Merge pull request #47 from szymon-zadworny/f16-example

Use f16 data in kernel launch example

Author: bratpiorka

Cargo.lock

@@ -29,4 +29,5 @@ pin-project = "1.1.13"
 thiserror = "2.0.18"
 
 [dev-dependencies]
+half = { version = "2.7.1", features = ["bytemuck"] }
 tokio = { version = "1.52.3", features = ["macros", "rt", "rt-multi-thread"] }

sycl/sycl-rs/Cargo.toml

@@ -6,6 +6,7 @@
 // SPDX-License-Identifier: MIT OR Apache-2.0
 //
 
+use half::f16;
 use sycl_rs::prelude::*;
 
 static IOTA_SRC: &str = r#"
@@ -15,16 +16,16 @@ namespace syclexp = sycl::ext::oneapi::experimental;
 
 extern "C"
 SYCL_EXT_ONEAPI_FUNCTION_PROPERTY((syclexp::nd_range_kernel<1>))
-void iota(float start, float *ptr) {
+void iota(sycl::half start, sycl::half *ptr) {
     size_t id = syclext::this_work_item::get_nd_item<1>().get_global_linear_id();
-    ptr[id] = start + static_cast<float>(id);
+    ptr[id] = start + static_cast<sycl::half>(id);
 }
 "#;
 
 #[tokio::main]
 async fn main() -> sycl_rs::Result<()> {
     let mut queue = Queue::new();
-    let mut device_array = queue.alloc_device::<f32>(1024)?.await?;
+    let mut device_array = queue.alloc_device::<f16>(1024)?.await?;
 
     let kernel = queue
         .get_context()
@@ -36,12 +37,12 @@ async fn main() -> sycl_rs::Result<()> {
         queue.launch(
             NdRange::new([1024], [16]),
             &kernel,
-            (3.14_f32, &mut device_array),
+            (f16::from_f32(3.14), &mut device_array),
         )
     }?
     .await?;
 
-    let mut host_array = queue.alloc_host::<f32>(1024)?.await?;
+    let mut host_array = queue.alloc_host::<f16>(1024)?.await?;
 
     queue.copy(&device_array, &mut host_array)?.await?;