triton icon indicating copy to clipboard operation
triton copied to clipboard

Python 3.11 - Triton not available?

Open Piscabo opened this issue 1 year ago • 3 comments

Python 3.11 - Triton not available? Where can find a version that works?

(comfyVENV) C:\comfyVENV>pip install -U --index-url https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/Triton-Nightly/pypi/simple/ triton-nightly Looking in indexes: https://aiinfra.pkgs.visualstudio.com/PublicPackages/_packaging/Triton-Nightly/pypi/simple/, https://pypi.nvidia.com ERROR: Could not find a version that satisfies the requirement triton-nightly (from versions: none) ERROR: No matching distribution found for triton-nightly

Piscabo avatar Jan 06 '24 09:01 Piscabo

same error on 3.12.0

jhirschibar avatar Jan 06 '24 18:01 jhirschibar

Are you by any chance using Windows? I got this error when I tried to install it on Windows without realizing that Triton is not available for the same

aflah02 avatar Jan 06 '24 21:01 aflah02

Are you by any chance using Windows? I got this error when I tried to install it on Windows without realizing that Triton is not available for the same

Triton for Windows is available. Just not for Python 3.11 pip install https://huggingface.co/r4ziel/xformers_pre_built/resolve/main/triton-2.0.0-cp310-cp310-win_amd64.whl

Piscabo avatar Jan 08 '24 18:01 Piscabo

Are you by any chance using Windows? I got this error when I tried to install it on Windows without realizing that Triton is not available for the same

Triton for Windows is available. Just not for Python 3.11 pip install https://huggingface.co/r4ziel/xformers_pre_built/resolve/main/triton-2.0.0-cp310-cp310-win_amd64.whl

cd python
python setup.py bdist_wheel

version triton 3 is compatible with python 3.11 & 12

johnnynunez avatar Mar 12 '24 09:03 johnnynunez

Are you by any chance using Windows? I got this error when I tried to install it on Windows without realizing that Triton is not available for the same

Triton for Windows is available. Just not for Python 3.11 pip install https://huggingface.co/r4ziel/xformers_pre_built/resolve/main/triton-2.0.0-cp310-cp310-win_amd64.whl

ERROR: triton-2.0.0-cp310-cp310-win_amd64.whl is not a supported wheel on this platform.

Keramatfar avatar May 28 '24 14:05 Keramatfar

Are you by any chance using Windows? I got this error when I tried to install it on Windows without realizing that Triton is not available for the same

Triton for Windows is available. Just not for Python 3.11 pip install https://huggingface.co/r4ziel/xformers_pre_built/resolve/main/triton-2.0.0-cp310-cp310-win_amd64.whl

Well, the official Readme.md file says that the trition only support Linux:

Compatibility
Supported Platforms:

Linux
Supported Hardware:

NVIDIA GPUs (Compute Capability 7.0+)
Under development: AMD GPUs, CPUs

If you can run trition on Windows, please tell me how:)

Marches7 avatar Jun 08 '24 15:06 Marches7

pip install https://huggingface.co/madbuda/triton-windows-builds/resolve/main/triton-2.1.0-cp311-cp311-win_amd64.whl

I found this build of triton for python 3.11 on Windows.

fancifulcrow avatar Jun 08 '24 18:06 fancifulcrow

I installed triton to python 3.11 on windows 11. Got error of ImportError: cannot import name 'get_cuda_stream' from 'triton.runtime.jit' (C:\Users\xx\AppData\Roaming\Python\Python311\site-packages\triton\runtime\jit.py)

Here is the trace: Traceback (most recent call last): File "H:\github\nanoGPT\train.py", line 210, in <module> model = torch.compile(model) # requires PyTorch 2.0 ^^^^^^^^^^^^^^^^^^^^ File "C:\Users\xx\AppData\Roaming\Python\Python311\site-packages\torch\__init__.py", line 1826, in compile return torch._dynamo.optimize(backend=backend, nopython=fullgraph, dynamic=dynamic, disable=disable)(model) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\xx\AppData\Roaming\Python\Python311\site-packages\torch\_dynamo\eval_frame.py", line 795, in optimize compiler_config=backend.get_compiler_config() ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\Users\xx\AppData\Roaming\Python\Python311\site-packages\torch\__init__.py", line 1671, in get_compiler_config from torch._inductor.compile_fx import get_patched_config_dict File "C:\Users\xx\AppData\Roaming\Python\Python311\site-packages\torch\_inductor\compile_fx.py", line 49, in <module> from .fx_passes.joint_graph import joint_graph_passes File "C:\Users\xx\AppData\Roaming\Python\Python311\site-packages\torch\_inductor\fx_passes\joint_graph.py", line 12, in <module> from ..pattern_matcher import ( File "C:\Users\xx\AppData\Roaming\Python\Python311\site-packages\torch\_inductor\pattern_matcher.py", line 44, in <module> from .lowering import fallback_node_due_to_unsupported_type File "C:\Users\xx\AppData\Roaming\Python\Python311\site-packages\torch\_inductor\lowering.py", line 5394, in <module> from . import kernel File "C:\Users\xx\AppData\Roaming\Python\Python311\site-packages\torch\_inductor\kernel\__init__.py", line 1, in <module> from . import mm, mm_common, mm_plus_mm, unpack_mixed_mm File "C:\Users\xx\AppData\Roaming\Python\Python311\site-packages\torch\_inductor\kernel\mm.py", line 8, in <module> from ..codegen.cuda.gemm_template import CUTLASSGemmTemplate File "C:\Users\xx\AppData\Roaming\Python\Python311\site-packages\torch\_inductor\codegen\cuda\gemm_template.py", line 11, in <module> from .cuda_kernel import CUDATemplateKernel File "C:\Users\xx\AppData\Roaming\Python\Python311\site-packages\torch\_inductor\codegen\cuda\cuda_kernel.py", line 7, in <module> from ...select_algorithm import ChoiceCaller File "C:\Users\xx\AppData\Roaming\Python\Python311\site-packages\torch\_inductor\select_algorithm.py", line 24, in <module> from .codegen.triton import texpr, TritonKernel, TritonPrinter, TritonScheduling File "C:\Users\xx\AppData\Roaming\Python\Python311\site-packages\torch\_inductor\codegen\triton.py", line 31, in <module> from ..triton_heuristics import AutotuneHint File "C:\Users\xx\AppData\Roaming\Python\Python311\site-packages\torch\_inductor\triton_heuristics.py", line 54, in <module> from triton.runtime.jit import get_cuda_stream ImportError: cannot import name 'get_cuda_stream' from 'triton.runtime.jit' (C:\Users\xx\AppData\Roaming\Python\Python311\site-packages\triton\runtime\jit.py)

seanzhou321 avatar Jul 08 '24 02:07 seanzhou321