[WIP] Make Linear Elements Fast #850
Open
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
ax3l
added
component: elements
ax3l
commented
Feb 12, 2025
src/elements/Quad.H
Outdated
| @@ -96,24 +118,14 @@ namespace impactx::elements | |||
| amrex::ParticleReal & AMREX_RESTRICT py, | |||
| amrex::ParticleReal & AMREX_RESTRICT pt, | |||
| uint64_t & AMREX_RESTRICT idcpu, | |||
| RefPart const & refpart | |||
| [[maybe_unused]] RefPart const & refpart | |||
There was a problem hiding this comment.
Idea: I would remove this fully from the interface to prevent that we can write inefficient code by accident.
There was a problem hiding this comment.
Discussed: this will not work to be fully removed, since the refpart for elements that integrate through maps is non-const.
ax3l
force-pushed
the
performance-quad-constants
branch
from
3c74ef7 to
9ed66b3
Compare
Compute and cache the constants for the push. In particular, used to pre-compute and cache variables that are independent of the individually tracked particle.
ax3l
force-pushed
the
performance-quad-constants
branch
from
9ed66b3 to
e788211
Compare
ax3l
force-pushed
the
performance-quad-constants
branch
3 times, most recently
from
6c6f5ef to
6af0a54
Compare
Make the `Quad` fast by calculating constants that depend on the element config and reference particle properties before pushing all particles.
ax3l
force-pushed
the
performance-quad-constants
branch
2 times, most recently
from
63298bf to
55bc5e9
Compare
ax3l
force-pushed
the
performance-quad-constants
branch
from
55bc5e9 to
86c0c2b
Compare
ax3l
force-pushed
the
performance-quad-constants
branch
from
db2f5ac to
483e3df
Compare
ax3l
commented
Feb 15, 2025
Comment on lines
+159
to
+166
| xout = m_cos_kxds * x + m_sincx * px; | ||
| pxout = m_const_x * x + m_cos_kxds * px; | ||
|
|
||
| yout = std::cos(m_ky*slice_ds)*y + sincy*py; | ||
| pyout = -m_ky * std::sin(m_ky*slice_ds)*y + std::cos(m_ky*slice_ds)*py; | ||
| yout = m_cos_kyds * y + m_sincy * py; | ||
| pyout = m_const_y * y + m_cos_kyds * py; | ||
|
|
||
| tout = std::cos(m_kt*slice_ds)*t + sinct/(betgam2)*pt; | ||
| ptout = -(m_kt*betgam2) * std::sin(m_kt*slice_ds)*t + std::cos(m_kt*slice_ds)*pt; | ||
| tout = m_cos_ktds * t + m_const_pt * pt; | ||
| ptout = m_const_y * t + m_cos_ktds * pt; |
There was a problem hiding this comment.
We could actually give those members proper m_Rij names...
There was a problem hiding this comment.
We will need to return the linear map for envelope tracking anyway, in a separate function within ConstF. Those matrix element values could be used here.
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.
Make linear elements, like
Quad&Drift, as fast as possible by:Example: 1 CPU core run of a single
Quad, 100M particlesBefore:
Now:
Close #849