Loading...

Xporter

XML

Word

Printable

Details

Type: Feature
Priority: Not Assigned
Fix Version/s: PI8
Component/s: COM SDP SW
Labels:
- Team_YANDA

ARTs:

Data Processing
Benefit hypothesis:

Hide

The capability to correct for the direction-dependent ionosphere in tied-array beamforming enables a more complete characterization of the LEAP algorithm's technical as well as scientific performance.

Show
The capability to correct for the direction-dependent ionosphere in tied-array beamforming enables a more complete characterization of the LEAP algorithm's technical as well as scientific performance.
Acceptance criteria:
Hide

finalize multi-threaded CUDA implementation based on ~~SP-994~~

demo of parallel workflow using an execution engine

comparison of technical LEAP CUDA with other implementation(s) on a single device
Show
finalize multi-threaded CUDA implementation based on SP-994 demo of parallel workflow using an execution engine comparison of technical LEAP CUDA with other implementation(s) on a single device
Feature Points:
2
Initial Size:
3
WSJF:
5
Epic Link:
SDP workflow prototyping
Agile Teams:

Team_YANDA
Due Sprint:
Sprint 5
Story Point Burn-up:
Overdue:
Outcomes:

Hide

Code: https://github.com/ICRAR/leap-accelerate/tree/beta_experimental
EAGLE palette: https://github.com/ICRAR/EAGLE-graph-repo/blob/master/leap/LeapAccelerateCLI.palette
Logical Graph (MVP): https://github.com/ICRAR/EAGLE-graph-repo/blob/master/leap/LeapMVP.graph
Logical Graph (Dynlib components): https://github.com/ICRAR/EAGLE-graph-repo/blob/master/leap/LeapAccelerateCLI.graph
performance - some initial stats: https://confluence.skatelescope.org/display/SE/LEAP+Benchmarking

Show
Code: https://github.com/ICRAR/leap-accelerate/tree/beta_experimental EAGLE palette: https://github.com/ICRAR/EAGLE-graph-repo/blob/master/leap/LeapAccelerateCLI.palette Logical Graph (MVP): https://github.com/ICRAR/EAGLE-graph-repo/blob/master/leap/LeapMVP.graph Logical Graph (Dynlib components): https://github.com/ICRAR/EAGLE-graph-repo/blob/master/leap/LeapAccelerateCLI.graph performance - some initial stats: https://confluence.skatelescope.org/display/SE/LEAP+Benchmarking
Resolved PI.Sprint:
8.6

Feature Checklist:

Satisfies Acceptance Criteria

Demos:
- DP_ART_9.1
Requirement Status:

PI22 - UNCOVERED

Description

LEAP (Rioja, Dodson & Franzen 2018; MN 478, 2337) is a method for evaluating delays for a grid of sources across the sky, from which solutions can be interpolated to arbitrary directions and provided to CSP for beam-forming.The goal is to run it on ionospheric variation timescales, at least for compact configurations. This calls for a high performance implementation, that exploits data parallelism to suffice the real time constraint.

For proof of concept a CPU-based prototype has been put together, which serves as a starting point and reference when porting it to CUDA.

Attachments

Issue Links

depends on

SP-994 CUDA port of LEAP algorithm

Done

is required by

SP-1433 Onboard LEAP and explore Scope for Parallelization

Done

relates to

SP-311 CI platform for CUDA and specialist architectures

Discarded

SP-1166 Summarise and report on status of state-of-the-art DD calibration and imaging analysis for SKA1-LOW

Discarded

mentioned in: Page Loading...; Page Loading...; Page Loading...; Page Loading...; Page Loading...; Page Loading...

(5 mentioned in)

Structure

Activity

People

Assignee:: Nikolic, Bojan

Reporter:: Dolensky, Markus [X] (Inactive)

Votes:: 0 Vote for this issue

Watchers:: 4 Start watching this issue

Feature Progress

Story Point Burn-up: (100.00%)

Feature Estimate: 2.0

	Issues	Story Points
To Do	0	0.0
In Progress	0	0.0
Complete	13	26.0
Total	13	26.0

Dates

Created:: 07/Aug/20 6:55 AM

Updated:: 22/Aug/22 10:07 PM

Resolved:: 08/Dec/20 11:25 AM

Finalize CUDA port of LEAP and characterize Performance