361
267
48,367
87
488,531
301
123,022
11
296
337,959
156
19,600
78
331,233
146
130,384
12
524
708,113
169
29,745
392
1,712,240
276
253,892
13
241
483,033
179
30,444
71*
875,896*
73
70,692
14
294
630,240
123
13,951
165
193,271
15
1,182
1,007,887
126
20,756
144
169,496
16
614
1,163,519
98
20,818
140
241,138
17
582
786,058
116
24,234
183
97,850
18
1,808
3,759,460
1,417
389,880
950
1,259,276
19+
71
662,810
96
17,260
62
106,016
(*BlueGene Data has a normal job limit of 12 hours, so all jobs of 13 hours or longer are counted in these cells.)
This data will be pulled and provided centrally by the GIG. (Dave Hart will pull from POPS)
Page 22 of 23
TeraGrid Data Reporting from TeraGrid.org catalog
Table 1.5.1. SDSC-based Data Collections
TITLE
URL
FIELD OF
SCIENCE
ALLOCATED
Arctic Ocean Diversity Project (Databases for DIGGIR
as part of Census of Marine Life/OBIS)
LUSciD - Scientific Data Management
NVO – Digital Sky
Sloan Digital Sky Survey
The Galactic ALFA HI Surveys
50-year Downscaling of Global Analysis over California
Region
ak.aoos.org:9000/digir
ASC
nvo.sdsc.edu
www.sdss.org
astron.berkeley.edu/~sstanimi/GALFA/galfa_page.html
AST
AST
AST
AST
meteora.ucsd.edu/cap
ATM
stat.hpwren.ucsd.edu
CNS
LEGACY
HPWREN Sensor Network Data & Wireless Network
Analysis Data
Backbone Header Traces
IMDC
Security Logs and Archives
Backscatter Data
www.caida.org/analysis/measurement/oc48_data_request.xml
www.caida.org/funding/trends/imdc
www.caida.org/data/passive/backscatter_request.xml
Point to teragrid.org catalog of data collections.
Page 23 of 23
CNS
CNS
CNS
CNS
Download