Fix CUDA kernel compilation with new linker and higher recursion limit

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/juntyr/rust-cuda", rev = "b26635f", features = ["derive"], optional = true }
+rust-cuda = { git = "https://github.com/juntyr/rust-cuda", rev = "6b53e88", features = ["derive"], optional = true }
 
 [target.'cfg(not(target_os = "cuda"))'.dependencies]
-rust-cuda = { git = "https://github.com/juntyr/rust-cuda", rev = "b26635f", features = ["derive", "host"], optional = true }
+rust-cuda = { git = "https://github.com/juntyr/rust-cuda", rev = "6b53e88", features = ["derive", "host"], optional = true }

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/juntyr/rust-cuda", rev = "b26635f", features = ["derive"] }
+rust-cuda = { git = "https://github.com/juntyr/rust-cuda", rev = "6b53e88", features = ["derive"] }
 
 [target.'cfg(not(target_os = "cuda"))'.dependencies]
-rust-cuda = { git = "https://github.com/juntyr/rust-cuda", rev = "b26635f", features = ["derive", "host"] }
+rust-cuda = { git = "https://github.com/juntyr/rust-cuda", rev = "6b53e88", features = ["derive", "host"] }

necsim/impls/no-std/Cargo.toml

@@ -30,7 +30,7 @@ fnv = { version = "1.0", default-features = false, features = [] }
 rand_core = "0.6"
 
 [target.'cfg(target_os = "cuda")'.dependencies]
-rust-cuda = { git = "https://github.com/juntyr/rust-cuda", rev = "b26635f", features = ["derive"], optional = true }
+rust-cuda = { git = "https://github.com/juntyr/rust-cuda", rev = "6b53e88", features = ["derive"], optional = true }
 
 [target.'cfg(not(target_os = "cuda"))'.dependencies]
-rust-cuda = { git = "https://github.com/juntyr/rust-cuda", rev = "b26635f", features = ["derive", "host"], optional = true }
+rust-cuda = { git = "https://github.com/juntyr/rust-cuda", rev = "6b53e88", features = ["derive", "host"], optional = true }

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/juntyr/rust-cuda", rev = "b26635f", features = ["host"] }
+rust-cuda = { git = "https://github.com/juntyr/rust-cuda", rev = "6b53e88", features = ["host"] }

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/juntyr/rust-cuda", rev = "b26635f", features = ["host"] }
+rust-cuda = { git = "https://github.com/juntyr/rust-cuda", rev = "6b53e88", features = ["host"] }

rustcoalescence/algorithms/cuda/cpu-kernel/src/lib.rs

@@ -1,6 +1,7 @@
 #![deny(clippy::pedantic)]
 #![allow(incomplete_features)]
 #![feature(specialization)]
+#![recursion_limit = "1024"]
 
 use necsim_core::{
     cogs::{

rustcoalescence/algorithms/cuda/gpu-kernel/.cargo/config.toml

@@ -2,7 +2,7 @@
 pipelining = false
 
 [target.nvptx64-nvidia-cuda]
-rustflags = ["-Clink-args=--arch=sm_35", "-Cpanic=abort", "-Clinker-plugin-lto", "-Ccodegen-units=1", "-Clink-arg=-Olto"]
+rustflags = ["-Clink-args=--arch=sm_35", "-Cpanic=abort", "-Clinker-plugin-lto", "-Ccodegen-units=1", "-Clink-arg=-O3", "-Clink-arg=--lto"]
 
 [unstable]
 build-std = ["core", "alloc"]

rustcoalescence/algorithms/cuda/gpu-kernel/Cargo.toml

@@ -16,4 +16,4 @@ necsim-core-bond = { path = "../../../../necsim/core/bond" }
 necsim-impls-no-std = { path = "../../../../necsim/impls/no-std", features = ["cuda"] }
 necsim-impls-cuda = { path = "../../../../necsim/impls/cuda" }
 
-rust-cuda = { git = "https://github.com/juntyr/rust-cuda", rev = "b26635f", features = ["derive"] }
+rust-cuda = { git = "https://github.com/juntyr/rust-cuda", rev = "6b53e88", features = ["derive"] }

rustcoalescence/algorithms/cuda/gpu-kernel/src/lib.rs

@@ -7,6 +7,8 @@
 #![cfg_attr(target_os = "cuda", feature(asm_experimental_arch))]
 #![cfg_attr(target_os = "cuda", feature(stdsimd))]
 #![cfg_attr(target_os = "cuda", feature(control_flow_enum))]
+#![allow(long_running_const_eval)]
+#![recursion_limit = "1024"]
 
 extern crate alloc;
 

rustcoalescence/algorithms/cuda/src/initialiser/fixup.rs

@@ -126,7 +126,7 @@ where
                     .extend(out_of_deme.into_iter().map(ExceptionalLineage::OutOfDeme));
             },
             OutOfDemeStrategy::Dispersal => {
-                fixable_lineages.extend(out_of_deme.into_iter());
+                fixable_lineages.extend(out_of_deme);
             },
         }
 
@@ -139,7 +139,7 @@ where
                 );
             },
             OutOfHabitatStrategy::UniformDispersal => {
-                fixable_lineages.extend(out_of_habitat.into_iter());
+                fixable_lineages.extend(out_of_habitat);
             },
         }
 

rustcoalescence/algorithms/cuda/src/launch.rs

@@ -230,7 +230,7 @@ where
             steps,
             lineages: lineages
                 .into_iter()
-                .chain(passthrough.into_iter())
+                .chain(passthrough)
                 .collect(),
             rng: simulation.rng_mut().clone(),
             marker: PhantomData::<M>,