Fix AtPoints transpose shift #1723
Open
+20
−20
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
jeremylt
force-pushed
the
jeremy/at-points-shifts
branch
from
47bab60 to
e244d91
Compare
|
Note: Rust failure is an unrelated issue in their nightly build that they are working on |
zatkins-dev
reviewed
Jan 10, 2025
| @@ -74,7 +74,7 @@ inline __device__ void InterpTransposeAtPoints1d(SharedData_Cuda &data, const Ce | |||
| // Contract x direction | |||
| if (p < NUM_POINTS) { | |||
| for (CeedInt i = 0; i < Q_1D; i++) { | |||
| atomicAdd(&data.slice[comp * Q_1D + (i + p) % Q_1D], chebyshev_x[(i + p) % Q_1D] * r_U[comp]); | |||
| atomicAdd(&data.slice[comp * Q_1D + (i + p) % Q_1D], chebyshev_x[(i + data.t_id_x) % Q_1D] * r_U[comp]); | |||
There was a problem hiding this comment.
should the first i + p in this line also be changed?
There was a problem hiding this comment.
Oops, yeah. Goes to show that most of the time the two will be the same
| @@ -120,7 +120,7 @@ inline __device__ void GradTransposeAtPoints1d(SharedData_Cuda &data, const Ceed | |||
| // Contract x direction | |||
| if (p < NUM_POINTS) { | |||
| for (CeedInt i = 0; i < Q_1D; i++) { | |||
| atomicAdd(&data.slice[comp * Q_1D + (i + p) % Q_1D], chebyshev_x[(i + p) % Q_1D] * r_U[comp]); | |||
| @@ -74,7 +74,7 @@ inline __device__ void InterpTransposeAtPoints1d(SharedData_Hip &data, const Cee | |||
| // Contract x direction | |||
| if (p < NUM_POINTS) { | |||
| for (CeedInt i = 0; i < Q_1D; i++) { | |||
| atomicAdd(&data.slice[comp * Q_1D + (i + p) % Q_1D], chebyshev_x[(i + p) % Q_1D] * r_U[comp]); | |||
| @@ -120,7 +120,7 @@ inline __device__ void GradTransposeAtPoints1d(SharedData_Hip &data, const CeedI | |||
| // Contract x direction | |||
| if (p < NUM_POINTS) { | |||
| for (CeedInt i = 0; i < Q_1D; i++) { | |||
| atomicAdd(&data.slice[comp * Q_1D + (i + p) % Q_1D], chebyshev_x[(i + p) % Q_1D] * r_U[comp]); | |||
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Minor performance fix to help prevent collisions in atomic adds for AtPoints transpose basis action on GPUs
Note that previously
i = threadIdx.x + threadIdx.y * blockDim.xwas the value ofp, so this should be largely a more clear replacement for the same logic we had before