Merge pull request #87 from williamfgc/release

Bump to v0.0.4

Project.toml

@@ -1,7 +1,7 @@
 name = "JACC"
 uuid = "0979c8fe-16a4-4796-9b82-89a9f10403ea"
 authors = ["pedrovalerolara <[email protected]>", "williamfgc <[email protected]>"]
-version = "0.0.3"
+version = "0.0.4"
 
 [deps]
 Atomix = "a9b6321e-bd34-4604-b9c9-b65b8de01458"

README.md

@@ -7,41 +7,49 @@
 
 CPU/GPU performance portable layer for Julia
 
-JACC.jl follows a function as a argument approach in combination with the power of Julia's ecosystem for multiple dispatch, GPU back end access, and weak dependencies since Julia v1.9 . Similar to portable layers like Kokkos, users would pass a size and a function including its arguments to a `parallel_for` or `parallel_reduce` function.
-The overall goal is to write a single source code that can be executed in multiple CPU and GPU parallel programming environments. Following the principle for OpenACC, it's meant to simplify programming for heterogeneous CPU and GPU systems.
+JACC.jl follows a function as a argument approach in combination with the power of Julia's ecosystem for multiple dispatch, GPU access via [JuliaGPU back ends](https://juliagpu.org/), and [package extensions](https://julialang.org/blog/2023/04/julia-1.9-highlights/#package_extensions) since Julia v1.9 . Similar to portable layers like Kokkos, users would pass a size and a function including its arguments to a `parallel_for` or `parallel_reduce` function.
+The overall goal is to write a single source code that can be executed on multiple vendor CPU and GPU parallel programming environments. JACC meant to simplify CPU/GPU kernel programming using a simple application programming interface (API).
 
 JuliaCon 2023 presentation [video](https://live.juliacon.org/talk/AY8EUX).
 
 1. Set a back end: "cuda", "amdgpu", or "threads" (default) with `JACC.JACCPreferences` generating a `LocalPreferences.toml` file
 
-```
-julia> import JACC.JACCPreferences
-julia> JACCPreferences.set_backend("cuda")
-```
+    ```
+    julia> import JACC.JACCPreferences
+    julia> JACCPreferences.set_backend("cuda")
+    ```
 
 2. Run a kernel example (see tests directory)
 
-```
-import JACC
+    ```
+    import JACC
 
-function axpy(i, alpha, x, y)
-  if i <= length(x)
-    @inbounds x[i] += alpha * y[i]
-  end
-end
+    function axpy(i, alpha, x, y)
+      if i <= length(x)
+        @inbounds x[i] += alpha * y[i]
+      end
+    end
 
-N = 10
-# Generate random vectors x and y of length N for the interval [0, 100]
-x = round.(rand(Float32, N) * 100)
-y = round.(rand(Float32, N) * 100)
-alpha = 2.5
+    N = 10
+    # Generate random vectors x and y of length N for the interval [0, 100]
+    x = round.(rand(Float32, N) * 100)
+    y = round.(rand(Float32, N) * 100)
+    alpha = 2.5
 
-x_d = JACC.Array(x)
-y_d = JACC.Array(y)
-JACC.parallel_for(N, axpy, alpha, x_d, y_d)
-```
+    x_d = JACC.Array(x)
+    y_d = JACC.Array(y)
+    JACC.parallel_for(N, axpy, alpha, x_d, y_d)
+    ```
 
 We currently have a limited number of configurations. 
 We hope to study and incoorporate more relevant cases and dimensions shapes as needed.
+For an app integration example see the [GrayScott.jl JACC branch](https://github.com/JuliaORNL/GrayScott.jl/tree/GrayScott-JACC) and the [Simulation.jl](https://github.com/JuliaORNL/GrayScott.jl/blob/GrayScott-JACC/src/simulation/Simulation.jl) for writing kernels with JACC.jl and selecting specific vendor back ends in Julia.
 
-Funded by the Exascale Computing Project (ECP) [PROTEAS-TUNE](https://www.ornl.gov/project/proteas-tune) and the [ASCR Bluestone](https://csmd.ornl.gov/Bluestone) projects.
+
+Funded by the US Department of Energy Advanced Scientific Computing Research (ASCR) projects:
+
+- PESO and S4PST as part of the Next Generation of Scientific Software Technologies (NGSST)
+- [Bluestone X-Stack](https://csmd.ornl.gov/Bluestone)
+
+Past sponsors:
+- The Exascale Computing Project (ECP) [PROTEAS-TUNE](https://www.ornl.gov/project/proteas-tune)