Skip to content
Benchmarks

Works with uint8 models #8

Sign in to view logs

Works with uint8 models

Works with uint8 models #8

Workflow file for this run

.github/workflows/benchmark.yml at 9dda6d2
name: Benchmarks
on:
push:
branches:
- master
- update_benchmark
jobs:
testmacbenchmark:
name: Mac Benchmark
runs-on: [self-hosted, macOS]
if: github.repository_owner == 'tinygrad'
env:
PYTHONPATH: .
steps:
- name: Checkout Code
uses: actions/checkout@v3
- name: Symlink models and datasets
run: |
ln -s ~/tinygrad/disassemblers/applegpu disassemblers/applegpu
ln -s ~/tinygrad/weights/sd-v1-4.ckpt weights/sd-v1-4.ckpt
ln -s ~/tinygrad/weights/bpe_simple_vocab_16e6.txt.gz weights/bpe_simple_vocab_16e6.txt.gz
ln -s ~/tinygrad/weights/LLaMA weights/LLaMA
ln -s ~/tinygrad/extra/datasets/cifar-10-python.tar.gz extra/datasets/cifar-10-python.tar.gz
- name: Run model inference benchmark
run: python3 test/external/external_model_benchmark.py
- name: Test speed vs torch
run: BIG=2 MPS=1 python3 test/test_speed_v_torch.py | tee torch_speed.txt
shell: bash
- name: Run Tensor Core GEMM
run: DEBUG=2 python3 extra/gemm/simple_matmul.py | tee matmul.txt
shell: bash
- name: Run Stable Diffusion
run: python3 examples/stable_diffusion.py --noshow --timing | tee sd.txt
shell: bash
- name: Run LLaMA
run: |
JIT=0 python3 examples/llama.py --prompt "Hello." --count 10 --temperature 0 --timing | tee llama_unjitted.txt
JIT=1 python3 examples/llama.py --prompt "Hello." --count 10 --temperature 0 --timing | tee llama_jitted.txt
shell: bash
- name: Run GPT2
run: |
JIT=0 python3 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_unjitted.txt
JIT=1 python3 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_jitted.txt
shell: bash
- name: Run 10 CIFAR training steps
run: STEPS=10 python3 examples/hlb_cifar10.py | tee train_cifar.txt
shell: bash
- name: Run 10 CIFAR training steps w winograd
run: WINO=1 STEPS=10 python3 examples/hlb_cifar10.py | tee train_cifar_wino.txt
shell: bash
- uses: actions/upload-artifact@v3
with:
name: Speed (Mac)
path: |
onnx_inference_speed.csv
torch_speed.txt
train_cifar.txt
train_cifar_wino.txt
llama_unjitted.txt
llama_jitted.txt
gpt2_unjitted.txt
gpt2_jitted.txt
matmul.txt
sd.txt
testnvidiabenchmark:
name: NVIDIA Benchmark
runs-on: [self-hosted, Linux, CUDA]
if: github.repository_owner == 'tinygrad'
env:
PYTHONPATH: .
steps:
- name: Checkout Code
uses: actions/checkout@v3
- name: Run model inference benchmark
run: CUDA=1 python3 test/external/external_model_benchmark.py
- name: Test speed vs torch
run: CUDA=1 BIG=2 TORCHCUDA=1 python3 test/test_speed_v_torch.py | tee torch_speed.txt
shell: bash
- name: Run GPT2
run: |
CUDA=1 JIT=0 python3 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_unjitted.txt
CUDA=1 JIT=1 python3 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_jitted.txt
shell: bash
- uses: actions/upload-artifact@v3
with:
name: Speed (NVIDIA)
path: |
onnx_inference_speed.csv
torch_speed.txt
gpt2_unjitted.txt
gpt2_jitted.txt
testamdbenchmark:
name: AMD Benchmark
runs-on: [self-hosted, Linux, ROCM]
if: github.repository_owner == 'tinygrad'
env:
PYTHONPATH: .
steps:
- name: Checkout Code
uses: actions/checkout@v3
- name: Symlink models and datasets
run: |
ln -s ~/tinygrad/weights/sd-v1-4.ckpt weights/sd-v1-4.ckpt
ln -s ~/tinygrad/weights/bpe_simple_vocab_16e6.txt.gz weights/bpe_simple_vocab_16e6.txt.gz
ln -s ~/tinygrad/weights/LLaMA weights/LLaMA
ln -s ~/tinygrad/extra/datasets/cifar-10-python.tar.gz extra/datasets/cifar-10-python.tar.gz
- name: Run model inference benchmark
run: python3 test/external/external_model_benchmark.py
- name: Test speed vs torch
run: BIG=2 TORCHCUDA=1 python3 test/test_speed_v_torch.py | tee torch_speed.txt
shell: bash
- name: Run Tensor Core GEMM
run: HIP=1 HALF=1 DEBUG=2 python3 extra/gemm/simple_matmul.py | tee matmul.txt
shell: bash
- name: Run Stable Diffusion
run: python3 examples/stable_diffusion.py --noshow --timing | tee sd.txt
shell: bash
- name: Run LLaMA
run: |
JIT=0 python3 examples/llama.py --prompt "Hello." --count 10 --temperature 0 --timing | tee llama_unjitted.txt
JIT=1 python3 examples/llama.py --prompt "Hello." --count 10 --temperature 0 --timing | tee llama_jitted.txt
shell: bash
- name: Run GPT2
run: |
JIT=0 python3 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_unjitted.txt
JIT=1 python3 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_jitted.txt
shell: bash
- name: Run 10 CIFAR training steps
run: STEPS=10 python3 examples/hlb_cifar10.py | tee train_cifar.txt
- name: Run 10 CIFAR training steps w winograd
run: WINO=1 STEPS=10 python3 examples/hlb_cifar10.py | tee train_cifar_wino.txt
shell: bash
- name: Run 10 CIFAR training steps w WINO/HALF/HIP
run: HALF=1 HIP=1 WINO=1 STEPS=10 python3 examples/hlb_cifar10.py | tee train_cifar_wino_half_hip.txt
- uses: actions/upload-artifact@v3
with:
name: Speed (AMD)
path: |
onnx_inference_speed.csv
torch_speed.txt
train_cifar.txt
train_cifar_wino.txt
train_cifar_wino_half_hip.txt
llama_unjitted.txt
llama_jitted.txt
gpt2_unjitted.txt
gpt2_jitted.txt
matmul.txt
sd.txt