Skip to content

Always build with JIT+LTO #1923

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.

Sign up for GitHub

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

Merged
rapids-bot merged 28 commits into from
Mar 25, 2026
Merged

Always build with JIT+LTO #1923

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
28 commits
Select commit Hold shift + click to select a range
fcffe8a
Always build with JIT+LTO
KyleFromNVIDIA Mar 16, 2026
6c91f9d
Use the driver API instead
KyleFromNVIDIA Mar 16, 2026
e858407
Conda recipe
KyleFromNVIDIA Mar 16, 2026
1972a74
deps
KyleFromNVIDIA Mar 16, 2026
4503307
PRIVATE
KyleFromNVIDIA Mar 16, 2026
a42ede0
auditwheel
KyleFromNVIDIA Mar 16, 2026
e26519f
Conda recipe
KyleFromNVIDIA Mar 17, 2026
697d1d0
Merge branch 'release/26.04' into jit-lto-cuda-12
KyleFromNVIDIA Mar 17, 2026
3269055
Revert "Conda recipe"
KyleFromNVIDIA Mar 17, 2026
07c50e6
COMPILE_ONLY
KyleFromNVIDIA Mar 17, 2026
788fd34
PUBLIC
KyleFromNVIDIA Mar 17, 2026
e16b88f
Revert "Use the driver API instead"
KyleFromNVIDIA Mar 17, 2026
96e9162
Remove driver dep
KyleFromNVIDIA Mar 17, 2026
8027d97
Default to static linking of libcudart (#1627)
bdice Mar 16, 2026
56229e8
Opt out of rmm's cudart dependency
KyleFromNVIDIA Mar 18, 2026
0a0540a
Make rmm interface dependency COMPILE_ONLY
KyleFromNVIDIA Mar 18, 2026
f379ad4
Merge branch 'main' into jit-lto-cuda-12
KyleFromNVIDIA Mar 18, 2026
38c9e9d
Merge branch 'main' into jit-lto-cuda-12
KyleFromNVIDIA Mar 19, 2026
17c5cd7
Push
KyleFromNVIDIA Mar 19, 2026
af0a04e
Merge branch 'main' into cudart-static
KyleFromNVIDIA Mar 19, 2026
8c771a5
Merge branch 'cudart-static' into jit-lto-cuda-12
KyleFromNVIDIA Mar 23, 2026
b6560be
Debugging
KyleFromNVIDIA Mar 23, 2026
84ddcf9
Downgrade to compute 7.0 for CUDA 12
KyleFromNVIDIA Mar 24, 2026
b08a35d
Merge branch 'main' into jit-lto-cuda-12
KyleFromNVIDIA Mar 24, 2026
a8493a3
Remove JIT_LTO_COMPILATION variable
KyleFromNVIDIA Mar 24, 2026
997ab66
Remove CUVS_ENABLE_JIT_LTO preprocessor definition
KyleFromNVIDIA Mar 24, 2026
fe67525
Use libnvjitlink run exports
KyleFromNVIDIA Mar 24, 2026
cb61d86
Merge branch 'main' into jit-lto-cuda-12
KyleFromNVIDIA Mar 24, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions conda/environments/all_cuda-129_arch-aarch64.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,6 +30,7 @@ dependencies:
- libcurand-dev
- libcusolver-dev
- libcusparse-dev
- libnvjitlink-dev
- librmm==26.4.*,>=0.0.0a0
- make
- nccl>=2.19
Expand Down
1 change: 1 addition & 0 deletions conda/environments/all_cuda-129_arch-x86_64.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,6 +30,7 @@ dependencies:
- libcurand-dev
- libcusolver-dev
- libcusparse-dev
- libnvjitlink-dev
- librmm==26.4.*,>=0.0.0a0
- make
- nccl>=2.19
Expand Down
1 change: 1 addition & 0 deletions conda/environments/bench_ann_cuda-129_arch-aarch64.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -29,6 +29,7 @@ dependencies:
- libcusolver-dev
- libcusparse-dev
- libcuvs==26.4.*,>=0.0.0a0
- libnvjitlink-dev
- librmm==26.4.*,>=0.0.0a0
- matplotlib-base>=3.9
- nccl>=2.19
Expand Down
1 change: 1 addition & 0 deletions conda/environments/bench_ann_cuda-129_arch-x86_64.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,6 +31,7 @@ dependencies:
- libcusolver-dev
- libcusparse-dev
- libcuvs==26.4.*,>=0.0.0a0
- libnvjitlink-dev
- librmm==26.4.*,>=0.0.0a0
- matplotlib-base>=3.9
- mkl-devel=2023
Expand Down
1 change: 1 addition & 0 deletions conda/environments/go_cuda-129_arch-aarch64.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,6 +25,7 @@ dependencies:
- libcusolver-dev
- libcusparse-dev
- libcuvs==26.4.*,>=0.0.0a0
- libnvjitlink-dev
- libraft==26.4.*,>=0.0.0a0
- nccl>=2.19
- ninja
Expand Down
1 change: 1 addition & 0 deletions conda/environments/go_cuda-129_arch-x86_64.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,6 +25,7 @@ dependencies:
- libcusolver-dev
- libcusparse-dev
- libcuvs==26.4.*,>=0.0.0a0
- libnvjitlink-dev
- libraft==26.4.*,>=0.0.0a0
- nccl>=2.19
- ninja
Expand Down
1 change: 1 addition & 0 deletions conda/environments/rust_cuda-129_arch-aarch64.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,6 +22,7 @@ dependencies:
- libcusolver-dev
- libcusparse-dev
- libcuvs==26.4.*,>=0.0.0a0
- libnvjitlink-dev
- libraft==26.4.*,>=0.0.0a0
- make
- nccl>=2.19
Expand Down
1 change: 1 addition & 0 deletions conda/environments/rust_cuda-129_arch-x86_64.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,6 +22,7 @@ dependencies:
- libcusolver-dev
- libcusparse-dev
- libcuvs==26.4.*,>=0.0.0a0
- libnvjitlink-dev
- libraft==26.4.*,>=0.0.0a0
- make
- nccl>=2.19
Expand Down
52 changes: 13 additions & 39 deletions conda/recipes/libcuvs/recipe.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -72,9 +72,7 @@ cache:
- ninja
- ${{ stdlib("c") }}
host:
- if: cuda_major == "13"
then:
- libnvjitlink-dev
- libnvjitlink-dev
- librmm =${{ minor_version }}
- libraft-headers =${{ minor_version }}
- nccl ${{ nccl_version }}
Expand Down Expand Up @@ -121,9 +119,7 @@ outputs:
- libcurand-dev
- libcusolver-dev
- libcusparse-dev
- if: cuda_major == "13"
then:
- libnvjitlink-dev
- libnvjitlink-dev
run:
- ${{ pin_compatible("cuda-version", upper_bound="x", lower_bound="x") }}
- libraft-headers =${{ minor_version }}
Expand All @@ -133,9 +129,7 @@ outputs:
- libcurand
- libcusolver
- libcusparse
- if: cuda_major == "13"
then:
- libnvjitlink
- libnvjitlink
ignore_run_exports:
by_name:
- cuda-cudart
Expand All @@ -149,9 +143,7 @@ outputs:
- librmm
- mkl
- nccl
- if: cuda_major == "13"
then:
- libnvjitlink
- libnvjitlink
about:
homepage: ${{ load_from_file("python/libcuvs/pyproject.toml").project.urls.Homepage }}
license: ${{ load_from_file("python/libcuvs/pyproject.toml").project.license }}
Expand Down Expand Up @@ -188,9 +180,7 @@ outputs:
- libcurand-dev
- libcusolver-dev
- libcusparse-dev
- if: cuda_major == "13"
then:
- libnvjitlink-dev
- libnvjitlink-dev
run:
- ${{ pin_compatible("cuda-version", upper_bound="x", lower_bound="x") }}
- ${{ pin_subpackage("libcuvs-headers", exact=True) }}
Expand All @@ -201,9 +191,7 @@ outputs:
- libcurand
- libcusolver
- libcusparse
- if: cuda_major == "13"
then:
- libnvjitlink
- libnvjitlink
ignore_run_exports:
by_name:
- cuda-cudart
Expand All @@ -217,9 +205,7 @@ outputs:
- librmm
- mkl
- nccl
- if: cuda_major == "13"
then:
- libnvjitlink
- libnvjitlink
about:
homepage: ${{ load_from_file("python/libcuvs/pyproject.toml").project.urls.Homepage }}
license: ${{ load_from_file("python/libcuvs/pyproject.toml").project.license }}
Expand Down Expand Up @@ -254,9 +240,7 @@ outputs:
- libcurand-dev
- libcusolver-dev
- libcusparse-dev
- if: cuda_major == "13"
then:
- libnvjitlink-dev
- libnvjitlink-dev
run:
- ${{ pin_compatible("cuda-version", upper_bound="x", lower_bound="x") }}
- ${{ pin_subpackage("libcuvs-headers", exact=True) }}
Expand All @@ -267,9 +251,7 @@ outputs:
- libcurand
- libcusolver
- libcusparse
- if: cuda_major == "13"
then:
- libnvjitlink
- libnvjitlink
ignore_run_exports:
by_name:
- cuda-cudart
Expand All @@ -280,9 +262,7 @@ outputs:
- libcurand
- libcusolver
- libcusparse
- if: cuda_major == "13"
then:
- libnvjitlink
- libnvjitlink
- librmm
- mkl
- nccl
Expand Down Expand Up @@ -419,9 +399,7 @@ outputs:
- libcurand-dev
- libcusolver-dev
- libcusparse-dev
- if: cuda_major == "13"
then:
- libnvjitlink-dev
- libnvjitlink-dev
run:
- ${{ pin_subpackage("libcuvs-headers", exact=True) }}
- ${{ pin_subpackage("libcuvs", exact=True) }}
Expand All @@ -431,9 +409,7 @@ outputs:
- libcurand
- libcusolver
- libcusparse
- if: cuda_major == "13"
then:
- libnvjitlink
- libnvjitlink
Comment thread
KyleFromNVIDIA marked this conversation as resolved.
Outdated
ignore_run_exports:
by_name:
- cuda-cudart
Expand All @@ -444,9 +420,7 @@ outputs:
- libcurand
- libcusolver
- libcusparse
- if: cuda_major == "13"
then:
- libnvjitlink
- libnvjitlink
- librmm
- mkl
- nccl
Expand Down
8 changes: 2 additions & 6 deletions cpp/CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -355,12 +355,8 @@ if(NOT BUILD_CPU_ONLY)
)
endif()

set(JIT_LTO_TARGET_ARCHITECTURE "")
set(JIT_LTO_COMPILATION OFF)
if(CMAKE_CUDA_COMPILER_VERSION VERSION_GREATER_EQUAL 13.0)
set(JIT_LTO_TARGET_ARCHITECTURE "75-real")
set(JIT_LTO_COMPILATION ON)
endif()
set(JIT_LTO_TARGET_ARCHITECTURE "75-real")
set(JIT_LTO_COMPILATION ON)
Comment thread
KyleFromNVIDIA marked this conversation as resolved.
Outdated

if(JIT_LTO_COMPILATION)
# Generate interleaved scan kernel files at build time
Expand Down
11 changes: 1 addition & 10 deletions dependencies.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -326,16 +326,7 @@ dependencies:
- libcurand-dev
- libcusolver-dev
- libcusparse-dev
specific:
- output_types: conda
matrices:
- matrix:
cuda: "13.*"
packages:
- libnvjitlink-dev
- matrix:
cuda: "12.*"
packages:
- libnvjitlink-dev
Comment thread
KyleFromNVIDIA marked this conversation as resolved.
cuda_wheels:
specific:
- output_types: [requirements, pyproject]
Expand Down
Loading