Skip to main content

Table 1 (abstract A4). Comparison of the time and memory required by the C-PAC and AFNI implementations to calculate DC (sparsity and correlation threshold) and lFCD on the first resting state scan of the first scanning session for all 36 participants’ data in the IBATRT dataset

From: 2015 Brainhack Proceedings

  

r > 0.6

r > 0.6

0.1 % Sparsity

Impl.

Thr.

Mem GB

TD s

Mem GB

TD s

Mem GB

TD s

Python

1

2.17 (0.078)

67.7 (3.90)

5.62 (0.176)

342.2 (12.25)

2.16 (0.082)

88.3 (6.40)

C

1

0.84 (0.003)

62.6 (9.23)

0.85 (0.002)

86.3 (13.83)

0.86 (0.003)

8.8 (1.27)

C

2

0.86 (0.002)

39.0 (4.62)

0.86 (0.003)

38.2 (0.55)

0.86 (0.003)

5.1 (0.25)

C

4

0.86 (0.003)

18.2 (1.93)

0.87 (0.003)

19.0 (0.45)

0.87 (0.003)

4.3 (0.23)

C

8

0.87 (0.002)

11.2 (0.25)

0.87 (0.000)

11.3 (0.31)

0.87 (0.000)

4.1 (0.15)

  1. Values are averaged across the 36 datasets and presented along with standard deviations in parenthesis. Impl: Implementation, Thr: Number of threads used to process a single dataset, Mem: average (standard deviation) memory in gigabytes used to process a single dataset, TD: the average (standard deviation) time in seconds to process a dataset. These statistics were collected on a C3.xlarge Amazon Web Services Elastic Compute Cloud node with 8 hyperthreads and 15 GB of RAM