🧠 SIL (SIL Intermediate Language)

SIL is a minimalist, typed intermediate language designed to be compiled into SPIR-V, the low-level intermediate representation used in GPU programming.

This project includes a full compiler pipeline:

• Lexical analysis
• Parsing into an AST
• Optional preprocessing (MiniSIL)
• SPIR-V code generation
• Runtime execution on the GPU via pyopencl

Note: minisil.py is an optional preprocessor that expands high-level array constructs into scalar operations.

---

✨ Features

• ✔️ Typed variables: uint, float, bool, pointers
• ✔️ Kernel definitions with parameters
• ✔️ Arithmetic, logical, bitwise, and cast expressions
• ✔️ Control flow: if, loop, break, return
• ✔️ @cpu blocks for CPU-side assertions and I/O
• ✔️ Array unrolling support via MiniSIL
• ✔️ SPIR-V validation and execution using pyopencl

---

📚 SIL Language Overview

kernel sum(out: uint) {
    var a: uint = 12;
    var b: uint = 18;
    out = a + b;
}

@cpu
import numpy as np, pyopencl as cl
buf = np.zeros(1, dtype=np.uint32)
out = rt.create_buffer(buf, cl.mem_flags.READ_WRITE)
rt.run_kernel("sum", 1, {"out": out})
result = rt.read_buffer(out, np.uint32, (1,))[0]
print("Result:", result)
assert result == 30, "Fail: incorrect result"

• kernel: Defines a GPU kernel.
• var: Declares a local variable.
• @cpu: Embeds a Python code block that runs after the GPU finishes.

---

⚙️ Project Structure

sil/
├── generator/       # SPIR-V code generation logic
├── parser/          # Parser (SIL → AST)
├── runtime/         # pyopencl runtime interface
├── sil_tests/       # Test suite in .sil files
├── sil_ast.py       # AST node definitions
├── minisil.py       # Preprocessor for arrays
├── test_runner.py   # Runs and validates SIL tests
├── lexer.py         # Simple handwritten lexer for SIL
└── main.py          # Compiler entry point

---

🚀 Getting Started

1. Set up environment

python -m venv .venv
source .venv/bin/activate      # or .venv\Scripts\activate on Windows
pip install -r requirements.txt

2. Run a SIL file

python main.py sil_tests/basic_ops/basic_ops.sil

3. Run all tests

python test_runner.py

---

🧪 Example Test File

kernel bitwise_extreme(out: uint = array[2]) {
    var val: uint = 1;
    out[0] = bitwise{ val << 31 };
    out[1] = bitwise{ (val << 31) >> 31 };
}

@cpu
def bitwise_extreme(**inputs):
    val = 1
    shifted = (val << 31) & 0xFFFFFFFF
    recovered = (shifted >> 31) & 0xFFFFFFFF

    actual0 = inputs["OUT_0"][0]
    actual1 = inputs["OUT_1"][0]

    print(f"OUT_0 = {actual0}, esperado = {shifted}")
    print(f"OUT_1 = {actual1}, esperado = {recovered}")

    if actual0 != shifted or actual1 != recovered:
        raise Exception("Teste bitwise_extreme falhou")

---

📌 Limitations

• ❌ No native support for dynamic-length arrays
• ❌ No structs, functions, or recursion
• ❌ Limited type inference and no type polymorphism
• 🔧 Array support is emulated via the MiniSIL preprocessor
• 🚫 No full GPU thread model:
  • No get_global_id, get_local_id, or workgroup support
  • No atomics or barriers
  • Parallelism is limited due to lack of indexable memory

⚠️ The lack of true arrays made it extremely difficult to implement GPU features like get_global_id(), which limits the parallelism model.

---

🧰 Requirements

To compile and run SIL programs, make sure the following tools are installed:

• Vulkan SDK
  Includes spirv-as and spirv-val, which are required to assemble and validate SPIR-V binaries.

  After installing, make sure the SDK's bin/ directory is added to your system PATH.

• PyOpenCL
  Used as a runtime to load and launch SPIR-V kernels on the GPU.

You can verify installation with:

spirv-as --version
spirv-val --version

---

📄 License

This project is licensed under the GNU General Public License v3.0 (GPLv3).

© 2025 Hernani Samuel Diniz