inducer / codepy Goto Github PK

In this discussion of using thrust with pycuda a few years ago, it was stated that using multiple modules while creating both a Boost::Python module and cuda code was necessary because nvcc couldn't handle boost.

I can now compile a simple Boost::Python module together with pycuda code, as seen in this gist.

It might be possible to simplify cuda.py now?

Hello

Would it be possible to burn a new version as the latest commit (sleep_delay) is not picked up by setup.py like an installation that depends on codepy.

Thank

I don't understand why it is necessary to use CudaGlobal in nvcc-test.py, why can't we just add those functions to the CudaModule directly like we see here?

I'm really interesting about codepy an pycuda, but i got some bug when i run python demo.py.

/home/a01/anaconda3/envs/pycuda/compiler_compat/ld: cannot find -lboost_python-py38
/home/a01/anaconda3/envs/pycuda/compiler_compat/ld: cannot find -lpython3.8m
collect2: error: ld returned 1 exit status

i have google about how to fix it but it doesn't work , and it same when i try pycuda/from-wiki python thrust_interop.py.

my computer enviroment:
ubuntu1804
conda virtual python:3.8
pycuda and codepy have been successfully installed

Hi,

So I'm a user of a package which depends on codepy (devitoproject.org). Since the compilations it generates can be quite slow (several minutes each), parallelism is obviously desirable if possible. The current locking system in codepy precludes this, if we want to also make use of the cache (which we do - because several minutes per compilation!).

Just so I don't start down a completely unwelcome path - can you please advise if a patch which changes the locking to be one-lockfile-per-hash (rather than a single lock on the whole cache directory) would be welcome?

Thanks!

When running this code from:
http://pastebin.com/pr15PirN

I get this error.

http://pastebin.com/Rbjdi7Dj

Hi there,

I'm struggling to make codepy to work with the following gcc version:

gcc (Ubuntu 11.2.0-19ubuntu1) 11.2.0
Copyright (C) 2021 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

Now, I really need it because it is required by a tool that performs scaffolding and leverages both GPU acceleration and PyCUDA. If it is helpful I cna report the error; however, based on what the developer for the tool said codepy for working need gcc vesion not greater than 8.x.x.

On the other hand, I really do not want to remove my gcc version since this might cause issues with XOrg driver (which I experienced before and are rather unpleasant to deal with). I also tried to install gcc 8.5.0 from shell, but unsuccessfully; later, I resorted to install it from the .bin file but even so, there is no way for me to use it in combination with the tool I need.

Thus, I was wondering is there any chance I can make it work with gcc 11.2.0

Moving discussion of inducer/pycuda#25 over here, where it should have been in the first place.

The docs righly state that dynamic compilation and linking is now working on Windows. Is there any plans to remedy this situation?

I have had a small peek on how the toolchains are being set up in CodePy, and I guess it should be possible to add MSVC to it as well. Continuing on this path, will one not potentially end up with having duplicated a cross-platform build system in Python. Would it not be better to utilize Bjam? It should be available on any system using Boost anyway to abstract away the platform independent anyways?

(On CUDA systems nvcc also does the build platform abstraction, so adding a third one
seems a bit excessive, CMake is of course another alternative to cross platform building, which also can call NVCC.)

Or is this new modePy-stuff going to replace CodePy