OpenCL detects GPU but sycl-ls does not display GPU

[Nilp]

Hi,

I recently ran the below commands on my Ubunutu 20.04 machine with an 11th Gen Intel(R) Core(TM) i9-11950H and a discrete Nvidia RTX A3000.

sudo apt update
sudo apt upgrade

The following modules were installed and upgraded.

Install: linux-oem-5.14-headers-5.14.0-1057:amd64 (5.14.0-1057.64, automatic), linux-headers-5.14.0-1057-oem:amd64 (5.14.0-1057.64, automatic), linux-image-5.14.0-1057-oem:amd64 (5.14.0-1057.64, automatic), docker-buildx-plugin:amd64 (0.10.2-1~ubuntu.20.04~focal, automatic), linux-modules-5.14.0-1057-oem:amd64 (5.14.0-1057.64, automatic)
Upgrade: libpam0g:amd64 (1.3.1-5ubuntu4.4, 1.3.1-5ubuntu4.6), nvidia-docker2:amd64 (2.11.0-1, 2.12.0-1), linux-image-oem-20.04d:amd64 (5.14.0.1056.54, 5.14.0.1057.55), libnvidia-container-tools:amd64 (1.11.0-1, 1.12.0-1), libpam-modules:amd64 (1.3.1-5ubuntu4.4, 1.3.1-5ubuntu4.6), update-notifier-common:amd64 (3.192.30.14, 3.192.30.16), ubuntu-advantage-tools:amd64 (27.13.2~20.04.1, 27.13.3~20.04.1), linux-headers-oem-20.04d:amd64 (5.14.0.1056.54, 5.14.0.1057.55), docker-ce-rootless-extras:amd64 (5:20.10.23~3-0~ubuntu-focal, 5:23.0.0-1~ubuntu.20.04~focal), libpam-runtime:amd64 (1.3.1-5ubuntu4.4, 1.3.1-5ubuntu4.6), thunderbird-gnome-support:amd64 (1:102.4.2+build2-0ubuntu0.20.04.1, 1:102.7.1+build2-0ubuntu0.20.04.1), nvidia-container-toolkit-base:amd64 (1.11.0-1, 1.12.0-1), libnvidia-container1:amd64 (1.11.0-1, 1.12.0-1), libpam-modules-bin:amd64 (1.3.1-5ubuntu4.4, 1.3.1-5ubuntu4.6), thunderbird:amd64 (1:102.4.2+build2-0ubuntu0.20.04.1, 1:102.7.1+build2-0ubuntu0.20.04.1), code:amd64 (1.74.3-1673284829, 1.75.0-1675266613), firefox:amd64 (109.0+build2-0ubuntu0.20.04.1, 109.0.1+build1-0ubuntu0.20.04.2), tmux:amd64 (3.0a-2ubuntu0.3, 3.0a-2ubuntu0.4), linux-oem-20.04c:amd64 (5.14.0.1056.54, 5.14.0.1057.55), linux-oem-20.04d:amd64 (5.14.0.1056.54, 5.14.0.1057.55), nvidia-container-toolkit:amd64 (1.11.0-1, 1.12.0-1), update-notifier:amd64 (3.192.30.14, 3.192.30.16), docker-ce:amd64 (5:20.10.23~3-0~ubuntu-focal, 5:23.0.0-1~ubuntu.20.04~focal), docker-ce-cli:amd64 (5:20.10.23~3-0~ubuntu-focal, 5:23.0.0-1~ubuntu.20.04~focal)

I then rebooted the system and found out that initramfs was not installed and had to install that. Once it finally booted without errors, I was trying to run some SYCL based software and found out that "sycl-ls" command was not detecting the GPU.

$ sycl-ls
[opencl:acc:0] Intel(R) FPGA Emulation Platform for OpenCL(TM), Intel(R) FPGA Emulation Device 1.2 [2022.15.12.0.01_081451]
[opencl:cpu:1] Intel(R) OpenCL, 11th Gen Intel(R) Core(TM) i9-11950H @ 2.60GHz 3.0 [2022.15.12.0.01_081451]
[opencl:acc:2] Intel(R) FPGA Emulation Platform for OpenCL(TM), Intel(R) FPGA Emulation Device 1.2 [2022.15.12.0.01_081451]
[opencl:cpu:3] Intel(R) OpenCL, 11th Gen Intel(R) Core(TM) i9-11950H @ 2.60GHz 3.0 [2022.15.12.0.01_081451]

I then ran the "cliinfo" command and found that openCL was detecting the GPU.

$ cliinfo
...
 Platform Name                                   NVIDIA CUDA
Number of devices                                 1
  Device Name                                     NVIDIA RTX A3000 Laptop GPU
  Device Vendor                                   NVIDIA Corporation
  Device Vendor ID                                0x10de
  Device Version                                  OpenCL 3.0 CUDA
  Driver Version                                  525.78.01
  Device OpenCL C Version                         OpenCL C 1.2 
  Device Type                                     GPU
  Device Topology (NV)                            PCI-E, 01:00.0
  Device Profile                                  FULL_PROFILE
  Device Available                                Yes
  Compiler Available                              Yes
  Linker Available                                Yes
  Max compute units                               32
  Max clock frequency                             1560MHz
  Compute Capability (NV)                         8.6
...

Can I get some help to understand why SYCL is not detecting a GPU anymore after the upgrade. My inital guess is that it could be the possible kernal upgrade that occurred in the upgrade.

Cheers

[ShivaniK_Intel]

Hi,

“Thank you for your inquiry. We can only offer direct support for Intel hardware platforms that the Intel® oneAPI product supports. Intel provides instructions on how to compile oneAPI code for both CPU and a wide range of GPU accelerators.

 https://intel.github.io/llvm-docs/GetStartedGuide.html”

Thanks & Regards

Shivani

[ShivaniK_Intel]

Hi,

If you need any additional information, please post a new question as this thread will no longer be monitored by Intel.

Thanks & Regards

Shivani