hls4ml installation

[amassahb]

Hi to all,
I am new here. I just started to install hls4ml library by running the following command in my home folder:
pip3 install hls4ml
but I am getting the following error:

Collecting hls4ml
  Using cached https://files.pythonhosted.org/packages/e9/64/30fe14f8ce47277c2a69d48a63538ed5bd54241dec388048d457e5cf4df3/hls4ml-0.6.0-py3-none-any.whl
Requirement already satisfied: pyyaml in /usr/lib64/python3.6/site-packages (from hls4ml)
Requirement already satisfied: numpy in /usr/lib64/python3.6/site-packages (from hls4ml)
Collecting onnx>=1.4.0 (from hls4ml)
  Using cached https://files.pythonhosted.org/packages/8f/71/1543d8dad6a26df1da8953653ebdbedacea9f1a5bcd023fe10f8c5f66d63/onnx-1.14.1.tar.gz
    Complete output from command python setup.py egg_info:
    fatal: Not a git repository (or any parent up to mount point /tmp)
    Stopping at filesystem boundary (GIT_DISCOVERY_ACROSS_FILESYSTEM not set).
    Traceback (most recent call last):
      File "<string>", line 1, in <module>
      File "/tmp/pip-build-svy0ktzx/onnx/setup.py", line 310, in <module>
        packages = setuptools.find_packages() + setuptools.find_namespace_packages(
    AttributeError: module 'setuptools' has no attribute 'find_namespace_packages'
    
    ----------------------------------------
Command "python setup.py egg_info" failed with error code 1 in /tmp/pip-build-svy0ktzx/onnx/

Do you have any idea?

Thank you in advance for your support.

Regards,
Amir

[vloncar]

You're using an old version of python (3.6), so you're attempting to install a last version of hls4ml that supported that python. You should upgrade to at least python 3.8 but preferably newer. We'll be moving to 3.10 as the requirement soon.

[amassahb]

Thank you for your reply. I am quite new with Python world.
I finally could install it without any error. However, when I try to import it I get the following errors:

`import hls4ml
WARNING:tensorflow:From C:\Program Files\Python310\lib\site-packages\keras\src\losses.py:2976: The name tf.losses.sparse_softmax_cross_entropy is deprecated. Please use tf.compat.v1.losses.sparse_softmax_cross_entropy instead.

Traceback (most recent call last):
File "<pyshell#0>", line 1, in
import hls4ml
File "C:\Program Files\Python310\lib\site-packages\hls4ml_init_.py", line 1, in
from hls4ml import converters, report, utils # noqa: F401
File "C:\Program Files\Python310\lib\site-packages\hls4ml\converters_init_.py", line 7, in
from hls4ml.converters.keras_to_hls import KerasFileReader # noqa: F401
File "C:\Program Files\Python310\lib\site-packages\hls4ml\converters\keras_to_hls.py", line 5, in
from hls4ml.model import ModelGraph
File "C:\Program Files\Python310\lib\site-packages\hls4ml\model_init_.py", line 1, in
from hls4ml.model.graph import HLSConfig, ModelGraph # noqa: F401
File "C:\Program Files\Python310\lib\site-packages\hls4ml\model\graph.py", line 9, in
from hls4ml.backends import get_backend
File "C:\Program Files\Python310\lib\site-packages\hls4ml\backends_init_.py", line 1, in
from hls4ml.backends.backend import Backend, get_available_backends, get_backend, register_backend # noqa: F401
File "C:\Program Files\Python310\lib\site-packages\hls4ml\backends\backend.py", line 5, in
from hls4ml.backends.template import Template
File "C:\Program Files\Python310\lib\site-packages\hls4ml\backends\template.py", line 1, in
from hls4ml.model.optimizer.optimizer import OptimizerPass
File "C:\Program Files\Python310\lib\site-packages\hls4ml\model\optimizer_init_.py", line 4, in
from hls4ml.model.optimizer.optimizer import ( # noqa: F401
File "C:\Program Files\Python310\lib\site-packages\hls4ml\model\optimizer\optimizer.py", line 5, in
from hls4ml.utils.string_utils import convert_to_snake_case
File "C:\Program Files\Python310\lib\site-packages\hls4ml\utils_init_.py", line 1, in
from hls4ml.utils.config import config_from_keras_model, config_from_onnx_model, config_from_pytorch_model # noqa: F401
File "C:\Program Files\Python310\lib\site-packages\hls4ml\utils\config.py", line 3, in
import qkeras
File "C:\Program Files\Python310\lib\site-packages\qkeras_init_.py", line 23, in
from .estimate import * # pylint: disable=wildcard-import
File "C:\Program Files\Python310\lib\site-packages\qkeras\estimate.py", line 41, in
from .qlayers import QActivation
File "C:\Program Files\Python310\lib\site-packages\qkeras\qlayers.py", line 55, in
from .quantizers import *
File "C:\Program Files\Python310\lib\site-packages\qkeras\quantizers.py", line 29, in
from .safe_eval import safe_eval
File "C:\Program Files\Python310\lib\site-packages\qkeras\safe_eval.py", line 21, in
from pyparsing import delimitedList
ModuleNotFoundError: No module named 'pyparsing'`

[amassahb]

Hi,
I found the reason and solved it :). Maybe it is worth it to give some information abut my platform. I installed hls4ml on Windows 11, Python 3.10.10, Vitis_HLS 2023.2.
I finally could install it and tried to run this example code which is given in your website:

import hls4ml
config = hls4ml.utils.fetch_example_model('KERAS_3layer.json')
print(config)
hls_model = hls4ml.converters.keras_to_hls(config)
hls_model.compile()

I can see that the model is created successfully, but I get an error for compiling the model which is:

'bash' is not recognized as an internal or external command, operable program or batch file. Traceback (most recent call last): File "c:\hls4ml\test.py", line 11, in <module> hls_model.compile() File "C:\Program Files\Python310\lib\site-packages\hls4ml\model\graph.py", line 662, in compile self._compile() File "C:\Program Files\Python310\lib\site-packages\hls4ml\model\graph.py", line 665, in _compile lib_name = self.config.backend.compile(self) File "C:\Program Files\Python310\lib\site-packages\hls4ml\backends\fpga\fpga_backend.py", line 131, in compile raise Exception(f'Failed to compile project "{model.config.get_project_name()}"') Exception: Failed to compile project "myproject"

It seems the problem is in sth related to bash files or sth like that.

I would be really grateful if you could help me or give me some hints dude.

Thank you so much and looking forward to hearing from you.

Regards,
Amir

[vloncar]

Here's a hint dude, hls4ml doesn't work on Windows (see the docs). You'll have to use Linux (via VM or WSL). Sorry :-(

[amassahb]

Ok, I installed an Ubuntu for that, Ubuntu 22.4 with python 3.10.12. I cold install the hls4ml without any error. I can see it by running:

pip show hls4ml

But when I import it into my code and run it, the import line code leads to a crash and error is "Illegal instruction (core dumped)".
This is my code with only a print and the import line:

print("#######################################################Test1") import hls4ml print("Test2")

I tried to run it in a verbose mode (by using this command "python -v test.py") to see if there is any error, but I cannot see any. I have attached the log file.
log.txt

At the beginning I though it might be because of Python version, and I tried with python 3.9.18 but not difference.

I would appreciate any support.

Sincerely yours,
Amir

[amassahb]

One more point. I have installed my Ubuntu on the VirtualBox. Could it be the reason?

[jmitrevs]

It seems like there's an issue with your tensorflow installation. I tend to setup my python environment with miniforge with the following packages:

conda create -n fastml python=3.10
conda activate fastml
conda install tensorflow pytorch numpy matplotlib scikit-learn pandas pytest pytest-cov jupyterlab toposort clize seaborn pydot

And then pip install hls4ml (not conda install).

[brtsfr]

hello amassahb, how was your problem? did you solve it then?

[brtsfr]

hi i'm new learner here. i also encounted the same situation. i am running ubuntu24.04 in virtualbox of Win11. "pip show hls4ml" indicated a successful installation. but when i typed “hls4ml -h” in cmd, or "import hls4ml" in python, only got "illegal instruction". please help

[Rezef]

Thanks a lot, managed to run the first tutorial notebook completely

[bluemoon0321]

hello Rezef, may I ask which tutorial did you finish?
And did you use it on a ubuntu 24.04 and did these steps after cloning to setup the env?

Thanks!

[Rezef]

@bluemoon0321 I was talking about this tutorial repo:
https://github.com/fastmachinelearning/hls4ml-tutorial
I installed this on ubuntu 22.02. You just need to create an environment based on the environment.yml mentioned in the tutorial repo. But you should have a Vivado(2020.1) or Vitis backend(23.2) installed in your local machine or virtual machine for this to work

[bluemoon0321]

@Rezef oh ok got it.
One more question .Did you use Vivado through the whole tutorial? because I have vitis(24.2) installed and can successful run the first few cells in part 1, but I see that for part 7 you need to use Vivado. I'm not sure do I also need to install Vivado(2019.2/2020.1) or can I do it through vitis installed only.
Thanks a lot!

[bo3z]

For part 7 you currently need access to Vivado HLS. In the future, there will be support to run part 7 using Vitis HLS, as there is a draft PR for that currently.