Commit Graph
6 Commits
Author SHA1 Message Date
Erik Fröjdh 941325ea15 More uniform binding of Numpy arrays (#381)
Build on RHEL9 / build (push) Successful in 2m54s
Build on RHEL8 / build (push) Successful in 3m31s
Run tests using data on local RHEL8 / build (push) Failing after 4m28s
Build on local RHEL8 / build (push) Successful in 3m6s
- Introduce a standard way to bind numpy arrays: py::array_t<T> and
.noconvert() on the argument
- Take py::array_t by value since it's just a small wrapper
- Update bindings
2026-09-25 14:40:33 +02:00
mazzol_a afce91b358 Fix/clusterfinder test (#359)
Build on RHEL9 / build (push) Successful in 2m41s
Build on RHEL8 / build (push) Successful in 3m19s
Run tests using data on local RHEL8 / build (push) Successful in 4m21s
Build on local RHEL8 / build (push) Successful in 2m53s
- fixed test in ClusterFinderMT (use minimum one pedestal to accumulate
dummy pedestal)
- added minimum_pedestal_samples as additional constructor argument in
ClusterFinder and ClusterFinderMT
- bugfix: race conditions for empty queues in collect() - check directly
when needed
2026-09-09 16:38:54 +02:00
Erik Fröjdh eddb919328 FastPedestal + optimizations in the cluster finder (#341)
Build on RHEL9 / build (push) Successful in 2m42s
Build on RHEL8 / build (push) Successful in 3m21s
Run tests using data on local RHEL8 / build (push) Failing after 3m56s
Build on local RHEL8 / build (push) Successful in 2m54s
Optimized pedestal tracking with FastPedestal and performance improvements to the cluster finder. 
- ~2x faster hitting 14k FPS on benchmark dataset on 16 core threadripper pro


**Changes**
- No bounds check on interior of frame. (only within half of a cluster)
(~15% improvement)
- Cache threshold (std*n_rms) Biggest improvement comes from not
computing std for each access
- Cache pedestal subtracted frame
- Switched to FastPedestal which assumes we reached steady state

**Things tried but rejected:**
- Always store cluster values, commit on val==max (~15% drop in frame
rate)

**Options**
- using 16 bit clusters and 16 bit pedestal. (slows down the single
threaded case but allow us to reach 12k with multi threading)

**Other notes on performance**
- Passing in memory frames from python and doing nothing: 0.8us/frame
- Passing + pedestal subtraction: 28us/frame
- Passing + pedestal + cluster finding (no store): 652us/frame
- Passing + pedestal + cluster finding: 886 us/frame
2026-09-01 18:41:12 +02:00
mazzol_a 05e6e69c66 added getter and setter for nSigma (#293)
Build on RHEL9 / build (push) Successful in 2m46s
Build on RHEL8 / build (push) Successful in 2m51s
Run tests using data on local RHEL8 / build (push) Successful in 3m15s
Build on local RHEL8 / build (push) Successful in 1m57s
2026-03-26 14:52:32 +01:00
Erik Fröjdh 200ae91622 also hpp 2025-11-21 10:14:14 +01:00
Erik Fröjdh 69964e08d5 Refactor cluster bindings (#185)
Build on RHEL9 / build (push) Successful in 2m19s
Build on RHEL8 / build (push) Successful in 2m34s
- Split up the file for cluster bindings
- new file names according to bind_ClassName.hpp
2025-06-03 08:43:40 +02:00