Intel® oneAPI Base Toolkit
Support for core tools and libraries to build and deploy high-performance data-centric applications
273 Discussions

Installation questions and intel-nvidia

RN1
New Contributor I
451 Views

Hi,

I am asking these in case you can confirm my questions before spending days trying to build and do tests without luck.

 

1. I own Xeon E5-2620. It is capable of running oneAPI. It has a kepler NVIDIA board (compute capability 3.5). The CUDA support for oneAPI (intel/llvm SYCL) is with 5.0 or higher, although some docs says with less value some features could not work (which ones? it is worth, maybe I can launch simple code?)

 

2. I own an old i3-4160 with a Intel graphics 4400. It seems oneAPI cannot run, isn't it? Can this platform run a NVIDIA SYCL GPU with oneAPI? or does it needs support also oneAPI from the host (generation over 6000)?

 

3. Regarding interoperability between Intel oneAPI and CUDA (SYCL backend). Can I build a program that has some regions where the code is launched to the intel GPU (gpu_selector for HD graphics) and others where it is offloaded to the NVIDIA GPU (gpu_selector for nvidia)? Or when using the llvm (these instructions https://intel.github.io/llvm-docs/GetStartedGuide.html#build-dpc-toolchain) the whole binary can only offload to the GPU? Currently I compile with dpcpp from oneAPI installation, but the instructions above always use "clang++". What would be the difference between both ways of compiling/linking/executing?

 

Regards

0 Kudos
3 Replies
GouthamK_Intel
Moderator
417 Views

Hi,

Thanks for reaching out to us!

This forum is intended to support Intel oneAPI Products only. As we can see that you are using an open-source version of oneAPI for Nvidia devices support, we request you to raise an issue in GitHub for faster response: https://github.com/intel/llvm/issues

 If you intended to use Intel oneAPI Products, one place to download a free community version is the Intel oneAPI Base Toolkit and follow the installation guide to install the toolkit and required driver.

After installing the Intel oneAPI Base toolkit, you can refer to the below Get-Started guide for Linux.

https://software.intel.com/content/www/us/en/develop/documentation/get-started-with-intel-oneapi-bas...

 

>>I own an old i3-4160 with a Intel graphics 4400. It seems oneAPI cannot run, isn't it?

In order to run your DPCPP codes on GPU, you need to have GEN9 (and higher) GPUs. System Requirements: https://software.intel.com/content/www/us/en/develop/articles/intel-oneapi-base-toolkit-system-requi...

 

>> What would be the difference between both ways of compiling/linking/executing?

Intel oneAPI DPCPP is a compiler driver which uses an ICX compiler under the hood. Whereas the open-source version of oneAPI uses a clang++ compiler. The behavior of Compiling/Linking/Execution of the codes is pretty much the same. 

 

Have a Good day!

 

Thanks & Regards

Goutham

 

GouthamK_Intel
Moderator
342 Views

Hi Raul,

Reminder: Could you please let us know if your issue is resolved? and whether we can close this thread from our end.


Regards

Goutham


GouthamK_Intel
Moderator
312 Views

Hi,

As we have not heard back from you, we are considering that your issue has been resolved and we have answered all your queries. So we will no longer respond to this thread. If you require additional assistance from Intel, please start a new thread. Any further interaction in this thread will be considered community only 

Have a Good day!


Thanks & Regards

Goutham


Reply