Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

catching up with upstream #1

Merged
merged 11 commits into from
Dec 20, 2020
Merged

catching up with upstream #1

merged 11 commits into from
Dec 20, 2020

Conversation

kaxap
Copy link
Owner

@kaxap kaxap commented Dec 20, 2020

No description provided.

favadi and others added 11 commits December 7, 2020 19:36
The version string will be injected at release step with the output of `git describe --tags --always`. This fixes #83.
This fixes #143 by parsing elements with VR=AT as Ints (instead of the default fallback of parsing as strings).
…sion. (#155)

This adds a more comprehensive test for converting NativeFrame data into an image, and also fixes the row selector in GetImage.
This updates the way the VERSION Makefile variable works.
…Hub Action (#158)

* Add -count=5 to benchmark GitHub Action

* Update to go 1.15
This change adds dedicated benchmark GitHub actions (to be triggered on pushes and pull requests), in addition to modernizing and cleaning up the Go Build/Test GitHub action.
This adds comprehensive benchmarks (and tests) specifically for readNativeFrames.
This change reduces the number of allocations in `readNativeFrames`. Instead of allocating a slice for each pixel's samples, a single flat buffer slice is allocated upfront of the size `pixelsPerFrame*samplesPerPixel`. Later, ranges in that slice are referred to in the larger 2D slice. This leads to there only being two calls to `make`, leading to significant performance gains.

On my machine running `make bench-diff`:
```
name           old time/op  new time/op  delta
Parse/1.dcm-8  2.29ms ± 1%  1.95ms ± 3%  -15.00%  (p=0.008 n=5+5)
Parse/2.dcm-8  2.27ms ± 1%  1.94ms ± 0%  -14.41%  (p=0.008 n=5+5)
Parse/3.dcm-8  2.10ms ± 0%  1.81ms ± 2%  -13.77%  (p=0.008 n=5+5)
Parse/4.dcm-8   240ms ± 1%   196ms ± 4%  -18.27%  (p=0.008 n=5+5)
Parse/5.dcm-8  33.6ms ± 1%  27.9ms ± 0%  -17.00%  (p=0.008 n=5+5)
```

We see similar results in the [GitHub action benchmark](https://github.com/suyashkumar/dicom/pull/157/checks?check_run_id=1574909456#step:5:26).

Presumably the percentage performance gains would be even higher for DICOMs with more Native PixelData (e.g. more frames, pixels per frame, samples per pixel, etc).

This helps address #161.
@kaxap kaxap merged commit 1780bba into kaxap:main Dec 20, 2020
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants