Pipeline CUDA mem transfer with async

juntyr · juntyr · commit c0f0205a83d7 · 2022-11-20T01:11:15.000-08:00
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/necsim/core/Cargo.toml b/necsim/core/Cargo.toml
@@ -20,7 +20,7 @@ contracts = "0.6.3"
 serde = { version = "1.0", default-features = false, features = ["derive"] }
 
 [target.'cfg(target_os = "cuda")'.dependencies]
-rust-cuda = { git = "https://github.com/MomoLangenstein/rust-cuda", rev = "0befd6d", features = ["derive"], optional = true }
+rust-cuda = { git = "https://github.com/MomoLangenstein/rust-cuda", rev = "f18ae16", features = ["derive"], optional = true }
 
 [target.'cfg(not(target_os = "cuda"))'.dependencies]
-rust-cuda = { git = "https://github.com/MomoLangenstein/rust-cuda", rev = "0befd6d", features = ["derive", "host"], optional = true }
+rust-cuda = { git = "https://github.com/MomoLangenstein/rust-cuda", rev = "f18ae16", features = ["derive", "host"], optional = true }
diff --git a/necsim/impls/cuda/Cargo.toml b/necsim/impls/cuda/Cargo.toml
@@ -15,7 +15,7 @@ contracts = "0.6.3"
 serde = { version = "1.0", default-features = false, features = ["derive"] }
 
 [target.'cfg(target_os = "cuda")'.dependencies]
-rust-cuda = { git = "https://github.com/MomoLangenstein/rust-cuda", rev = "0befd6d", features = ["derive"] }
+rust-cuda = { git = "https://github.com/MomoLangenstein/rust-cuda", rev = "f18ae16", features = ["derive"] }
 
 [target.'cfg(not(target_os = "cuda"))'.dependencies]
-rust-cuda = { git = "https://github.com/MomoLangenstein/rust-cuda", rev = "0befd6d", features = ["derive", "host"] }
+rust-cuda = { git = "https://github.com/MomoLangenstein/rust-cuda", rev = "f18ae16", features = ["derive", "host"] }
diff --git a/necsim/impls/no-std/Cargo.toml b/necsim/impls/no-std/Cargo.toml
@@ -31,7 +31,7 @@ rand_core = "0.6"
 rand_distr = { version = "0.4", default-features = false, features = [] }
 
 [target.'cfg(target_os = "cuda")'.dependencies]
-rust-cuda = { git = "https://github.com/MomoLangenstein/rust-cuda", rev = "0befd6d", features = ["derive"], optional = true }
+rust-cuda = { git = "https://github.com/MomoLangenstein/rust-cuda", rev = "f18ae16", features = ["derive"], optional = true }
 
 [target.'cfg(not(target_os = "cuda"))'.dependencies]
-rust-cuda = { git = "https://github.com/MomoLangenstein/rust-cuda", rev = "0befd6d", features = ["derive", "host"], optional = true }
+rust-cuda = { git = "https://github.com/MomoLangenstein/rust-cuda", rev = "f18ae16", features = ["derive", "host"], optional = true }
diff --git a/necsim/impls/no-std/src/array2d.rs b/necsim/impls/no-std/src/array2d.rs
@@ -11,6 +11,7 @@ use core::ops::{Index, IndexMut};
 /// A fixed sized two-dimensional array.
 #[derive(Clone, Eq, PartialEq)]
 #[cfg_attr(feature = "cuda", derive(rust_cuda::common::LendRustToCuda))]
+#[cfg_attr(feature = "cuda", cuda(async = false))]
 #[cfg_attr(
     feature = "cuda",
     cuda(
diff --git a/necsim/impls/no-std/src/cogs/dispersal_sampler/in_memory/packed_alias/mod.rs b/necsim/impls/no-std/src/cogs/dispersal_sampler/in_memory/packed_alias/mod.rs
@@ -47,6 +47,7 @@ impl From<AliasSamplerRange> for Range<usize> {
 #[allow(clippy::module_name_repetitions)]
 #[cfg_attr(feature = "cuda", derive(rust_cuda::common::LendRustToCuda))]
 #[cfg_attr(feature = "cuda", cuda(free = "M", free = "H", free = "G"))]
+#[cfg_attr(feature = "cuda", cuda(async = false))]
 pub struct InMemoryPackedAliasDispersalSampler<
     M: MathsCore,
     H: Habitat<M>,
diff --git a/necsim/impls/no-std/src/cogs/habitat/in_memory.rs b/necsim/impls/no-std/src/cogs/habitat/in_memory.rs
@@ -19,6 +19,7 @@ use crate::array2d::Array2D;
 #[derive(Debug)]
 #[cfg_attr(feature = "cuda", derive(rust_cuda::common::LendRustToCuda))]
 #[cfg_attr(feature = "cuda", cuda(free = "M"))]
+#[cfg_attr(feature = "cuda", cuda(async = false))]
 pub struct InMemoryHabitat<M: MathsCore> {
     #[cfg_attr(feature = "cuda", cuda(embed))]
     habitat: Final<Box<[u32]>>,
diff --git a/necsim/impls/no-std/src/cogs/turnover_rate/in_memory.rs b/necsim/impls/no-std/src/cogs/turnover_rate/in_memory.rs
@@ -13,6 +13,7 @@ use crate::{array2d::Array2D, cogs::habitat::in_memory::InMemoryHabitat};
 #[allow(clippy::module_name_repetitions)]
 #[derive(Debug)]
 #[cfg_attr(feature = "cuda", derive(rust_cuda::common::LendRustToCuda))]
+#[cfg_attr(feature = "cuda", cuda(async = false))]
 pub struct InMemoryTurnoverRate {
     #[cfg_attr(feature = "cuda", cuda(embed))]
     turnover_rate: Final<Box<[NonNegativeF64]>>,
diff --git a/rustcoalescence/algorithms/cuda/Cargo.toml b/rustcoalescence/algorithms/cuda/Cargo.toml
@@ -23,4 +23,4 @@ thiserror = "1.0"
 serde = { version = "1.0", features = ["derive"] }
 serde_state = "0.4"
 serde_derive_state = "0.4"
-rust-cuda = { git = "https://github.com/MomoLangenstein/rust-cuda", rev = "0befd6d", features = ["host"] }
+rust-cuda = { git = "https://github.com/MomoLangenstein/rust-cuda", rev = "f18ae16", features = ["host"] }
diff --git a/rustcoalescence/algorithms/cuda/cpu-kernel/Cargo.toml b/rustcoalescence/algorithms/cuda/cpu-kernel/Cargo.toml
@@ -14,4 +14,4 @@ necsim-impls-no-std = { path = "../../../../necsim/impls/no-std", features = ["c
 necsim-impls-cuda = { path = "../../../../necsim/impls/cuda" }
 rustcoalescence-algorithms-cuda-gpu-kernel = { path = "../gpu-kernel" }
 
-rust-cuda = { git = "https://github.com/MomoLangenstein/rust-cuda", rev = "0befd6d", features = ["host"] }
+rust-cuda = { git = "https://github.com/MomoLangenstein/rust-cuda", rev = "f18ae16", features = ["host"] }
diff --git a/rustcoalescence/algorithms/cuda/cpu-kernel/src/lib.rs b/rustcoalescence/algorithms/cuda/cpu-kernel/src/lib.rs
@@ -20,11 +20,10 @@ use necsim_impls_no_std::cogs::{
 
 use rust_cuda::{
     common::RustToCuda,
-    host::{CudaDropWrapper, LaunchConfig, LaunchPackage, Launcher, TypedKernel},
+    host::{LaunchConfig, LaunchPackage, Launcher, TypedKernel},
     rustacuda::{
         error::CudaResult,
         function::{BlockSize, Function, GridSize},
-        stream::Stream,
     },
 };
 
@@ -71,7 +70,6 @@ pub struct SimulationKernel<
             ReportDispersal,
         >,
     >,
-    stream: CudaDropWrapper<Stream>,
     grid: GridSize,
     block: BlockSize,
     ptx_jit: bool,
@@ -99,7 +97,6 @@ impl<
     ///
     /// Returns a `CudaError` if loading the CUDA kernel failed.
     pub fn try_new(
-        stream: Stream,
         grid: GridSize,
         block: BlockSize,
         ptx_jit: bool,
@@ -123,12 +120,10 @@ impl<
             ReportDispersal,
         >,
     {
-        let stream = CudaDropWrapper::from(stream);
         let kernel = Self::new_kernel()?;
 
         Ok(Self {
             kernel,
-            stream,
             grid,
             block,
             ptx_jit,
@@ -181,10 +176,7 @@ impl<
                 shared_memory_size: 0_u32,
                 ptx_jit: self.ptx_jit,
             },
-
             kernel: &mut self.kernel,
-            stream: &mut self.stream,
-
             watcher: &mut self.watcher,
         }
     }
diff --git a/rustcoalescence/algorithms/cuda/cpu-kernel/src/patch.rs b/rustcoalescence/algorithms/cuda/cpu-kernel/src/patch.rs
diff --git a/rustcoalescence/algorithms/cuda/gpu-kernel/Cargo.toml b/rustcoalescence/algorithms/cuda/gpu-kernel/Cargo.toml
diff --git a/rustcoalescence/algorithms/cuda/src/launch.rs b/rustcoalescence/algorithms/cuda/src/launch.rs
diff --git a/rustcoalescence/algorithms/cuda/src/parallelisation/monolithic.rs b/rustcoalescence/algorithms/cuda/src/parallelisation/monolithic.rs
