topic Re:x86GPUs occurs problem that douple precision is not supported in Intel® oneAPI Math Kernel Library

x86GPUs occurs problem that douple precision is not supported

qingfeng — Thu, 28 Oct 2021 09:49:14 GMT

when I use MKL on x86 GPU, compilation can pass but the double precision is not supported when executing. So can I solve it ?

回复： x86GPUs occurs problem that douple precision is not supported

qingfeng — Thu, 28 Oct 2021 10:23:12 GMT

I went to check the calling interface of the function and defined the data type of the input parameter as double, but my GPU is x86

Re:x86GPUs occurs problem that douple precision is not supported

VidyalathaB_Intel — Fri, 29 Oct 2021 10:01:29 GMT

Hi,

Thanks for reaching out to us.

>>my GPU is x86

Could you please let us know which GPU you are using for running your program along with your OS environment details and MKL version?

>>the calling interface of the function and defined the data type of the input parameter as double,

It would be a great help if you provide us with a sample reproducer so that we can try it from our end.

Please refer to the below link for more details regarding the system requirements of Intel oneMKL

https://www.intel.com/content/www/us/en/developer/articles/system-requirements/oneapi-math-kernel-library-system-requirements.html

Regards,

Vidya.

Re:x86GPUs occurs problem that douple precision is not supported

VidyalathaB_Intel — Fri, 05 Nov 2021 12:19:44 GMT

Hi,

Reminder:

As we have not heard back from you, do you have any updates? Please provide us with a sample reproducer and the above-mentioned details so that we can work on it from our end.

Regards,

Vidya.

回复： x86GPUs occurs problem that douple precision is not supported

qingfeng — Mon, 08 Nov 2021 02:10:06 GMT

Hi Vidya

The samples is block_cholesky_decomposition that it can be found in oneAPI-samples/Libraries/oneMKL/block_cholesky_decomposition at 18e90f52840ea509023caa8e8b909211ad26edc3 · oneapi-src/oneAPI-samples (github.com)

I looked at the source code and there is a check for double determinations.

Another error occurs when I comment out the check item. It is " terminate called after throwing an instance of 'cl::sycl::memory_allocation_error'"

Re:x86GPUs occurs problem that douple precision is not supported

VidyalathaB_Intel — Mon, 08 Nov 2021 06:18:48 GMT

Hi,

Thanks for providing the details of your code.

We tried it from our end and were able to run the block_cholesky_decomposition successfully.

Below are the environment details on which I have tried.

OS: Ubuntu 18.04

MKL: 2021.4.0

Device: Intel(R) UHD Graphics 630 [0x3e98]

>>can pass but the double precision is not supported when executing. So can I solve it

As per the system requirements, Gen9 or higher versions of GPU are supported.

Please refer to the below link for more details

https://www.intel.com/content/www/us/en/developer/articles/guide/intel-graphics-developers-guides.html

Could you please let us know which GPU you are using for running your program and also MKL version?

Regards,

Vidya.

回复： Re:x86GPUs occurs problem that douple precision is not supported

qingfeng — Mon, 08 Nov 2021 07:26:56 GMT

hi Vidya

Thanks for your reply.

Below are the environment details on which I have tried.

OS: 18.04.1-Ubuntu

MKL: 2021.4.0

Device: Vivante OpenCL Device CC8400.6305.0000

Regards,

qingfeng.

Re:x86GPUs occurs problem that douple precision is not supported

VidyalathaB_Intel — Tue, 09 Nov 2021 05:31:56 GMT

Hi,

Thanks for providing the details.

The following GPUs are the only supported devices for using intel oneMKL as of now.

Intel® HD Graphics

Intel® UHD Graphics for 9th, 10th, and 11th Gen Intel® Processors

Intel® Iris® Plus Graphics

Intel® Iris® Xe Graphics

Intel® Iris® Xe Max Graphics

Intel® Iris® Graphics

Intel® Iris® Pro Graphics

Please refer to the below link for more details.

https://www.intel.com/content/www/us/en/developer/articles/system-requirements/oneapi-math-kernel-library-system-requirements.html

>> Device: Vivante OpenCL Device CC8400.6305.0000... but the double precision is not supported when executing

We suggest you try the supported devices.

Please do let us know if you face any issues with the supported devices (Intel GPU's) if there is nothing else please confirm whether we can close this thread from our end.

Regards,

Vidya.

Re:x86GPUs occurs problem that douple precision is not supported

VidyalathaB_Intel — Tue, 09 Nov 2021 07:46:14 GMT

Hi,

Thanks for accepting our solution.

As the issue is resolved we are closing this thread. Please post a new question if you need any additional information from Intel as this thread will no longer be monitored.

Have a Nice Day!

Regards,

Vidya.