Skip to content

perf(cuda): migrate consumers to vortex-cuda-driver #10722

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
0ax1 merged 1 commit into from
Oct 23, 2026
Merged

perf(cuda): migrate consumers to vortex-cuda-driver #10722

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
38 changes: 8 additions & 30 deletions Cargo.lock
View file
Open in desktop

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

6 changes: 0 additions & 6 deletions Cargo.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -132,12 +132,6 @@ clap = "4.5"
criterion = { package = "codspeed-criterion-compat-walltime", version = "5.0.0" }
crossbeam-channel = "0.5.16"
crossterm = "0.29"
cudarc = { version = "0.19.0", features = [
# The NSight Compute version available on lambda.ai hosts does not inject a symbol for
# several functions that cudarc expects to load when this is set to "cuda-12080". We
# don't use any CUDA 12.8 specific features currently so this is fine.
"cuda-12050",
] }
custom-labels = "0.4.4"
dashmap = "6.2.1"
datafusion = { version = "55.0.0", default-features = false, features = [
Expand Down
2 changes: 0 additions & 2 deletions vortex-array/Cargo.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -28,7 +28,6 @@ arcref = { workspace = true }
arrow-buffer = { workspace = true }
bytes = { workspace = true }
cfg-if = { workspace = true }
cudarc = { workspace = true, optional = true }
enum-iterator = { workspace = true }
fearless_simd = { workspace = true }
fearless_simd_macros = { workspace = true }
Expand Down Expand Up @@ -76,7 +75,6 @@ vortex-utils = { workspace = true, features = ["dyn-traits"] }
[features]
arbitrary = ["dep:arbitrary", "dep:primitive-types"]
canonical_counter = []
cudarc = ["dep:cudarc"]
table-display = ["dep:tabled"]
_test-harness = ["dep:goldenfile", "dep:rstest", "dep:rstest_reuse"]
serde = ["dep:serde", "vortex-buffer/serde", "vortex-mask/serde"]
Expand Down
3 changes: 0 additions & 3 deletions vortex-array/src/dtype/bigint/mod.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -39,9 +39,6 @@ use vortex_error::VortexExpect;
#[derive(Debug, Copy, Clone, Default, Eq, PartialEq, Hash, PartialOrd, Ord)]
pub struct i256(arrow_buffer::i256);

#[cfg(feature = "cudarc")]
unsafe impl cudarc::driver::DeviceRepr for i256 {}

#[expect(
clippy::same_name_method,
reason = "inherent methods intentionally shadow arrow_buffer::i256 methods"
Expand Down
4 changes: 2 additions & 2 deletions vortex-cuda/Cargo.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -23,7 +23,6 @@ _test-harness = []
[dependencies]
arrow-schema = { workspace = true, features = ["ffi"] }
async-trait = { workspace = true }
cudarc = { workspace = true, features = ["f16"] }
futures = { workspace = true, features = ["executor"] }
itertools = { workspace = true }
kanal = { workspace = true }
Expand All @@ -36,9 +35,10 @@ prost = { workspace = true }
tokio = { workspace = true, features = ["fs"] }
tracing = { workspace = true, features = ["std", "attributes"] }
vortex = { workspace = true, features = ["zstd"] }
vortex-array = { workspace = true, features = ["cudarc"] }
vortex-array = { workspace = true }
vortex-arrow = { workspace = true }
vortex-cub = { path = "cub" }
vortex-cuda-driver = { workspace = true, features = ["f16", "i256"] }
vortex-cuda-macros = { workspace = true }
vortex-error = { workspace = true, features = ["object_store"] }
vortex-nvcomp = { path = "nvcomp" }
Expand Down
2 changes: 1 addition & 1 deletion vortex-cuda/benches/alp_cuda.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,7 +19,6 @@ use std::time::Duration;
use criterion::BenchmarkId;
use criterion::Criterion;
use criterion::Throughput;
use cudarc::driver::DeviceRepr;
use futures::executor::block_on;
use mimalloc::MiMalloc;
use vortex::array::IntoArray;
Expand All @@ -43,6 +42,7 @@ use vortex_array::session::ArraySession;
use vortex_cuda::CudaDispatchMode;
use vortex_cuda::CudaSession;
use vortex_cuda::executor::CudaArrayExt;
use vortex_cuda_driver::DeviceRepr;
use vortex_cuda_macros::cuda_available;
use vortex_cuda_macros::cuda_not_available;

Expand Down
2 changes: 1 addition & 1 deletion vortex-cuda/benches/bitpacked_cuda.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,7 +19,6 @@ use std::time::Duration;
use criterion::BenchmarkId;
use criterion::Criterion;
use criterion::Throughput;
use cudarc::driver::DeviceRepr;
use futures::executor::block_on;
use mimalloc::MiMalloc;
use vortex::array::IntoArray;
Expand All @@ -42,6 +41,7 @@ use vortex_array::session::ArraySession;
use vortex_cuda::CudaDispatchMode;
use vortex_cuda::CudaSession;
use vortex_cuda::executor::CudaArrayExt;
use vortex_cuda_driver::DeviceRepr;
use vortex_cuda_macros::cuda_available;
use vortex_cuda_macros::cuda_not_available;

Expand Down
20 changes: 10 additions & 10 deletions vortex-cuda/benches/core_primitives.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,15 +15,15 @@ use criterion::BatchSize;
use criterion::BenchmarkId;
use criterion::Criterion;
use criterion::Throughput;
use cudarc::driver::CudaContext;
use cudarc::driver::CudaStream;
use cudarc::driver::HostSlice;
use cudarc::driver::SyncOnDrop;
use cudarc::driver::result;
use cudarc::driver::sys;
use cudarc::driver::sys::CU_MEMHOSTALLOC_WRITECOMBINED;
use cudarc::driver::sys::CUmemPool_attribute_enum;
use mimalloc::MiMalloc;
use vortex_cuda_driver::CudaContext;
use vortex_cuda_driver::CudaStream;
use vortex_cuda_driver::HostSlice;
use vortex_cuda_driver::SyncOnDrop;
use vortex_cuda_driver::result;
use vortex_cuda_driver::sys;
use vortex_cuda_driver::sys::CU_MEMHOSTALLOC_WRITECOMBINED;
use vortex_cuda_driver::sys::CUmemPool_attribute_enum;
use vortex_cuda_macros::cuda_available;
use vortex_cuda_macros::cuda_not_available;

Expand Down Expand Up @@ -92,7 +92,7 @@ impl HostSlice<u8> for CudaHostBuffer {
) -> (&'a [u8], SyncOnDrop<'a>) {
(
unsafe { std::slice::from_raw_parts(self.ptr, self.len) },
SyncOnDrop::Sync(None),
SyncOnDrop::none(),
)
}

Expand All @@ -102,7 +102,7 @@ impl HostSlice<u8> for CudaHostBuffer {
) -> (&'a mut [u8], SyncOnDrop<'a>) {
(
unsafe { std::slice::from_raw_parts_mut(self.ptr, self.len) },
SyncOnDrop::Sync(None),
SyncOnDrop::none(),
)
}
}
Expand Down
2 changes: 1 addition & 1 deletion vortex-cuda/benches/delta_cuda.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,7 +21,6 @@ use std::time::Duration;
use criterion::BenchmarkId;
use criterion::Criterion;
use criterion::Throughput;
use cudarc::driver::DeviceRepr;
use futures::executor::block_on;
use mimalloc::MiMalloc;
use vortex::array::IntoArray;
Expand All @@ -43,6 +42,7 @@ use vortex_array::session::ArraySession;
use vortex_cuda::CudaDispatchMode;
use vortex_cuda::CudaSession;
use vortex_cuda::executor::CudaArrayExt;
use vortex_cuda_driver::DeviceRepr;
use vortex_cuda_macros::cuda_available;
use vortex_cuda_macros::cuda_not_available;

Expand Down
2 changes: 1 addition & 1 deletion vortex-cuda/benches/dict_cuda.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,7 +18,6 @@ use std::time::Duration;
use criterion::BenchmarkId;
use criterion::Criterion;
use criterion::Throughput;
use cudarc::driver::DeviceRepr;
use futures::executor::block_on;
use mimalloc::MiMalloc;
use vortex::array::IntoArray;
Expand All @@ -31,6 +30,7 @@ use vortex::error::VortexExpect;
use vortex_cuda::CudaDispatchMode;
use vortex_cuda::CudaSession;
use vortex_cuda::executor::CudaArrayExt;
use vortex_cuda_driver::DeviceRepr;
use vortex_cuda_macros::cuda_available;
use vortex_cuda_macros::cuda_not_available;

Expand Down
14 changes: 7 additions & 7 deletions vortex-cuda/benches/dynamic_dispatch_cuda.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,13 +18,6 @@ use std::time::Duration;
use criterion::BenchmarkId;
use criterion::Criterion;
use criterion::Throughput;
use cudarc::driver::CudaFunction;
use cudarc::driver::CudaSlice;
use cudarc::driver::DevicePtr;
use cudarc::driver::DeviceRepr;
use cudarc::driver::LaunchConfig;
use cudarc::driver::PushKernelArg;
use cudarc::driver::sys::CUevent_flags;
use futures::executor::block_on;
use mimalloc::MiMalloc;
use vortex::array::ArrayRef;
Expand Down Expand Up @@ -76,6 +69,13 @@ use vortex_cuda::cuda_session;
use vortex_cuda::dynamic_dispatch::CudaDispatchPlan;
use vortex_cuda::dynamic_dispatch::DispatchPlan;
use vortex_cuda::dynamic_dispatch::MaterializedPlan;
use vortex_cuda_driver::CudaFunction;
use vortex_cuda_driver::CudaSlice;
use vortex_cuda_driver::DevicePtr;
use vortex_cuda_driver::DeviceRepr;
use vortex_cuda_driver::LaunchConfig;
use vortex_cuda_driver::PushKernelArg;
use vortex_cuda_driver::sys::CUevent_flags;
use vortex_cuda_macros::cuda_available;
use vortex_cuda_macros::cuda_not_available;

Expand Down
14 changes: 7 additions & 7 deletions vortex-cuda/benches/filter_cuda.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,13 +16,6 @@ use std::time::Duration;
use criterion::BenchmarkId;
use criterion::Criterion;
use criterion::Throughput;
use cudarc::driver::CudaSlice;
use cudarc::driver::CudaView;
use cudarc::driver::DevicePtr;
use cudarc::driver::DevicePtrMut;
use cudarc::driver::DeviceRepr;
use cudarc::driver::ValidAsZeroBits;
use cudarc::driver::sys::CUevent_flags;
use futures::executor::block_on;
use mimalloc::MiMalloc;
use vortex::error::VortexExpect;
Expand All @@ -33,6 +26,13 @@ use vortex_cub::filter::cudaStream_t;
use vortex_cuda::CudaDeviceBuffer;
use vortex_cuda::CudaExecutionCtx;
use vortex_cuda::CudaSession;
use vortex_cuda_driver::CudaSlice;
use vortex_cuda_driver::CudaView;
use vortex_cuda_driver::DevicePtr;
use vortex_cuda_driver::DevicePtrMut;
use vortex_cuda_driver::DeviceRepr;
use vortex_cuda_driver::ValidAsZeroBits;
use vortex_cuda_driver::sys::CUevent_flags;
use vortex_cuda_macros::cuda_available;
use vortex_cuda_macros::cuda_not_available;

Expand Down
2 changes: 1 addition & 1 deletion vortex-cuda/benches/for_cuda.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,7 +19,6 @@ use std::time::Duration;
use criterion::BenchmarkId;
use criterion::Criterion;
use criterion::Throughput;
use cudarc::driver::DeviceRepr;
use futures::executor::block_on;
use mimalloc::MiMalloc;
use vortex::array::IntoArray;
Expand All @@ -44,6 +43,7 @@ use vortex_array::session::ArraySession;
use vortex_cuda::CudaDispatchMode;
use vortex_cuda::CudaSession;
use vortex_cuda::executor::CudaArrayExt;
use vortex_cuda_driver::DeviceRepr;
use vortex_cuda_macros::cuda_available;
use vortex_cuda_macros::cuda_not_available;

Expand Down
2 changes: 1 addition & 1 deletion vortex-cuda/benches/runend_cuda.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,7 +17,6 @@ use std::time::Duration;
use criterion::BenchmarkId;
use criterion::Criterion;
use criterion::Throughput;
use cudarc::driver::DeviceRepr;
use futures::executor::block_on;
use mimalloc::MiMalloc;
use vortex::array::ExecutionCtx;
Expand All @@ -30,6 +29,7 @@ use vortex::encodings::runend::RunEnd;
use vortex::encodings::runend::RunEndArray;
use vortex_cuda::CudaSession;
use vortex_cuda::executor::CudaArrayExt;
use vortex_cuda_driver::DeviceRepr;
use vortex_cuda_macros::cuda_available;
use vortex_cuda_macros::cuda_not_available;

Expand Down
8 changes: 4 additions & 4 deletions vortex-cuda/benches/throughput_cuda.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,13 +17,13 @@ use std::time::Duration;
use criterion::BenchmarkId;
use criterion::Criterion;
use criterion::Throughput;
use cudarc::driver::DevicePtr;
use cudarc::driver::DevicePtrMut;
use cudarc::driver::sys;
use cudarc::driver::sys::CUevent_flags;
use mimalloc::MiMalloc;
use vortex_cuda::CudaExecutionCtx;
use vortex_cuda::CudaSession;
use vortex_cuda_driver::DevicePtr;
use vortex_cuda_driver::DevicePtrMut;
use vortex_cuda_driver::sys;
use vortex_cuda_driver::sys::CUevent_flags;
use vortex_cuda_macros::cuda_available;
use vortex_cuda_macros::cuda_not_available;

Expand Down
Loading
Loading