TinyGPU 🐉⚡

TinyGPU is a tiny educational GPU simulator - inspired by Tiny8, designed to demonstrate how GPUs execute code in parallel. It models a small SIMT (Single Instruction, Multiple Threads) system with per-thread registers, global memory, synchronization barriers, branching, and a minimal GPU-like instruction set.

🎓 Built for learning and visualization - see how threads, registers, and memory interact across cycles!

Odd-Even Sort	Reduction

---

🚀 What's New in v2.0.0

• Enhanced Instruction Set:
  • Added SHLD and SHST for robust shared memory operations.
  • Improved SYNC semantics for better thread coordination.
• Visualizer Improvements:
  • Export execution as GIFs with enhanced clarity.
  • Added support for saving visuals directly from the simulator.
• Refactored Core:
  • Simplified step semantics for better extensibility.
  • Optimized performance for larger thread counts.
• CI/CD Updates:
  • Integrated linting (ruff, black) and testing workflows.
  • Automated builds and tests on GitHub Actions.
• Documentation:
  • Expanded examples and added detailed usage instructions.

---

💡 Why TinyGPU?

TinyGPU was built as a learning-first GPU simulator - simple enough for beginners, but powerful enough to illustrate real GPU execution concepts like threads, synchronization, and divergence.

• ⚡ Lightweight:
  Zero heavy dependencies - runs entirely in Python with clean, readable code.
• 🧩 Educational:
  Demonstrates GPU fundamentals: thread execution, shared memory, branching, and synchronization barriers.
• 🔁 Fast feedback loop:
  Write a .tgpu file, run it, and instantly visualize how your threads interact.
• 🧠 Extensible:
  Add your own instructions, modify architecture parameters, or visualize new metrics.
• 🎨 Visual clarity:
  Each program can be rendered as a heatmap or exported as a GIF - perfect for teaching and demos.

---

👥 Who Should Use It?

• 🧑‍🎓 Students - learn GPU basics through hands-on examples.
• 👩‍🏫 Educators - build step-by-step visual lessons about parallel computing.
• 🧑‍🔬 Researchers & hobbyists - prototype kernel logic or explore synchronization models.
• 💡 Developers - experiment with SIMD-style concepts without real GPU dependencies.

---

🚦 Get Started

1. Install & run locally - see the Quickstart section below.
2. Explore the examples - try vector_add, odd_even_sort, and reduce_sum kernels.
3. Visualize execution - watch registers, memory, and threads evolve in real-time.
4. Experiment! - tweak instruction behavior or add your own to understand how GPUs schedule and synchronize work.

🧭 TinyGPU aims to make GPU learning intuitive, visual, and interactive - from classroom demos to self-guided exploration.

---

✨ Highlights

• 🧩 GPU-like instruction set:
  SET, ADD, MUL, LD, ST, JMP, BNE, BEQ, SYNC, CSWAP, SHLD, SHST.
• 🧠 Per-thread registers & PCs - each thread executes the same kernel independently.
• 🧱 Shared global memory for inter-thread operations.
• 🔄 Synchronization barriers (SYNC) for parallel coordination.
• 🎨 Visualizer & GIF exporter - view execution as heatmaps or export to animated GIFs.
• 🧮 Example kernels included:
  • Vector addition
  • Odd-even sort (GPU-style bubble sort)
  • Parallel reduction (sum of array)
  • Loop test & synchronization demo

---

🖼️ Example Visuals

Located in src/outputs/ — run the example scripts to generate these GIFs (they're saved under src/outputs/<script_name>/).

Example	Description	GIF Preview
Vector Add	Parallel vector addition (A+B -> C)
Block Shared Sum	Per-block shared memory sum example
Odd-Even Sort	GPU-style odd-even transposition sort
Parallel Reduction	Sum reduction across an array
Sync Test	Synchronization / barrier demonstration
Loop Test	Branching and loop behavior demo
Compare Test	Comparison and branching example
Kernel Args Test	Demonstrates passing kernel arguments

---

🚀 Quickstart

Clone and install

git clone https://github.com/deaneeth/tinygpu.git
cd tinygpu
pip install -e .
pip install -r requirements-dev.txt

Run an example

python -m examples.run_odd_even_sort

Produces: src/outputs/run_odd_even_sort/run_odd_even_sort_*.gif — a visual GPU-style sorting process.

Other examples

python -m examples.run_vector_add
python -m examples.run_reduce_sum
python -m examples.run_test_loop
python -m examples.run_sync_test

---

🧩 Project Layout

.
├─ .github/
│ └─ workflows/
│ └─ ci.yml
├─ docs/
│ └─ index.md
├─ examples/
│ ├─ odd_even_sort_tmp.tgpu
│ ├─ odd_even_sort.tgpu
│ ├─ reduce_sum.tgpu
│ ├─ run_odd_even_sort.py
│ ├─ run_reduce_sum.py
│ ├─ run_sync_test.py
│ ├─ run_test_loop.py
│ ├─ run_vector_add.py
│ ├─ sync_test.tgpu
│ ├─ test_loop.tgpu
│ └─ vector_add.tgpu
├─ src/outputs/
│ ├─ run_block_shared_sum/
│ ├─ run_odd_even_sort/
│ ├─ run_reduce_sum/
│ ├─ run_sync_test/
│ ├─ run_test_cmp/
│ ├─ run_test_kernel_args/
│ ├─ run_test_loop/
│ └─ run_vector_add/
├─ src/
│ └─ tinygpu/
│ ├─ __init__.py
│ ├─ assembler.py
│ ├─ gpu.py
│ ├─ instructions.py
│ └─ visualizer.py
├─ tests/
│ ├─ test_assembler.py
│ ├─ test_gpu_core.py
│ ├─ test_gpu.py
│ └─ test_programs.py
├─ LICENSE
├─ pyproject.toml
├─ README.md
└─ requirements-dev.txt

---

🧠 Assembly Reference (Mini ISA)

TinyGPU uses a minimal instruction set designed for clarity and education - simple enough for beginners, but expressive enough to build realistic GPU-style kernels.

Instruction	Operands	Description
SET Rd, imm	Rd = destination register, imm = immediate value	Set register Rd to an immediate constant.
ADD Rd, Ra, Rb	Rd = destination, Ra + Rb	Add two registers and store result in Rd.
ADD Rd, Ra, imm	Rd = destination, Ra + immediate	Add register and immediate value.
MUL Rd, Ra, Rb	Multiply two registers.	Rd = Ra * Rb
MUL Rd, Ra, imm	Multiply register by immediate.	Rd = Ra * imm
LD Rd, addr	Load from memory address into register.	Rd = mem[addr]
LD Rd, Rk	Load from address in register Rk.	Rd = mem[Rk]
ST addr, Rs	Store register into memory address.	mem[addr] = Rs
ST Rk, Rs	Store value from Rs into memory at address in register Rk.	mem[Rk] = Rs
JMP target	Label or immediate.	Unconditional jump — sets PC to target.
BEQ Ra, Rb, target	Branch if equal.	Jump to target if Ra == Rb.
BNE Ra, Rb, target	Branch if not equal.	Jump to target if Ra != Rb.
SYNC	(no operands)	Synchronization barrier — all threads must reach this point before continuing.
CSWAP addrA, addrB	Compare-and-swap memory values.	If mem[addrA] > mem[addrB], swap them. Used for sorting.
SHLD addr, Rs	Load shared memory into register.	Rs = shared_mem[addr]
SHST addr, Rs	Store register into shared memory.	shared_mem[addr] = Rs
CMP Rd, Ra, Rb(optional)	Compare and set flag or register.	Used internally for extended examples (e.g., prefix-scan).
NOP(optional)	(no operands)	No operation; placeholder instruction.

---

💡 Notes & Conventions

• Registers:
  • Each thread has its own register file (R0–R7 or more depending on configuration).
  • R7 is automatically initialized to the thread ID (tid).
• Memory:
  • Global memory is shared between all threads.
  • Addresses can be immediate (16) or register-based (R2).
• Labels & Comments:
  • Use labels like loop: for jumps.
  • Comments start with ;.
• Synchronization:
  • Use SYNC when threads must finish a phase before the next (e.g., sorting or reduction).
• Execution:
  • Each instruction executes per-thread.
  • Threads can branch independently (SIMT divergence is handled via per-thread PCs).

---

🧮 Example: Odd-Even Sort Kernel

SET R0,0 ; phase_counterSET R1,8 ; num_phases == N (set to array length here, adjust in runner)SET R3,0 ; parity (0 = even phase, 1 = odd phase)loop_phase: ; compute base index = tid * 2MUL R4, R7,2 ; R4 = tid * 2ADD R5, R4, R3 ; R5 = index = tid*2 + parityADD R6, R5,1 ; R6 = index + 1 CSWAP R5, R6 ; compare & swap memory[index], memory[index+1] SYNC ; synchronize threads across the phaseADD R3, R3,1 ; parity = parity + 1 BNE R3,2, noreset SET R3,0noreset:ADD R0, R0,1 BNE R0, R1, loop_phasedone:JMP done

---

🧩 Writing & Running Programs

1. Write your .tgpu assembly file in examples/.

2. Assemble and run it in Python:

   fromsrc.tinygpu.assemblerimportassemble_filefromsrc.tinygpu.gpuimportTinyGPUprog, labels=assemble_file("examples/vector_add.tgpu")
   gpu=TinyGPU(num_threads=8, num_registers=8, mem_size=64)
   gpu.load_program(prog, labels)
   gpu.run(max_cycles=100)

3. Visualize the run:

   fromsrc.tinygpu.visualizerimportvisualizevisualize(gpu, show_pc=True)

4. Export as GIF:

   fromsrc.tinygpu.visualizerimportsave_animationsave_animation(gpu, out_path="examples/my_run.gif", fps=10, max_frames=200)

---

🧰 Development & Testing

Run tests:

pytest

Run linters:

ruff .
black --check src/ tests/

CI builds and runs tests automatically on push/pull.

---

📘 License

MIT - see LICENSE

---

🌟 Credits & Inspiration

❤️ Built by Deaneeth

Inspired by the educational design of Tiny8 CPU Simulator.

TinyGPU extends these ideas into the world of parallel GPU computing, emphasizing clarity, simplicity, and visualization for all learners.