Auto merge of #144222 - Kobzol:stdarch-push, r=folkertdev

stdarch subtree update

Subtree update of `stdarch` to https://github.com/rust-lang/stdarch/commit/5531955678494ee28ec02130a6d94082ad4532da.

Created using https://github.com/rust-lang/josh-sync.

I saw that there were non-trivial changes made to `std_detect` in `stdarch` recently. So I want to get them merged here before we move forward with https://github.com/rust-lang/rust/pull/143412.

r? `@folkertdev`

15 files changed, 500 insertions, 477 deletions

diff --git a/library/stdarch/crates/intrinsic-test/src/arm/compile.rs b/library/stdarch/crates/intrinsic-test/src/arm/compile.rs
index 8276cd87c1c..48a8ed950e3 100644
--- a/library/stdarch/crates/intrinsic-test/src/arm/compile.rs
+++ b/library/stdarch/crates/intrinsic-test/src/arm/compile.rs
@@ -1,64 +1,51 @@
-use crate::common::compile_c::CompilationCommandBuilder;
-use crate::common::gen_c::compile_c_programs;
+use crate::common::cli::ProcessedCli;
+use crate::common::compile_c::{CompilationCommandBuilder, CppCompilation};
+
+pub fn build_cpp_compilation(config: &ProcessedCli) -> Option<CppCompilation> {
+    let cpp_compiler = config.cpp_compiler.as_ref()?;
 
-pub fn compile_c_arm(
-    intrinsics_name_list: &[String],
-    compiler: &str,
-    target: &str,
-    cxx_toolchain_dir: Option<&str>,
-) -> bool {
     // -ffp-contract=off emulates Rust's approach of not fusing separate mul-add operations
     let mut command = CompilationCommandBuilder::new()
         .add_arch_flags(vec!["armv8.6-a", "crypto", "crc", "dotprod", "fp16"])
-        .set_compiler(compiler)
-        .set_target(target)
+        .set_compiler(cpp_compiler)
+        .set_target(&config.target)
         .set_opt_level("2")
-        .set_cxx_toolchain_dir(cxx_toolchain_dir)
+        .set_cxx_toolchain_dir(config.cxx_toolchain_dir.as_deref())
         .set_project_root("c_programs")
         .add_extra_flags(vec!["-ffp-contract=off", "-Wno-narrowing"]);
 
-    if !target.contains("v7") {
+    if !config.target.contains("v7") {
         command = command.add_arch_flags(vec!["faminmax", "lut", "sha3"]);
     }
 
-    /*
-     * clang++ cannot link an aarch64_be object file, so we invoke
-     * aarch64_be-unknown-linux-gnu's C++ linker. This ensures that we
-     * are testing the intrinsics against LLVM.
-     *
-     * Note: setting `--sysroot=<...>` which is the obvious thing to do
-     * does not work as it gets caught up with `#include_next <stdlib.h>`
-     * not existing...
-     */
-    if target.contains("aarch64_be") {
-        command = command
-            .set_linker(
-                cxx_toolchain_dir.unwrap_or("").to_string() + "/bin/aarch64_be-none-linux-gnu-g++",
-            )
-            .set_include_paths(vec![
-                "/include",
-                "/aarch64_be-none-linux-gnu/include",
-                "/aarch64_be-none-linux-gnu/include/c++/14.2.1",
-                "/aarch64_be-none-linux-gnu/include/c++/14.2.1/aarch64_be-none-linux-gnu",
-                "/aarch64_be-none-linux-gnu/include/c++/14.2.1/backward",
-                "/aarch64_be-none-linux-gnu/libc/usr/include",
-            ]);
-    }
-
-    if !compiler.contains("clang") {
+    if !cpp_compiler.contains("clang") {
         command = command.add_extra_flag("-flax-vector-conversions");
     }
 
-    let compiler_commands = intrinsics_name_list
-        .iter()
-        .map(|intrinsic_name| {
-            command
-                .clone()
-                .set_input_name(intrinsic_name)
-                .set_output_name(intrinsic_name)
-                .make_string()
-        })
-        .collect::<Vec<_>>();
+    let mut cpp_compiler = command.into_cpp_compilation();
+
+    if config.target.contains("aarch64_be") {
+        let Some(ref cxx_toolchain_dir) = config.cxx_toolchain_dir else {
+            panic!(
+                "target `{}` must specify `cxx_toolchain_dir`",
+                config.target
+            )
+        };
+
+        cpp_compiler.command_mut().args([
+            &format!("--sysroot={cxx_toolchain_dir}/aarch64_be-none-linux-gnu/libc"),
+            "--include-directory",
+            &format!("{cxx_toolchain_dir}/aarch64_be-none-linux-gnu/include/c++/14.3.1"),
+            "--include-directory",
+            &format!("{cxx_toolchain_dir}/aarch64_be-none-linux-gnu/include/c++/14.3.1/aarch64_be-none-linux-gnu"),
+            "-L",
+            &format!("{cxx_toolchain_dir}/lib/gcc/aarch64_be-none-linux-gnu/14.3.1"),
+            "-L",
+            &format!("{cxx_toolchain_dir}/aarch64_be-none-linux-gnu/libc/usr/lib"),
+            "-B",
+            &format!("{cxx_toolchain_dir}/lib/gcc/aarch64_be-none-linux-gnu/14.3.1"),
+        ]);
+    }
 
-    compile_c_programs(&compiler_commands)
+    Some(cpp_compiler)
 }
diff --git a/library/stdarch/crates/intrinsic-test/src/arm/config.rs b/library/stdarch/crates/intrinsic-test/src/arm/config.rs
index cee80374ae9..9a7b37253d1 100644
--- a/library/stdarch/crates/intrinsic-test/src/arm/config.rs
+++ b/library/stdarch/crates/intrinsic-test/src/arm/config.rs
@@ -114,7 +114,6 @@ pub const AARCH_CONFIGURATIONS: &str = r#"
 #![cfg_attr(any(target_arch = "aarch64", target_arch = "arm64ec"), feature(stdarch_neon_fcma))]
 #![cfg_attr(any(target_arch = "aarch64", target_arch = "arm64ec"), feature(stdarch_neon_dotprod))]
 #![cfg_attr(any(target_arch = "aarch64", target_arch = "arm64ec"), feature(stdarch_neon_i8mm))]
-#![cfg_attr(any(target_arch = "aarch64", target_arch = "arm64ec"), feature(stdarch_neon_sha3))]
 #![cfg_attr(any(target_arch = "aarch64", target_arch = "arm64ec"), feature(stdarch_neon_sm4))]
 #![cfg_attr(any(target_arch = "aarch64", target_arch = "arm64ec"), feature(stdarch_neon_ftts))]
 #![feature(fmt_helpers_for_derive)]
diff --git a/library/stdarch/crates/intrinsic-test/src/arm/intrinsic.rs b/library/stdarch/crates/intrinsic-test/src/arm/intrinsic.rs
index 773dabf4d75..16572b2c03f 100644
--- a/library/stdarch/crates/intrinsic-test/src/arm/intrinsic.rs
+++ b/library/stdarch/crates/intrinsic-test/src/arm/intrinsic.rs
@@ -1,8 +1,8 @@
 use crate::common::argument::ArgumentList;
 use crate::common::indentation::Indentation;
 use crate::common::intrinsic::{Intrinsic, IntrinsicDefinition};
-use crate::common::intrinsic_helpers::{IntrinsicType, IntrinsicTypeDefinition, TypeKind};
-use std::ops::Deref;
+use crate::common::intrinsic_helpers::{IntrinsicType, IntrinsicTypeDefinition, Sign, TypeKind};
+use std::ops::{Deref, DerefMut};
 
 #[derive(Debug, Clone, PartialEq)]
 pub struct ArmIntrinsicType(pub IntrinsicType);
@@ -15,6 +15,12 @@ impl Deref for ArmIntrinsicType {
     }
 }
 
+impl DerefMut for ArmIntrinsicType {
+    fn deref_mut(&mut self) -> &mut Self::Target {
+        &mut self.0
+    }
+}
+
 impl IntrinsicDefinition<ArmIntrinsicType> for Intrinsic<ArmIntrinsicType> {
     fn arguments(&self) -> ArgumentList<ArmIntrinsicType> {
         self.arguments.clone()
@@ -73,8 +79,9 @@ impl IntrinsicDefinition<ArmIntrinsicType> for Intrinsic<ArmIntrinsicType> {
                     TypeKind::Float if self.results().inner_size() == 16 => "float16_t".to_string(),
                     TypeKind::Float if self.results().inner_size() == 32 => "float".to_string(),
                     TypeKind::Float if self.results().inner_size() == 64 => "double".to_string(),
-                    TypeKind::Int => format!("int{}_t", self.results().inner_size()),
-                    TypeKind::UInt => format!("uint{}_t", self.results().inner_size()),
+                    TypeKind::Int(Sign::Signed) => format!("int{}_t", self.results().inner_size()),
+                    TypeKind::Int(Sign::Unsigned) =>
+                        format!("uint{}_t", self.results().inner_size()),
                     TypeKind::Poly => format!("poly{}_t", self.results().inner_size()),
                     ty => todo!("print_result_c - Unknown type: {:#?}", ty),
                 },
diff --git a/library/stdarch/crates/intrinsic-test/src/arm/json_parser.rs b/library/stdarch/crates/intrinsic-test/src/arm/json_parser.rs
index 0ac47484b01..58d366c86a9 100644
--- a/library/stdarch/crates/intrinsic-test/src/arm/json_parser.rs
+++ b/library/stdarch/crates/intrinsic-test/src/arm/json_parser.rs
@@ -110,7 +110,7 @@ fn json_to_intrinsic(
     Ok(Intrinsic {
         name,
         arguments,
-        results: *results,
+        results: results,
         arch_tags: intr.architectures,
     })
 }
diff --git a/library/stdarch/crates/intrinsic-test/src/arm/mod.rs b/library/stdarch/crates/intrinsic-test/src/arm/mod.rs
index 6aaa49ff97f..0a64a24e731 100644
--- a/library/stdarch/crates/intrinsic-test/src/arm/mod.rs
+++ b/library/stdarch/crates/intrinsic-test/src/arm/mod.rs
@@ -4,15 +4,20 @@ mod intrinsic;
 mod json_parser;
 mod types;
 
+use std::fs::File;
+
+use rayon::prelude::*;
+
+use crate::arm::config::POLY128_OSTREAM_DEF;
 use crate::common::SupportedArchitectureTest;
 use crate::common::cli::ProcessedCli;
 use crate::common::compare::compare_outputs;
+use crate::common::gen_c::{write_main_cpp, write_mod_cpp};
 use crate::common::gen_rust::compile_rust_programs;
 use crate::common::intrinsic::{Intrinsic, IntrinsicDefinition};
 use crate::common::intrinsic_helpers::TypeKind;
-use crate::common::write_file::{write_c_testfiles, write_rust_testfiles};
-use compile::compile_c_arm;
-use config::{AARCH_CONFIGURATIONS, F16_FORMATTING_DEF, POLY128_OSTREAM_DEF, build_notices};
+use crate::common::write_file::write_rust_testfiles;
+use config::{AARCH_CONFIGURATIONS, F16_FORMATTING_DEF, build_notices};
 use intrinsic::ArmIntrinsicType;
 use json_parser::get_neon_intrinsics;
 
@@ -21,6 +26,13 @@ pub struct ArmArchitectureTest {
     cli_options: ProcessedCli,
 }
 
+fn chunk_info(intrinsic_count: usize) -> (usize, usize) {
+    let available_parallelism = std::thread::available_parallelism().unwrap().get();
+    let chunk_size = intrinsic_count.div_ceil(Ord::min(available_parallelism, intrinsic_count));
+
+    (chunk_size, intrinsic_count.div_ceil(chunk_size))
+}
+
 impl SupportedArchitectureTest for ArmArchitectureTest {
     fn create(cli_options: ProcessedCli) -> Box<Self> {
         let a32 = cli_options.target.contains("v7");
@@ -51,33 +63,58 @@ impl SupportedArchitectureTest for ArmArchitectureTest {
     }
 
     fn build_c_file(&self) -> bool {
-        let compiler = self.cli_options.cpp_compiler.as_deref();
-        let target = &self.cli_options.target;
-        let cxx_toolchain_dir = self.cli_options.cxx_toolchain_dir.as_deref();
         let c_target = "aarch64";
+        let platform_headers = &["arm_neon.h", "arm_acle.h", "arm_fp16.h"];
 
-        let intrinsics_name_list = write_c_testfiles(
-            &self
-                .intrinsics
-                .iter()
-                .map(|i| i as &dyn IntrinsicDefinition<_>)
-                .collect::<Vec<_>>(),
-            target,
+        let (chunk_size, chunk_count) = chunk_info(self.intrinsics.len());
+
+        let cpp_compiler = compile::build_cpp_compilation(&self.cli_options).unwrap();
+
+        let notice = &build_notices("// ");
+        self.intrinsics
+            .par_chunks(chunk_size)
+            .enumerate()
+            .map(|(i, chunk)| {
+                let c_filename = format!("c_programs/mod_{i}.cpp");
+                let mut file = File::create(&c_filename).unwrap();
+                write_mod_cpp(&mut file, notice, c_target, platform_headers, chunk).unwrap();
+
+                // compile this cpp file into a .o file
+                let output = cpp_compiler
+                    .compile_object_file(&format!("mod_{i}.cpp"), &format!("mod_{i}.o"))?;
+                assert!(output.status.success(), "{output:?}");
+
+                Ok(())
+            })
+            .collect::<Result<(), std::io::Error>>()
+            .unwrap();
+
+        let mut file = File::create("c_programs/main.cpp").unwrap();
+        write_main_cpp(
+            &mut file,
             c_target,
-            &["arm_neon.h", "arm_acle.h", "arm_fp16.h"],
-            &build_notices("// "),
-            &[POLY128_OSTREAM_DEF],
-        );
+            POLY128_OSTREAM_DEF,
+            self.intrinsics.iter().map(|i| i.name.as_str()),
+        )
+        .unwrap();
 
-        match compiler {
-            None => true,
-            Some(compiler) => compile_c_arm(
-                intrinsics_name_list.as_slice(),
-                compiler,
-                target,
-                cxx_toolchain_dir,
-            ),
-        }
+        // compile this cpp file into a .o file
+        info!("compiling main.cpp");
+        let output = cpp_compiler
+            .compile_object_file("main.cpp", "intrinsic-test-programs.o")
+            .unwrap();
+        assert!(output.status.success(), "{output:?}");
+
+        let object_files = (0..chunk_count)
+            .map(|i| format!("mod_{i}.o"))
+            .chain(["intrinsic-test-programs.o".to_owned()]);
+
+        let output = cpp_compiler
+            .link_executable(object_files, "intrinsic-test-programs")
+            .unwrap();
+        assert!(output.status.success(), "{output:?}");
+
+        true
     }
 
     fn build_rust_file(&self) -> bool {
@@ -104,7 +141,7 @@ impl SupportedArchitectureTest for ArmArchitectureTest {
     }
 
     fn compare_outputs(&self) -> bool {
-        if let Some(ref toolchain) = self.cli_options.toolchain {
+        if self.cli_options.toolchain.is_some() {
             let intrinsics_name_list = self
                 .intrinsics
                 .iter()
@@ -113,8 +150,7 @@ impl SupportedArchitectureTest for ArmArchitectureTest {
 
             compare_outputs(
                 &intrinsics_name_list,
-                toolchain,
-                &self.cli_options.c_runner,
+                &self.cli_options.runner,
                 &self.cli_options.target,
             )
         } else {
diff --git a/library/stdarch/crates/intrinsic-test/src/arm/types.rs b/library/stdarch/crates/intrinsic-test/src/arm/types.rs
index 9f3d6302f46..77f5e8d0e56 100644
--- a/library/stdarch/crates/intrinsic-test/src/arm/types.rs
+++ b/library/stdarch/crates/intrinsic-test/src/arm/types.rs
@@ -1,6 +1,6 @@
 use super::intrinsic::ArmIntrinsicType;
 use crate::common::cli::Language;
-use crate::common::intrinsic_helpers::{IntrinsicType, IntrinsicTypeDefinition, TypeKind};
+use crate::common::intrinsic_helpers::{IntrinsicType, IntrinsicTypeDefinition, Sign, TypeKind};
 
 impl IntrinsicTypeDefinition for ArmIntrinsicType {
     /// Gets a string containing the typename for this type in C format.
@@ -73,8 +73,8 @@ impl IntrinsicTypeDefinition for ArmIntrinsicType {
             format!(
                 "vld{len}{quad}_{type}{size}",
                 type = match k {
-                    TypeKind::UInt => "u",
-                    TypeKind::Int => "s",
+                    TypeKind::Int(Sign::Unsigned) => "u",
+                    TypeKind::Int(Sign::Signed) => "s",
                     TypeKind::Float => "f",
                     // The ACLE doesn't support 64-bit polynomial loads on Armv7
                     // if armv7 and bl == 64, use "s", else "p"
@@ -107,8 +107,8 @@ impl IntrinsicTypeDefinition for ArmIntrinsicType {
             format!(
                 "vget{quad}_lane_{type}{size}",
                 type = match k {
-                    TypeKind::UInt => "u",
-                    TypeKind::Int => "s",
+                    TypeKind::Int(Sign::Unsigned) => "u",
+                    TypeKind::Int(Sign::Signed) => "s",
                     TypeKind::Float => "f",
                     TypeKind::Poly => "p",
                     x => todo!("get_load_function TypeKind: {:#?}", x),
@@ -121,7 +121,7 @@ impl IntrinsicTypeDefinition for ArmIntrinsicType {
         }
     }
 
-    fn from_c(s: &str, target: &str) -> Result<Box<Self>, String> {
+    fn from_c(s: &str, target: &str) -> Result<Self, String> {
         const CONST_STR: &str = "const";
         if let Some(s) = s.strip_suffix('*') {
             let (s, constant) = match s.trim().strip_suffix(CONST_STR) {
@@ -131,9 +131,8 @@ impl IntrinsicTypeDefinition for ArmIntrinsicType {
             let s = s.trim_end();
             let temp_return = ArmIntrinsicType::from_c(s, target);
             temp_return.map(|mut op| {
-                let edited = op.as_mut();
-                edited.0.ptr = true;
-                edited.0.ptr_constant = constant;
+                op.ptr = true;
+                op.ptr_constant = constant;
                 op
             })
         } else {
@@ -163,7 +162,7 @@ impl IntrinsicTypeDefinition for ArmIntrinsicType {
                     ),
                     None => None,
                 };
-                Ok(Box::new(ArmIntrinsicType(IntrinsicType {
+                Ok(ArmIntrinsicType(IntrinsicType {
                     ptr: false,
                     ptr_constant: false,
                     constant,
@@ -172,14 +171,14 @@ impl IntrinsicTypeDefinition for ArmIntrinsicType {
                     simd_len,
                     vec_len,
                     target: target.to_string(),
-                })))
+                }))
             } else {
                 let kind = start.parse::<TypeKind>()?;
                 let bit_len = match kind {
-                    TypeKind::Int => Some(32),
+                    TypeKind::Int(_) => Some(32),
                     _ => None,
                 };
-                Ok(Box::new(ArmIntrinsicType(IntrinsicType {
+                Ok(ArmIntrinsicType(IntrinsicType {
                     ptr: false,
                     ptr_constant: false,
                     constant,
@@ -188,7 +187,7 @@ impl IntrinsicTypeDefinition for ArmIntrinsicType {
                     simd_len: None,
                     vec_len: None,
                     target: target.to_string(),
-                })))
+                }))
             }
         }
     }
diff --git a/library/stdarch/crates/intrinsic-test/src/common/argument.rs b/library/stdarch/crates/intrinsic-test/src/common/argument.rs
index 443ccb919f4..1df4f55995e 100644
--- a/library/stdarch/crates/intrinsic-test/src/common/argument.rs
+++ b/library/stdarch/crates/intrinsic-test/src/common/argument.rs
@@ -76,7 +76,7 @@ where
         Argument {
             pos,
             name: String::from(var_name),
-            ty: *ty,
+            ty: ty,
             constraint,
         }
     }
@@ -125,19 +125,23 @@ where
     /// Creates a line for each argument that initializes an array for C from which `loads` argument
     /// values can be loaded  as a sliding window.
     /// e.g `const int32x2_t a_vals = {0x3effffff, 0x3effffff, 0x3f7fffff}`, if loads=2.
-    pub fn gen_arglists_c(&self, indentation: Indentation, loads: u32) -> String {
-        self.iter()
-            .filter(|&arg| !arg.has_constraint())
-            .map(|arg| {
-                format!(
-                    "{indentation}const {ty} {name}_vals[] = {values};",
-                    ty = arg.ty.c_scalar_type(),
-                    name = arg.name,
-                    values = arg.ty.populate_random(indentation, loads, &Language::C)
-                )
-            })
-            .collect::<Vec<_>>()
-            .join("\n")
+    pub fn gen_arglists_c(
+        &self,
+        w: &mut impl std::io::Write,
+        indentation: Indentation,
+        loads: u32,
+    ) -> std::io::Result<()> {
+        for arg in self.iter().filter(|&arg| !arg.has_constraint()) {
+            writeln!(
+                w,
+                "{indentation}const {ty} {name}_vals[] = {values};",
+                ty = arg.ty.c_scalar_type(),
+                name = arg.name,
+                values = arg.ty.populate_random(indentation, loads, &Language::C)
+            )?
+        }
+
+        Ok(())
     }
 
     /// Creates a line for each argument that initializes an array for Rust from which `loads` argument
diff --git a/library/stdarch/crates/intrinsic-test/src/common/cli.rs b/library/stdarch/crates/intrinsic-test/src/common/cli.rs
index 1d572723008..beae6a4b044 100644
--- a/library/stdarch/crates/intrinsic-test/src/common/cli.rs
+++ b/library/stdarch/crates/intrinsic-test/src/common/cli.rs
@@ -60,7 +60,7 @@ pub struct ProcessedCli {
     pub filename: PathBuf,
     pub toolchain: Option<String>,
     pub cpp_compiler: Option<String>,
-    pub c_runner: String,
+    pub runner: String,
     pub target: String,
     pub linker: Option<String>,
     pub cxx_toolchain_dir: Option<String>,
@@ -70,7 +70,7 @@ pub struct ProcessedCli {
 impl ProcessedCli {
     pub fn new(cli_options: Cli) -> Self {
         let filename = cli_options.input;
-        let c_runner = cli_options.runner.unwrap_or_default();
+        let runner = cli_options.runner.unwrap_or_default();
         let target = cli_options.target;
         let linker = cli_options.linker;
         let cxx_toolchain_dir = cli_options.cxx_toolchain_dir;
@@ -102,7 +102,7 @@ impl ProcessedCli {
         Self {
             toolchain,
             cpp_compiler,
-            c_runner,
+            runner,
             target,
             linker,
             cxx_toolchain_dir,
diff --git a/library/stdarch/crates/intrinsic-test/src/common/compare.rs b/library/stdarch/crates/intrinsic-test/src/common/compare.rs
index 9e0cbe8cd6a..cb55922eb19 100644
--- a/library/stdarch/crates/intrinsic-test/src/common/compare.rs
+++ b/library/stdarch/crates/intrinsic-test/src/common/compare.rs
@@ -2,27 +2,25 @@ use super::cli::FailureReason;
 use rayon::prelude::*;
 use std::process::Command;
 
-pub fn compare_outputs(
-    intrinsic_name_list: &Vec<String>,
-    toolchain: &str,
-    runner: &str,
-    target: &str,
-) -> bool {
+pub fn compare_outputs(intrinsic_name_list: &Vec<String>, runner: &str, target: &str) -> bool {
+    fn runner_command(runner: &str) -> Command {
+        let mut it = runner.split_whitespace();
+        let mut cmd = Command::new(it.next().unwrap());
+        cmd.args(it);
+
+        cmd
+    }
+
     let intrinsics = intrinsic_name_list
         .par_iter()
         .filter_map(|intrinsic_name| {
-            let c = Command::new("sh")
-                .arg("-c")
-                .arg(format!("{runner} ./c_programs/{intrinsic_name}"))
+            let c = runner_command(runner)
+                .arg("./c_programs/intrinsic-test-programs")
+                .arg(intrinsic_name)
                 .output();
 
-            let rust = Command::new("sh")
-                .current_dir("rust_programs")
-                .arg("-c")
-                .arg(format!(
-                    "cargo {toolchain} run --target {target} --bin {intrinsic_name} --release",
-                ))
-                .env("RUSTFLAGS", "-Cdebuginfo=0")
+            let rust = runner_command(runner)
+                .arg(format!("target/{target}/release/{intrinsic_name}"))
                 .output();
 
             let (c, rust) = match (c, rust) {
@@ -42,8 +40,8 @@ pub fn compare_outputs(
             if !rust.status.success() {
                 error!(
                     "Failed to run Rust program for intrinsic {intrinsic_name}\nstdout: {stdout}\nstderr: {stderr}",
-                    stdout = std::str::from_utf8(&rust.stdout).unwrap_or(""),
-                    stderr = std::str::from_utf8(&rust.stderr).unwrap_or(""),
+                    stdout = String::from_utf8_lossy(&rust.stdout),
+                    stderr = String::from_utf8_lossy(&rust.stderr),
                 );
                 return Some(FailureReason::RunRust(intrinsic_name.clone()));
             }
diff --git a/library/stdarch/crates/intrinsic-test/src/common/compile_c.rs b/library/stdarch/crates/intrinsic-test/src/common/compile_c.rs
index aebb7b111e2..0c905a149e4 100644
--- a/library/stdarch/crates/intrinsic-test/src/common/compile_c.rs
+++ b/library/stdarch/crates/intrinsic-test/src/common/compile_c.rs
@@ -5,11 +5,7 @@ pub struct CompilationCommandBuilder {
     cxx_toolchain_dir: Option<String>,
     arch_flags: Vec<String>,
     optimization: String,
-    include_paths: Vec<String>,
     project_root: Option<String>,
-    output: String,
-    input: String,
-    linker: Option<String>,
     extra_flags: Vec<String>,
 }
 
@@ -21,11 +17,7 @@ impl CompilationCommandBuilder {
             cxx_toolchain_dir: None,
             arch_flags: Vec::new(),
             optimization: "2".to_string(),
-            include_paths: Vec::new(),
             project_root: None,
-            output: String::new(),
-            input: String::new(),
-            linker: None,
             extra_flags: Vec::new(),
         }
     }
@@ -57,37 +49,12 @@ impl CompilationCommandBuilder {
         self
     }
 
-    /// Sets a list of include paths for compilation.
-    /// The paths that are passed must be relative to the
-    /// "cxx_toolchain_dir" directory path.
-    pub fn set_include_paths(mut self, paths: Vec<&str>) -> Self {
-        self.include_paths = paths.into_iter().map(|path| path.to_string()).collect();
-        self
-    }
-
     /// Sets the root path of all the generated test files.
     pub fn set_project_root(mut self, path: &str) -> Self {
         self.project_root = Some(path.to_string());
         self
     }
 
-    /// The name of the output executable, without any suffixes
-    pub fn set_output_name(mut self, path: &str) -> Self {
-        self.output = path.to_string();
-        self
-    }
-
-    /// The name of the input C file, without any suffixes
-    pub fn set_input_name(mut self, path: &str) -> Self {
-        self.input = path.to_string();
-        self
-    }
-
-    pub fn set_linker(mut self, linker: String) -> Self {
-        self.linker = Some(linker);
-        self
-    }
-
     pub fn add_extra_flags(mut self, flags: Vec<&str>) -> Self {
         let mut flags: Vec<String> = flags.into_iter().map(|f| f.to_string()).collect();
         self.extra_flags.append(&mut flags);
@@ -100,55 +67,69 @@ impl CompilationCommandBuilder {
 }
 
 impl CompilationCommandBuilder {
-    pub fn make_string(self) -> String {
-        let arch_flags = self.arch_flags.join("+");
+    pub fn into_cpp_compilation(self) -> CppCompilation {
+        let mut cpp_compiler = std::process::Command::new(self.compiler);
+
+        if let Some(project_root) = self.project_root {
+            cpp_compiler.current_dir(project_root);
+        }
+
         let flags = std::env::var("CPPFLAGS").unwrap_or("".into());
-        let project_root = self.project_root.unwrap_or_default();
-        let project_root_str = project_root.as_str();
-        let mut output = self.output.clone();
-        if self.linker.is_some() {
-            output += ".o"
-        };
-        let mut command = format!(
-            "{} {flags} -march={arch_flags} \
-            -O{} \
-            -o {project_root}/{} \
-            {project_root}/{}.cpp",
-            self.compiler, self.optimization, output, self.input,
-        );
-
-        command = command + " " + self.extra_flags.join(" ").as_str();
+        cpp_compiler.args(flags.split_whitespace());
+
+        cpp_compiler.arg(format!("-march={}", self.arch_flags.join("+")));
+
+        cpp_compiler.arg(format!("-O{}", self.optimization));
+
+        cpp_compiler.args(self.extra_flags);
 
         if let Some(target) = &self.target {
-            command = command + " --target=" + target;
+            cpp_compiler.arg(format!("--target={target}"));
         }
 
-        if let (Some(linker), Some(cxx_toolchain_dir)) = (&self.linker, &self.cxx_toolchain_dir) {
-            let include_args = self
-                .include_paths
-                .iter()
-                .map(|path| "--include-directory=".to_string() + cxx_toolchain_dir + path)
-                .collect::<Vec<_>>()
-                .join(" ");
-
-            command = command
-                + " -c "
-                + include_args.as_str()
-                + " && "
-                + linker
-                + " "
-                + project_root_str
-                + "/"
-                + &output
-                + " -o "
-                + project_root_str
-                + "/"
-                + &self.output
-                + " && rm "
-                + project_root_str
-                + "/"
-                + &output;
-        }
-        command
+        CppCompilation(cpp_compiler)
+    }
+}
+
+pub struct CppCompilation(std::process::Command);
+
+fn clone_command(command: &std::process::Command) -> std::process::Command {
+    let mut cmd = std::process::Command::new(command.get_program());
+    if let Some(current_dir) = command.get_current_dir() {
+        cmd.current_dir(current_dir);
+    }
+    cmd.args(command.get_args());
+
+    for (key, val) in command.get_envs() {
+        cmd.env(key, val.unwrap_or_default());
+    }
+
+    cmd
+}
+
+impl CppCompilation {
+    pub fn command_mut(&mut self) -> &mut std::process::Command {
+        &mut self.0
+    }
+
+    pub fn compile_object_file(
+        &self,
+        input: &str,
+        output: &str,
+    ) -> std::io::Result<std::process::Output> {
+        let mut cmd = clone_command(&self.0);
+        cmd.args([input, "-c", "-o", output]);
+        cmd.output()
+    }
+
+    pub fn link_executable(
+        &self,
+        inputs: impl Iterator<Item = String>,
+        output: &str,
+    ) -> std::io::Result<std::process::Output> {
+        let mut cmd = clone_command(&self.0);
+        cmd.args(inputs);
+        cmd.args(["-o", output]);
+        cmd.output()
     }
 }
diff --git a/library/stdarch/crates/intrinsic-test/src/common/gen_c.rs b/library/stdarch/crates/intrinsic-test/src/common/gen_c.rs
index 1cfb66c39b9..905efb6d890 100644
--- a/library/stdarch/crates/intrinsic-test/src/common/gen_c.rs
+++ b/library/stdarch/crates/intrinsic-test/src/common/gen_c.rs
@@ -1,8 +1,3 @@
-use itertools::Itertools;
-use rayon::prelude::*;
-use std::collections::BTreeMap;
-use std::process::Command;
-
 use super::argument::Argument;
 use super::indentation::Indentation;
 use super::intrinsic::IntrinsicDefinition;
@@ -11,104 +6,16 @@ use super::intrinsic_helpers::IntrinsicTypeDefinition;
 // The number of times each intrinsic will be called.
 const PASSES: u32 = 20;
 
-// Formats the main C program template with placeholders
-pub fn format_c_main_template(
-    notices: &str,
-    header_files: &[&str],
-    arch_identifier: &str,
-    arch_specific_definitions: &[&str],
-    arglists: &str,
-    passes: &str,
-) -> String {
-    format!(
-        r#"{notices}{header_files}
-#include <iostream>
-#include <cstring>
-#include <iomanip>
-#include <sstream>
-
-template<typename T1, typename T2> T1 cast(T2 x) {{
-  static_assert(sizeof(T1) == sizeof(T2), "sizeof T1 and T2 must be the same");
-  T1 ret{{}};
-  memcpy(&ret, &x, sizeof(T1));
-  return ret;
-}}
-
-std::ostream& operator<<(std::ostream& os, float16_t value) {{
-    uint16_t temp = 0;
-    memcpy(&temp, &value, sizeof(float16_t));
-    std::stringstream ss;
-    ss << "0x" << std::setfill('0') << std::setw(4) << std::hex << temp;
-    os << ss.str();
-    return os;
-}}
-
-#ifdef __{arch_identifier}__
-{arch_specific_definitions}
-#endif
-
-{arglists}
-
-int main(int argc, char **argv) {{
-{passes}
-    return 0;
-}}"#,
-        header_files = header_files
-            .iter()
-            .map(|header| format!("#include <{header}>"))
-            .collect::<Vec<_>>()
-            .join("\n"),
-        arch_specific_definitions = arch_specific_definitions.join("\n"),
-    )
-}
-
-pub fn compile_c_programs(compiler_commands: &[String]) -> bool {
-    compiler_commands
-        .par_iter()
-        .map(|compiler_command| {
-            let output = Command::new("sh").arg("-c").arg(compiler_command).output();
-            if let Ok(output) = output {
-                if output.status.success() {
-                    true
-                } else {
-                    error!(
-                        "Failed to compile code for intrinsics: \n\nstdout:\n{}\n\nstderr:\n{}",
-                        std::str::from_utf8(&output.stdout).unwrap_or(""),
-                        std::str::from_utf8(&output.stderr).unwrap_or("")
-                    );
-                    false
-                }
-            } else {
-                error!("Command failed: {output:#?}");
-                false
-            }
-        })
-        .find_any(|x| !x)
-        .is_none()
-}
-
-// Creates directory structure and file path mappings
-pub fn setup_c_file_paths(identifiers: &Vec<String>) -> BTreeMap<&String, String> {
-    let _ = std::fs::create_dir("c_programs");
-    identifiers
-        .par_iter()
-        .map(|identifier| {
-            let c_filename = format!(r#"c_programs/{identifier}.cpp"#);
-
-            (identifier, c_filename)
-        })
-        .collect::<BTreeMap<&String, String>>()
-}
-
 pub fn generate_c_test_loop<T: IntrinsicTypeDefinition + Sized>(
+    w: &mut impl std::io::Write,
     intrinsic: &dyn IntrinsicDefinition<T>,
     indentation: Indentation,
     additional: &str,
     passes: u32,
-    _target: &str,
-) -> String {
+) -> std::io::Result<()> {
     let body_indentation = indentation.nested();
-    format!(
+    writeln!(
+        w,
         "{indentation}for (int i=0; i<{passes}; i++) {{\n\
             {loaded_args}\
             {body_indentation}auto __return_value = {intrinsic_call}({args});\n\
@@ -121,78 +28,172 @@ pub fn generate_c_test_loop<T: IntrinsicTypeDefinition + Sized>(
     )
 }
 
-pub fn generate_c_constraint_blocks<T: IntrinsicTypeDefinition>(
+pub fn generate_c_constraint_blocks<'a, T: IntrinsicTypeDefinition + 'a>(
+    w: &mut impl std::io::Write,
     intrinsic: &dyn IntrinsicDefinition<T>,
     indentation: Indentation,
-    constraints: &[&Argument<T>],
+    constraints: &mut (impl Iterator<Item = &'a Argument<T>> + Clone),
     name: String,
-    target: &str,
-) -> String {
-    if let Some((current, constraints)) = constraints.split_last() {
-        let range = current
-            .constraint
-            .iter()
-            .map(|c| c.to_range())
-            .flat_map(|r| r.into_iter());
-
-        let body_indentation = indentation.nested();
-        range
-            .map(|i| {
-                format!(
-                    "{indentation}{{\n\
-                        {body_indentation}{ty} {name} = {val};\n\
-                        {pass}\n\
-                    {indentation}}}",
-                    name = current.name,
-                    ty = current.ty.c_type(),
-                    val = i,
-                    pass = generate_c_constraint_blocks(
-                        intrinsic,
-                        body_indentation,
-                        constraints,
-                        format!("{name}-{i}"),
-                        target,
-                    )
-                )
-            })
-            .join("\n")
-    } else {
-        generate_c_test_loop(intrinsic, indentation, &name, PASSES, target)
+) -> std::io::Result<()> {
+    let Some(current) = constraints.next() else {
+        return generate_c_test_loop(w, intrinsic, indentation, &name, PASSES);
+    };
+
+    let body_indentation = indentation.nested();
+    for i in current.constraint.iter().flat_map(|c| c.to_range()) {
+        let ty = current.ty.c_type();
+
+        writeln!(w, "{indentation}{{")?;
+        writeln!(w, "{body_indentation}{ty} {} = {i};", current.name)?;
+
+        generate_c_constraint_blocks(
+            w,
+            intrinsic,
+            body_indentation,
+            &mut constraints.clone(),
+            format!("{name}-{i}"),
+        )?;
+
+        writeln!(w, "{indentation}}}")?;
     }
+
+    Ok(())
 }
 
 // Compiles C test programs using specified compiler
-pub fn create_c_test_program<T: IntrinsicTypeDefinition>(
+pub fn create_c_test_function<T: IntrinsicTypeDefinition>(
+    w: &mut impl std::io::Write,
     intrinsic: &dyn IntrinsicDefinition<T>,
-    header_files: &[&str],
-    target: &str,
-    c_target: &str,
-    notices: &str,
-    arch_specific_definitions: &[&str],
-) -> String {
+) -> std::io::Result<()> {
+    let indentation = Indentation::default();
+
+    writeln!(w, "int run_{}() {{", intrinsic.name())?;
+
+    // Define the arrays of arguments.
     let arguments = intrinsic.arguments();
-    let constraints = arguments
-        .iter()
-        .filter(|&i| i.has_constraint())
-        .collect_vec();
+    arguments.gen_arglists_c(w, indentation.nested(), PASSES)?;
 
-    let indentation = Indentation::default();
-    format_c_main_template(
-        notices,
-        header_files,
-        c_target,
-        arch_specific_definitions,
-        intrinsic
-            .arguments()
-            .gen_arglists_c(indentation, PASSES)
-            .as_str(),
-        generate_c_constraint_blocks(
-            intrinsic,
-            indentation.nested(),
-            constraints.as_slice(),
-            Default::default(),
-            target,
-        )
-        .as_str(),
-    )
+    generate_c_constraint_blocks(
+        w,
+        intrinsic,
+        indentation.nested(),
+        &mut arguments.iter().rev().filter(|&i| i.has_constraint()),
+        Default::default(),
+    )?;
+
+    writeln!(w, "    return 0;")?;
+    writeln!(w, "}}")?;
+
+    Ok(())
+}
+
+pub fn write_mod_cpp<T: IntrinsicTypeDefinition>(
+    w: &mut impl std::io::Write,
+    notice: &str,
+    architecture: &str,
+    platform_headers: &[&str],
+    intrinsics: &[impl IntrinsicDefinition<T>],
+) -> std::io::Result<()> {
+    write!(w, "{notice}")?;
+
+    for header in platform_headers {
+        writeln!(w, "#include <{header}>")?;
+    }
+
+    writeln!(
+        w,
+        r#"
+#include <iostream>
+#include <cstring>
+#include <iomanip>
+#include <sstream>
+
+template<typename T1, typename T2> T1 cast(T2 x) {{
+  static_assert(sizeof(T1) == sizeof(T2), "sizeof T1 and T2 must be the same");
+  T1 ret{{}};
+  memcpy(&ret, &x, sizeof(T1));
+  return ret;
+}}
+
+std::ostream& operator<<(std::ostream& os, float16_t value);
+
+
+
+"#
+    )?;
+
+    writeln!(w, "#ifdef __{architecture}__")?;
+    writeln!(
+        w,
+        "std::ostream& operator<<(std::ostream& os, poly128_t value);"
+    )?;
+    writeln!(w, "#endif")?;
+
+    for intrinsic in intrinsics {
+        create_c_test_function(w, intrinsic)?;
+    }
+
+    Ok(())
+}
+
+pub fn write_main_cpp<'a>(
+    w: &mut impl std::io::Write,
+    architecture: &str,
+    arch_specific_definitions: &str,
+    intrinsics: impl Iterator<Item = &'a str> + Clone,
+) -> std::io::Result<()> {
+    writeln!(w, "#include <iostream>")?;
+    writeln!(w, "#include <string>")?;
+
+    for header in ["arm_neon.h", "arm_acle.h", "arm_fp16.h"] {
+        writeln!(w, "#include <{header}>")?;
+    }
+
+    writeln!(
+        w,
+        r#"
+#include <cstring>
+#include <iomanip>
+#include <sstream>
+
+std::ostream& operator<<(std::ostream& os, float16_t value) {{
+    uint16_t temp = 0;
+    memcpy(&temp, &value, sizeof(float16_t));
+    std::stringstream ss;
+    ss << "0x" << std::setfill('0') << std::setw(4) << std::hex << temp;
+    os << ss.str();
+    return os;
+}}
+"#
+    )?;
+
+    writeln!(w, "#ifdef __{architecture}__")?;
+    writeln!(w, "{arch_specific_definitions }")?;
+    writeln!(w, "#endif")?;
+
+    for intrinsic in intrinsics.clone() {
+        writeln!(w, "extern int run_{intrinsic}(void);")?;
+    }
+
+    writeln!(w, "int main(int argc, char **argv) {{")?;
+    writeln!(w, "    std::string intrinsic_name = argv[1];")?;
+
+    writeln!(w, "    if (false) {{")?;
+
+    for intrinsic in intrinsics {
+        writeln!(w, "    }} else if (intrinsic_name == \"{intrinsic}\") {{")?;
+        writeln!(w, "        return run_{intrinsic}();")?;
+    }
+
+    writeln!(w, "    }} else {{")?;
+    writeln!(
+        w,
+        "        std::cerr << \"Unknown command: \" << intrinsic_name << \"\\n\";"
+    )?;
+    writeln!(w, "        return -1;")?;
+    writeln!(w, "    }}")?;
+
+    writeln!(w, "}}")?;
+
+    Ok(())
 }
diff --git a/library/stdarch/crates/intrinsic-test/src/common/gen_rust.rs b/library/stdarch/crates/intrinsic-test/src/common/gen_rust.rs
index 52bccaf905c..0e4a95ab528 100644
--- a/library/stdarch/crates/intrinsic-test/src/common/gen_rust.rs
+++ b/library/stdarch/crates/intrinsic-test/src/common/gen_rust.rs
@@ -2,7 +2,6 @@ use itertools::Itertools;
 use rayon::prelude::*;
 use std::collections::BTreeMap;
 use std::fs::File;
-use std::io::Write;
 use std::process::Command;
 
 use super::argument::Argument;
@@ -23,8 +22,8 @@ pub fn format_rust_main_template(
 ) -> String {
     format!(
         r#"{notices}#![feature(simd_ffi)]
-#![feature(link_llvm_intrinsics)]
 #![feature(f16)]
+#![allow(unused)]
 {configurations}
 {definitions}
 
@@ -38,6 +37,42 @@ fn main() {{
     )
 }
 
+fn write_cargo_toml(w: &mut impl std::io::Write, binaries: &[String]) -> std::io::Result<()> {
+    writeln!(
+        w,
+        concat!(
+            "[package]\n",
+            "name = \"intrinsic-test-programs\"\n",
+            "version = \"{version}\"\n",
+            "authors = [{authors}]\n",
+            "license = \"{license}\"\n",
+            "edition = \"2018\"\n",
+            "[workspace]\n",
+            "[dependencies]\n",
+            "core_arch = {{ path = \"../crates/core_arch\" }}",
+        ),
+        version = env!("CARGO_PKG_VERSION"),
+        authors = env!("CARGO_PKG_AUTHORS")
+            .split(":")
+            .format_with(", ", |author, fmt| fmt(&format_args!("\"{author}\""))),
+        license = env!("CARGO_PKG_LICENSE"),
+    )?;
+
+    for binary in binaries {
+        writeln!(
+            w,
+            concat!(
+                "[[bin]]\n",
+                "name = \"{binary}\"\n",
+                "path = \"{binary}/main.rs\"\n",
+            ),
+            binary = binary,
+        )?;
+    }
+
+    Ok(())
+}
+
 pub fn compile_rust_programs(
     binaries: Vec<String>,
     toolchain: Option<&str>,
@@ -45,56 +80,20 @@ pub fn compile_rust_programs(
     linker: Option<&str>,
 ) -> bool {
     let mut cargo = File::create("rust_programs/Cargo.toml").unwrap();
-    cargo
-        .write_all(
-            format!(
-                r#"[package]
-name = "intrinsic-test-programs"
-version = "{version}"
-authors = [{authors}]
-license = "{license}"
-edition = "2018"
-[workspace]
-[dependencies]
-core_arch = {{ path = "../crates/core_arch" }}
-{binaries}"#,
-                version = env!("CARGO_PKG_VERSION"),
-                authors = env!("CARGO_PKG_AUTHORS")
-                    .split(":")
-                    .format_with(", ", |author, fmt| fmt(&format_args!("\"{author}\""))),
-                license = env!("CARGO_PKG_LICENSE"),
-                binaries = binaries
-                    .iter()
-                    .map(|binary| {
-                        format!(
-                            r#"[[bin]]
-name = "{binary}"
-path = "{binary}/main.rs""#,
-                        )
-                    })
-                    .collect::<Vec<_>>()
-                    .join("\n")
-            )
-            .into_bytes()
-            .as_slice(),
-        )
-        .unwrap();
-
-    let toolchain = match toolchain {
-        None => return true,
-        Some(t) => t,
-    };
+    write_cargo_toml(&mut cargo, &binaries).unwrap();
 
     /* If there has been a linker explicitly set from the command line then
      * we want to set it via setting it in the RUSTFLAGS*/
 
-    let cargo_command = format!("cargo {toolchain} build --target {target} --release");
+    let mut cargo_command = Command::new("cargo");
+    cargo_command.current_dir("rust_programs");
 
-    let mut command = Command::new("sh");
-    command
-        .current_dir("rust_programs")
-        .arg("-c")
-        .arg(cargo_command);
+    if let Some(toolchain) = toolchain {
+        if !toolchain.is_empty() {
+            cargo_command.arg(toolchain);
+        }
+    }
+    cargo_command.args(["build", "--target", target, "--release"]);
 
     let mut rust_flags = "-Cdebuginfo=0".to_string();
     if let Some(linker) = linker {
@@ -102,11 +101,11 @@ path = "{binary}/main.rs""#,
         rust_flags.push_str(linker);
         rust_flags.push_str(" -C link-args=-static");
 
-        command.env("CPPFLAGS", "-fuse-ld=lld");
+        cargo_command.env("CPPFLAGS", "-fuse-ld=lld");
     }
 
-    command.env("RUSTFLAGS", rust_flags);
-    let output = command.output();
+    cargo_command.env("RUSTFLAGS", rust_flags);
+    let output = cargo_command.output();
 
     if let Ok(output) = output {
         if output.status.success() {
diff --git a/library/stdarch/crates/intrinsic-test/src/common/intrinsic_helpers.rs b/library/stdarch/crates/intrinsic-test/src/common/intrinsic_helpers.rs
index 3d200b19461..697f9c8754d 100644
--- a/library/stdarch/crates/intrinsic-test/src/common/intrinsic_helpers.rs
+++ b/library/stdarch/crates/intrinsic-test/src/common/intrinsic_helpers.rs
@@ -9,13 +9,21 @@ use super::indentation::Indentation;
 use super::values::value_for_array;
 
 #[derive(Debug, PartialEq, Copy, Clone)]
+pub enum Sign {
+    Signed,
+    Unsigned,
+}
+
+#[derive(Debug, PartialEq, Copy, Clone)]
 pub enum TypeKind {
     BFloat,
     Float,
-    Int,
-    UInt,
+    Int(Sign),
+    Char(Sign),
     Poly,
     Void,
+    Mask,
+    Vector,
 }
 
 impl FromStr for TypeKind {
@@ -23,12 +31,17 @@ impl FromStr for TypeKind {
 
     fn from_str(s: &str) -> Result<Self, Self::Err> {
         match s {
-            "bfloat" => Ok(Self::BFloat),
-            "float" => Ok(Self::Float),
-            "int" => Ok(Self::Int),
+            "bfloat" | "BF16" => Ok(Self::BFloat),
+            "float" | "double" | "FP16" | "FP32" | "FP64" => Ok(Self::Float),
+            "int" | "long" | "short" | "SI8" | "SI16" | "SI32" | "SI64" => {
+                Ok(Self::Int(Sign::Signed))
+            }
             "poly" => Ok(Self::Poly),
-            "uint" | "unsigned" => Ok(Self::UInt),
+            "char" => Ok(Self::Char(Sign::Signed)),
+            "uint" | "unsigned" | "UI8" | "UI16" | "UI32" | "UI64" => Ok(Self::Int(Sign::Unsigned)),
             "void" => Ok(Self::Void),
+            "MASK" => Ok(Self::Mask),
+            "M64" | "M128" | "M256" | "M512" => Ok(Self::Vector),
             _ => Err(format!("Impossible to parse argument kind {s}")),
         }
     }
@@ -42,10 +55,14 @@ impl fmt::Display for TypeKind {
             match self {
                 Self::BFloat => "bfloat",
                 Self::Float => "float",
-                Self::Int => "int",
-                Self::UInt => "uint",
+                Self::Int(Sign::Signed) => "int",
+                Self::Int(Sign::Unsigned) => "uint",
                 Self::Poly => "poly",
                 Self::Void => "void",
+                Self::Char(Sign::Signed) => "char",
+                Self::Char(Sign::Unsigned) => "unsigned char",
+                Self::Mask => "mask",
+                Self::Vector => "vector",
             }
         )
     }
@@ -56,9 +73,10 @@ impl TypeKind {
     pub fn c_prefix(&self) -> &str {
         match self {
             Self::Float => "float",
-            Self::Int => "int",
-            Self::UInt => "uint",
+            Self::Int(Sign::Signed) => "int",
+            Self::Int(Sign::Unsigned) => "uint",
             Self::Poly => "poly",
+            Self::Char(Sign::Signed) => "char",
             _ => unreachable!("Not used: {:#?}", self),
         }
     }
@@ -66,10 +84,13 @@ impl TypeKind {
     /// Gets the rust prefix for the type kind i.e. i, u, f.
     pub fn rust_prefix(&self) -> &str {
         match self {
+            Self::BFloat => "bf",
             Self::Float => "f",
-            Self::Int => "i",
-            Self::UInt => "u",
+            Self::Int(Sign::Signed) => "i",
+            Self::Int(Sign::Unsigned) => "u",
             Self::Poly => "u",
+            Self::Char(Sign::Unsigned) => "u",
+            Self::Char(Sign::Signed) => "i",
             _ => unreachable!("Unused type kind: {:#?}", self),
         }
     }
@@ -133,11 +154,14 @@ impl IntrinsicType {
     }
 
     pub fn c_scalar_type(&self) -> String {
-        format!(
-            "{prefix}{bits}_t",
-            prefix = self.kind().c_prefix(),
-            bits = self.inner_size()
-        )
+        match self.kind() {
+            TypeKind::Char(_) => String::from("char"),
+            _ => format!(
+                "{prefix}{bits}_t",
+                prefix = self.kind().c_prefix(),
+                bits = self.inner_size()
+            ),
+        }
     }
 
     pub fn rust_scalar_type(&self) -> String {
@@ -155,8 +179,8 @@ impl IntrinsicType {
                 bit_len: Some(8),
                 ..
             } => match kind {
-                TypeKind::Int => "(int)",
-                TypeKind::UInt => "(unsigned int)",
+                TypeKind::Int(Sign::Signed) => "(int)",
+                TypeKind::Int(Sign::Unsigned) => "(unsigned int)",
                 TypeKind::Poly => "(unsigned int)(uint8_t)",
                 _ => "",
             },
@@ -172,6 +196,21 @@ impl IntrinsicType {
                 128 => "",
                 _ => panic!("invalid bit_len"),
             },
+            IntrinsicType {
+                kind: TypeKind::Float,
+                bit_len: Some(bit_len),
+                ..
+            } => match bit_len {
+                16 => "(float16_t)",
+                32 => "(float)",
+                64 => "(double)",
+                128 => "",
+                _ => panic!("invalid bit_len"),
+            },
+            IntrinsicType {
+                kind: TypeKind::Char(_),
+                ..
+            } => "(char)",
             _ => "",
         }
     }
@@ -185,7 +224,7 @@ impl IntrinsicType {
         match self {
             IntrinsicType {
                 bit_len: Some(bit_len @ (8 | 16 | 32 | 64)),
-                kind: kind @ (TypeKind::Int | TypeKind::UInt | TypeKind::Poly),
+                kind: kind @ (TypeKind::Int(_) | TypeKind::Poly | TypeKind::Char(_)),
                 simd_len,
                 vec_len,
                 ..
@@ -201,7 +240,8 @@ impl IntrinsicType {
                         .format_with(",\n", |i, fmt| {
                             let src = value_for_array(*bit_len, i);
                             assert!(src == 0 || src.ilog2() < *bit_len);
-                            if *kind == TypeKind::Int && (src >> (*bit_len - 1)) != 0 {
+                            if *kind == TypeKind::Int(Sign::Signed) && (src >> (*bit_len - 1)) != 0
+                            {
                                 // `src` is a two's complement representation of a negative value.
                                 let mask = !0u64 >> (64 - *bit_len);
                                 let ones_compl = src ^ mask;
@@ -257,7 +297,7 @@ impl IntrinsicType {
                 ..
             } => false,
             IntrinsicType {
-                kind: TypeKind::Int | TypeKind::UInt | TypeKind::Poly,
+                kind: TypeKind::Int(_) | TypeKind::Poly,
                 ..
             } => true,
             _ => unimplemented!(),
@@ -282,7 +322,9 @@ pub trait IntrinsicTypeDefinition: Deref<Target = IntrinsicType> {
     fn get_lane_function(&self) -> String;
 
     /// can be implemented in an `impl` block
-    fn from_c(_s: &str, _target: &str) -> Result<Box<Self>, String>;
+    fn from_c(_s: &str, _target: &str) -> Result<Self, String>
+    where
+        Self: Sized;
 
     /// Gets a string containing the typename for this type in C format.
     /// can be directly defined in `impl` blocks
diff --git a/library/stdarch/crates/intrinsic-test/src/common/write_file.rs b/library/stdarch/crates/intrinsic-test/src/common/write_file.rs
index 0ba3e829a6b..92dd70b7c57 100644
--- a/library/stdarch/crates/intrinsic-test/src/common/write_file.rs
+++ b/library/stdarch/crates/intrinsic-test/src/common/write_file.rs
@@ -1,5 +1,3 @@
-use super::gen_c::create_c_test_program;
-use super::gen_c::setup_c_file_paths;
 use super::gen_rust::{create_rust_test_program, setup_rust_file_paths};
 use super::intrinsic::IntrinsicDefinition;
 use super::intrinsic_helpers::IntrinsicTypeDefinition;
@@ -11,37 +9,6 @@ pub fn write_file(filename: &String, code: String) {
     file.write_all(code.into_bytes().as_slice()).unwrap();
 }
 
-pub fn write_c_testfiles<T: IntrinsicTypeDefinition + Sized>(
-    intrinsics: &Vec<&dyn IntrinsicDefinition<T>>,
-    target: &str,
-    c_target: &str,
-    headers: &[&str],
-    notice: &str,
-    arch_specific_definitions: &[&str],
-) -> Vec<String> {
-    let intrinsics_name_list = intrinsics
-        .iter()
-        .map(|i| i.name().clone())
-        .collect::<Vec<_>>();
-    let filename_mapping = setup_c_file_paths(&intrinsics_name_list);
-
-    intrinsics.iter().for_each(|&i| {
-        let c_code = create_c_test_program(
-            i,
-            headers,
-            target,
-            c_target,
-            notice,
-            arch_specific_definitions,
-        );
-        if let Some(filename) = filename_mapping.get(&i.name()) {
-            write_file(filename, c_code)
-        };
-    });
-
-    intrinsics_name_list
-}
-
 pub fn write_rust_testfiles<T: IntrinsicTypeDefinition>(
     intrinsics: Vec<&dyn IntrinsicDefinition<T>>,
     rust_target: &str,
diff --git a/library/stdarch/crates/intrinsic-test/src/main.rs b/library/stdarch/crates/intrinsic-test/src/main.rs
index 054138a0dba..538f317a297 100644
--- a/library/stdarch/crates/intrinsic-test/src/main.rs
+++ b/library/stdarch/crates/intrinsic-test/src/main.rs
@@ -30,12 +30,15 @@ fn main() {
 
     let test_environment = test_environment_result.unwrap();
 
+    info!("building C binaries");
     if !test_environment.build_c_file() {
         std::process::exit(2);
     }
+    info!("building Rust binaries");
     if !test_environment.build_rust_file() {
         std::process::exit(3);
     }
+    info!("comaparing outputs");
     if !test_environment.compare_outputs() {
         std::process::exit(1);
     }