commaai
/
tinygrad
mirror of https://github.com/commaai/tinygrad.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

tinybox benchmark with HSA (#3603) 

* tinybox benchmark with HSA

* torch cuda init can fail

* no TORCHCUDA

* print torch version

* LD_PRELOAD="/opt/rocm/lib/libhsa-runtime64.so"
This commit is contained in:
chenyu 2024-03-05 11:03:52 -05:00 committed by GitHub
parent f500be1313
commit 3c3f846c45
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 15 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

28
.github/workflows/benchmark.yml vendored
View File

@ -140,30 +140,32 @@ jobs:
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: HIP=1 NOCLANG=1 python3 test/external/external_model_benchmark.py
run: LD_PRELOAD="/opt/rocm/lib/libhsa-runtime64.so" HSA=1 NOCLANG=1 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
run: |
python3 -c "import torch; print(torch.__version__)"
LD_PRELOAD="/opt/rocm/lib/libhsa-runtime64.so" HSA=1 BIG=2 TORCHCUDA=1 python3 test/test_speed_v_torch.py | tee torch_speed.txt
- name: Run Tensor Core GEMM
run: HIP=1 HALF=1 DEBUG=2 python3 extra/gemm/simple_matmul.py | tee matmul.txt
run: HSA=1 HALF=1 DEBUG=2 python3 extra/gemm/simple_matmul.py | tee matmul.txt
- name: Run Stable Diffusion
run: python3 examples/stable_diffusion.py --seed 0 --noshow --timing | tee sd.txt
- name: Run LLaMA (with HIP)
run: HSA=1 python3 examples/stable_diffusion.py --seed 0 --noshow --timing | tee sd.txt
- name: Run LLaMA
run: |
HIP=1 JIT=0 python3 examples/llama.py --gen 1 --prompt "Hello." --count 10 --temperature 0 --timing | tee llama_unjitted.txt
HIP=1 JIT=1 python3 examples/llama.py --gen 1 --prompt "Hello." --count 10 --temperature 0 --timing | tee llama_jitted.txt
- name: Run GPT2 (with HIP)
HSA=1 JIT=0 python3 examples/llama.py --gen 1 --prompt "Hello." --count 10 --temperature 0 --timing | tee llama_unjitted.txt
HSA=1 JIT=1 python3 examples/llama.py --gen 1 --prompt "Hello." --count 10 --temperature 0 --timing | tee llama_jitted.txt
- name: Run GPT2
run: |
HIP=1 JIT=0 python3 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_unjitted.txt
HIP=1 JIT=1 python3 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_jitted.txt
HSA=1 JIT=0 python3 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_unjitted.txt
HSA=1 JIT=1 python3 examples/gpt2.py --prompt "Hello." --count 10 --temperature 0 --timing | tee gpt2_jitted.txt
- name: Run 10 CIFAR training steps
run: HIP=1 STEPS=10 python3 examples/hlb_cifar10.py | tee train_cifar.txt
run: HSA=1 STEPS=10 python3 examples/hlb_cifar10.py | tee train_cifar.txt
- name: Run 10 CIFAR training steps w HALF
run: HIP=1 STEPS=10 HALF=1 python3 examples/hlb_cifar10.py | tee train_cifar_half.txt
run: HSA=1 STEPS=10 HALF=1 python3 examples/hlb_cifar10.py | tee train_cifar_half.txt
# # TODO: enable this. it took 3 minutes in CI and made the full training one more than 5 minutes
# - name: Run 10 CIFAR training steps w 6 GPUS
# run: time HALF=1 STEPS=10 BS=1536 GPUS=6 python3 examples/hlb_cifar10.py
- name: Run full CIFAR training
run: time HIP=1 HALF=1 LATEWINO=1 STEPS=1000 TARGET_EVAL_ACC_PCT=93 python3 examples/hlb_cifar10.py | tee train_cifar_one_gpu.txt
run: time HSA=1 HALF=1 LATEWINO=1 STEPS=1000 TARGET_EVAL_ACC_PCT=93 python3 examples/hlb_cifar10.py | tee train_cifar_one_gpu.txt
- uses: actions/upload-artifact@v4
with:
name: Speed (AMD)