Namespaces

Variants
Actions
Personal tools

Talk:NAMD Benchmarks

From UABgrid Documentation

(Difference between revisions)
Jump to: navigation, search
(Table)
Line 14: Line 14:
 
|-
 
|-
 
|1
 
|1
|
+
|15.4054
|
+
|18.0535
|
+
|19.1000
 
|-
 
|-
 
|2
 
|2
|
+
|7.7119
|
+
|9.5163
|
+
|9.7600
 
|-
 
|-
 
|4
 
|4
|
+
|3.8933
|
+
|4.9222
|
+
|4.7570
 
|-
 
|-
 
|8
 
|8
|
+
|1.9653
|
+
|2.5763
|
+
|2.5360
 
|-
 
|-
 
|16
 
|16
|
+
|0.9950
|
+
|1.2658
|
+
|1.3870
 
|-
 
|-
 
|32
 
|32
|
+
|0.5101
|
+
|0.6463
|
+
|0.7438
 
|-
 
|-
 
|64
 
|64
|
+
|0.2592
|
+
|0.3390
|
+
|0.3938
 
|-
 
|-
 
|128
 
|128
|
+
|0.1360
|
+
|NA
|
+
|NA
 
|-
 
|-
 
|256
 
|256
|
+
|0.07706
|
+
|NA
|
+
|NA
 
|}
 
|}
  

Revision as of 08:37, 26 May 2011

Benchmark introduction Calculation of efficiency Scaling to higher processor count- Calculation for scaling

Contents

Sample Benchmark comparing Days/ns for Cheaha, Biowulf, and DMC using IB

Table

Sample Benchmarks
Processors Days/ns Cheaha Days/ns Biowulf Days/ns DMC
1 15.4054 18.0535 19.1000
2 7.7119 9.5163 9.7600
4 3.8933 4.9222 4.7570
8 1.9653 2.5763 2.5360
16 0.9950 1.2658 1.3870
32 0.5101 0.6463 0.7438
64 0.2592 0.3390 0.3938
128 0.1360 NA NA
256 0.07706 NA NA

Graph's

Efficiency

Bench mark notes

Comparison of Ethernet and IB

Comparison on different nodes on Cheaha Gen 2 and Gen 3 with IB

Actual job benchmarks (Segrest job)

Perform Your Own Benchmarks