UPdated examples and documentations to not require glsl deps

docs/overview/advanced-examples.rst

@@ -55,7 +55,8 @@ The example below shows how you can enable the "VK_EXT_shader_atomic_float" exte
                  atomicAdd(pa[2], pcs.z);
              })");
 
-       std::vector<uint32_t> spirv = kp_test_utils::Shader::compileSource(shader);
+       // See shader documentation section for compileSource
+       std::vector<uint32_t> spirv = compileSource(shader);
 
        std::shared_ptr<kp::Sequence> sq = nullptr;
 
@@ -102,7 +103,8 @@ We also provide tools that allow you to `convert shaders into C++ headers <https
                 throw std::runtime_error("Kompute OpMult expected 3 tensors but got " + tensors.size());
             }
 
-            std::vector<uint32_t> spirv = kp_test_utils::Shader::compileSource(R"(
+           // See shader documentation section for compileSource
+            std::vector<uint32_t> spirv = compileSource(R"(
                 #version 450
 
                 layout(set = 0, binding = 0) buffer tensorLhs {
@@ -215,7 +217,8 @@ In this case we create a shader that should take a couple of milliseconds to run
         }
     )");
 
-    auto algo = mgr.algorithm({tensor}, kp_test_utils::Shader::compileSource(shader));
+    // See shader documentation section for compileSource
+    auto algo = mgr.algorithm({tensor}, compileSource(shader));
 
 Now we are able to run the await function on the default sequence. 
 
@@ -361,7 +364,8 @@ Similar to the asyncrhonous usecase above, we can still run synchronous commands
            }
        )");
 
-       std::vector<uint32_t> spirv = kp_test_utils::Shader::compileSource(shader);
+       // See shader documentation section for compileSource
+       std::vector<uint32_t> spirv = compileSource(shader);
 
        std::shared_ptr<kp::Algorithm> algo = mgr.algorithm({tensorA, tenssorB}, spirv);
 

docs/overview/build-system.rst

@@ -69,8 +69,6 @@ Compile Flags
      - Enable debug build including debug flags (enabled by cmake debug build)
    * - -DKOMPUTE_DISABLE_VK_DEBUG_LAYERS
      - Disable the debug Vulkan SDK Layers, mainly used for android builds
-   * - -DKOMPUTE_DISABLE_SHADER_UTILS
-     - Disable the shader utils and skip adding glslang as dependency
 
 Other CMake Flags
 ~~~~~~~~~~~~~~~~~

docs/overview/ci-tests.rst

@@ -38,10 +38,7 @@ Running on the CPU
 
 We use `Swiftshader <https://github.com/google/swiftshader>`_ to enable us to run the Kompute framework directly on the CPU for the CI tests.
 
-Even though Swiftshader is optimized to function as a high-performance CPU backend for the Vulkan SDK, there are several limitations, the most notable in context of Kompute are:
-
-* Loading files (spirv or text) leads to segfault
-* Loading raw text string shaders leads to segfault
+Even though Swiftshader is optimized to function as a high-performance CPU backend for the Vulkan SDK, there are several limitations, the most notable are limitations in extensions.
 
 This is one of the main reason why only a subset of the tests are run in the CI.
 

docs/overview/reference.rst

@@ -119,13 +119,4 @@ The :class:`kp::OpMemoryBarrier` is a tensor only operation which adds memory ba
 .. doxygenclass:: kp::OpTensorSyncDevice
    :members:
 
-Shader
---------
-
-The :class:`kp_test_utils::Shader` class contains a set of utilities to compile and process shaders.
-
-.. doxygenclass:: kp_test_utils::Shader
-   :members:
-
-
 

docs/overview/shaders-to-headers.rst

@@ -3,6 +3,29 @@
 Processing Shaders with Kompute
 =====================
 
+Demo / testing function to compile shaders
+----------------------------------
+
+GLSLANG was initially integrated as part of the framework but it now has been removed due to the license of the glslang pre-processor being under a custom NVIDIA license which explicitly excludes grant of any licenses to NVIDIA's patents in the preprocessor.
+
+For users that are looking to quickly test the processors it is possible to use the function that is provided in the examples which provides a (non-thread-safe / non-robust) implementation that compiles a shader string into spirv bytes. It is not recommended to use in production but it does enable for faster iteration cycles during development.
+
+.. code-block:: cpp
+    :linenos:
+    static std::vector<uint32_t>
+    compileSource(
+      const std::string& source)
+    {
+        if (system(std::string("glslangValidator --stdin -S comp -V -o tmp_kp_shader.comp.spv << END
+    " + source + "
+    END").c_str()))
+            throw std::runtime_error("Error running glslangValidator command");
+        std::ifstream fileStream("tmp_kp_shader.comp.spv", std::ios::binary);
+        std::vector<char> buffer;
+        buffer.insert(buffer.begin(), std::istreambuf_iterator<char>(fileStream), {});
+        return {(uint32_t*)buffer.data(), (uint32_t*)(buffer.data() + buffer.size())};
+    }
+
 Converting Shaders into C / C++ Header Files
 ----------------------------------