Residual noise covariance for Planck low-resolution data analysis

Free Access

Table 4

CPU time costs for 1270 GHz detector years = 3 × 10¹⁰ samples.

Case	N _side	w _corr^a	PEs	CPUh

MADping	32	131 073	14 000	80000^b
MADping	32	4097	14 000	7525
ROMA	32	4097	512	25 000
ROMA	4	16 385	256	480
ROMA	4	4097	256	410
Madam 1.25 s, 50^cc	64	1759	1024	901
Madam 1.25 s, 50	32	1759	512	114
Madam 1.25 s, 10	32	8285	512	438
Madam 10 s, 50	32	289	128	42
Madam 60 s, 50	32	84	64	34
Madam 1.25 s, 50	16	1759	256	34
Springtide, 10	32	1	1024	256
Springtide, 10	16	1	1024	51

Notes. Calculations are done on 2.6 GHz Quad-core Opterons.

^(a)

The filter or baseline correlation length.

^(b)

Performed on an earlier, dual-core version of the machine.

^(c)

10 and 50 refer to 1/f knee frequencies in mHz.

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.