[whahnsr]

Summary

Inline the common pointer-based distance routines in DistRoutines.h:

• DIST2()
• DIST2_ImageOrtho()
• DIST2_NoImage()

For orthogonal imaging, cache the three box lengths and use std::fabs() for coordinate differences.

This follows the suggestion in #1196 to refactor the common distance routines so Actions beyond Action_Radial can benefit from the same optimization.

Performance

Tested with Action_MinMaxDist using the 11,670-atom orthogonal trajectory from Test_AutoImage.

The benchmark evaluates approximately 170 million pair distances over 5 frames.

Five serial runs with OMP_NUM_THREADS=1:

Baseline user times:
1.26, 1.24, 1.25, 1.25, 1.24 s

Baseline mean:
1.248 s

Optimized user times:
0.92, 0.92, 0.93, 0.92, 0.92 s

Optimized mean:
0.922 s

CPU time reduction:
26.1%

Speedup:
approximately 1.35x

Before the change, perf showed:

• DIST2_ImageOrtho(): 56.71%
• DIST2(): 18.07%

After inlining, those routines no longer appear as separate hot symbols; their work is folded into the calling Action_MinMaxDist routine.

Validation

The MinMaxDist benchmark output was byte-for-byte identical between baseline and optimized builds.

Regression tests:

• Test_MinMaxDist: 6/6 comparisons passed
• Test_Distance: 5/5 comparisons passed
• Test_Radial: 17/17 comparisons passed

git diff --check reports no whitespace errors.

[drroe]

That's funny, I actually have a branch I'm working on that implements something similar. I'll just merge this for now in the meantime.

[drroe]

@whahnsr If you would like me to add your name and affiliation to the cpptraj README, let me know.

[whahnsr]

Hello Daniel, Yes I would very much like to get copies of the cpptraj README. My real name is Werner Hahn, and I am an independent consultant. Most of my previous career was devoted to optimizing and porting software to all kinds of platforms. I got a PhD in Chemical Engineering from the University of Arizona in 1979 (yes I am old). The highlights of my works was in Reservoir Simulation with Exxon and Saudi Aramco. I did porting and optimization with Kendall Square Research and Hewlett Packard. I decided to keep busy in retirement by running around GitHub and finding packages where developers welcome outside development of their packages. I came across cpptraj and hopefully contributed a little bit to the development of the program. I did this with help from ChatGpt that can read code and analyze it infinitely faster than I can. But the final decision of what to do and how much to do was always mine. Looking forward to keeping in touch with cpptraj, Werner

…

On Wed, Oct 14, 2026 at 11:47 AM Daniel R. Roe ***@***.***> wrote: *drroe* left a comment (Amber-MD/cpptraj#1197) <#1197 (comment)> @whahnsr <https://github.com/whahnsr> If you would like me to add your name and affiliation to the cpptraj README, let me know. — Reply to this email directly, view it on GitHub <#1197?email_source=notifications&email_token=CPA2VLUVSU4VUOEZKZYSS3T5T7C35A5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTMMJXGU3TONJSGEZ2M4TFMFZW63VHNVSW45DJN5XKKZLWMVXHJLDGN5XXIZLSL5RWY2LDNM#issuecomment-6175775213>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/CPA2VLW6G57RUEG5UHDFSYT5T7C35AVCNFSNUABEKJSXA33TNF2G64TZHMZDOOBSHAYDSNZ3JFZXG5LFHM2TQNBQGY3TQNJYGSQXMAQ> . Triage notifications, keep track of coding agent tasks and review pull requests on the go with GitHub Mobile for iOS <https://github.com/notifications/mobile/ios/CPA2VLS5PJC23G72LD3MPAD5T7C35A5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTMMJXGU3TONJSGEZ2M4TFMFZW63VHNVSW45DJN5XKKZLWMVXHJKTGN5XXIZLSL5UW64Y> and Android <https://github.com/notifications/mobile/android/CPA2VLXWYJ77UGCIFLU5W6L5T7C35A5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTMMJXGU3TONJSGEZ2M4TFMFZW63VHNVSW45DJN5XKKZLWMVXHJLTGN5XXIZLSL5QW4ZDSN5UWI>. Download it today! You are receiving this because you were mentioned.Message ID: ***@***.***>

[drroe]

I've added some details to the README, see #1203

Let me know if you'd like anything changed. Thanks again for your contributions!

-Dan