Skip to content

Feature/qgpu - #142

Open
goodstudyqaq wants to merge 238 commits into
mainfrom
feature/qgpu
Open

goodstudyqaq wants to merge 238 commits into
mainfrom
feature/qgpu

Conversation

@goodstudyqaq

Copy link
Copy Markdown
Collaborator

No description provided.

Chiel Jespers and others added 30 commits October 16, 2025 14:40
fix parse excluded file bug when accessing many atoms
Restore qgpu-qligfep functionality
goodstudyqaq and others added 29 commits July 8, 2026 15:11
Co-authored-by: shen.guo <g.shen@rug.nl>
Co-authored-by: shen.guo <g.shen@rug.nl>
* add cpu water boundary force class

* add cuda water boundary force

---------

Co-authored-by: shen.guo <g.shen@rug.nl>
Co-authored-by: shen <goodstudyqaq@gmail.com>
* make context cleaner

* start to use new context

* finish clean

---------

Co-authored-by: shen.guo <g.shen@rug.nl>
Co-authored-by: shen <goodstudyqaq@gmail.com>
Co-authored-by: shen.guo <g.shen@rug.nl>
* add multi-replica CLI and context metadata

* make energy buffer replica-aware

* allocate dynamic state for replica batches

* normalize replica inputs and initialize batched state

* make temperature calculation replica-aware

* make CUDA integrator replica-aware

* make CUDA SHAKE replica-aware

* make CUDA bonded forces replica-aware

* make CUDA restraint forces replica-aware

* make CUDA water-boundary forces replica-aware

* make CUDA nonbonded forces replica-aware

* write replica-aware native outputs

* remove the guard

---------

Co-authored-by: shen.guo <g.shen@rug.nl>
Co-authored-by: shen.guo <g.shen@rug.nl>
* Revert "update benchmark qgpu input to inp and support replicates"

This reverts commit 182ac36.

* Revert "centralize output buffer synchronization in Handler (#113)"

This reverts commit d0ddeaa.

* Revert "Feature/qgpu multi replicate v2 (#112)"

This reverts commit feedaea.

* Revert "Revert "centralize output buffer synchronization in Handler (#113)""

This reverts commit 6ffe3f5.

---------

Co-authored-by: shen.guo <g.shen@rug.nl>
Co-authored-by: shen <goodstudyqaq@gmail.com>
* Revert "update benchmark qgpu input to inp and support replicates"

This reverts commit 182ac36.

* Revert "centralize output buffer synchronization in Handler (#113)"

This reverts commit d0ddeaa.

* Revert "Feature/qgpu multi replicate v2 (#112)"

This reverts commit feedaea.

* Revert "Revert "centralize output buffer synchronization in Handler (#113)""

This reverts commit 6ffe3f5.

* rename Shake abstraction to ConstraintForce

* finish cpu lincs

* add cuda lincs class

* support cuda small group lincs

* support cuda big group lincs

* add lincs header

* Revert "support cuda big group lincs"

This reverts commit 79b5e8b.

* support cuda big group lincs

* support cuda big group lincs in grid

---------

Co-authored-by: shen.guo <g.shen@rug.nl>
* Revert "update benchmark qgpu input to inp and support replicates"

This reverts commit 182ac36.

* Revert "centralize output buffer synchronization in Handler (#113)"

This reverts commit d0ddeaa.

* Revert "Feature/qgpu multi replicate v2 (#112)"

This reverts commit feedaea.

* Revert "Revert "centralize output buffer synchronization in Handler (#113)""

This reverts commit 6ffe3f5.

* remove the reset logic

---------

Co-authored-by: shen.guo <g.shen@rug.nl>
* support mix constraint force class

* fix big lincs group segment fault

* fix(cuda): validate SETTLE inputs and restore serial SHAKE initialization

* support to parse constraint input
* Fix harmonic Q-bond and Q-angle FEP parsing

* implement q-improper and q-torsion parse

* support q bonds cpu version

* support q bonds gpu version

* [nonbonded force] initialize mask to avoid use LJ_matrix

* support morse bond force

* support softpair

* support bonded coupling

* add off-diagonal and #comment and change the way to find water bond in water boundary force

* add off-diagonal and #comment and change the way to find water bond in water boundary force

* add off-diagonal and #comment and change the way to find water bond in water boundary force

* remove no use file

* move 14 nb force to bonded force
* support cpu lrf

* simplify cpu lrf

* simplify cpu lrf

* support cuda lrf, unfinish

* support cuda lrf first version

* support cuda lrf first version

* optimize init lrf part

* optimize init lrf part

* optimize exact entry calculation part

* optimize lrf coefficient

* add z order when building atom list

* add 14 mask

---------

Co-authored-by: shen <goodstudyqaq@gmail.com>
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.

7 participants