# \#opencl **URL:** https://forums.developer.nvidia.com/tag/opencl/374.md [Latest](https://forums.developer.nvidia.com/latest.md) · [Categories](https://forums.developer.nvidia.com/categories.md) · [Tags](https://forums.developer.nvidia.com/tags.md) --- ## [OpenCL variable-amount rotate(x, y) extra mask instruction in PTX](https://forums.developer.nvidia.com/t/opencl-variable-amount-rotate-x-y-extra-mask-instruction-in-ptx/372144)
**Author:** [@mix](https://forums.developer.nvidia.com/u/mix)\ **Replies:** 4\ **Last updated:** [November 6, 2026, 10:13pm UTC](https://forums.developer.nvidia.com/t/opencl-variable-amount-rotate-x-y-extra-mask-instruction-in-ptx/372144 "2026-11-06T22:13:27Z")
NVidia Developers, I have an OpenCL macro: 'define ROTL(x, s) rotate((uint)(x), (uint)(s)) It appears that the NVVM Compiler generates an explicit bitwise AND (and.b32) before every variable rotate / funnel shift (shf… --- ## ["Launching the workload is taking more time than expected" - why?](https://forums.developer.nvidia.com/t/launching-the-workload-is-taking-more-time-than-expected-why/379628)
**Author:** [@eyal.rozenberg1](https://forums.developer.nvidia.com/u/eyal.rozenberg1)\ **Replies:** 1\ **Last updated:** [October 29, 2026, 12:44pm UTC](https://forums.developer.nvidia.com/t/launching-the-workload-is-taking-more-time-than-expected-why/379628 "2026-10-29T12:44:28Z")
I’m profiling a certain kernel using NSight Compute. The profiling is taking quite some time, and at some point I get the message: ==WARNING== Launching the workload is taking more time than expected and a suggestion… --- ## [Announcement: opencl-api-cpp - A 'sister' library to cuda-api-wrappers, but for OpenCL](https://forums.developer.nvidia.com/t/announcement-opencl-api-cpp-a-sister-library-to-cuda-api-wrappers-but-for-opencl/383931)
**Author:** [@epk](https://forums.developer.nvidia.com/u/epk)\ **Replies:** 0\ **Last updated:** [September 21, 2026, 7:17pm UTC](https://forums.developer.nvidia.com/t/announcement-opencl-api-cpp-a-sister-library-to-cuda-api-wrappers-but-for-opencl/383931 "2026-09-21T19:17:44Z")
For those of you who do work in OpenCL, I have published a library: –==\[ opencl-api-cpp \]==– which is a set of headers-only modern-C++ bindings for OpenCL. Yes, I know you might be aware of the CLHPP C++ bindings whic… --- ## [GTX 1660 Super (TU116) not exposing FP16 on driver 580.94.16](https://forums.developer.nvidia.com/t/gtx-1660-super-tu116-not-exposing-fp16-on-driver-580-94-16/359199)
**Author:** [@tegarargo](https://forums.developer.nvidia.com/u/tegarargo)\ **Replies:** 6\ **Last updated:** [June 11, 2026, 11:48am UTC](https://forums.developer.nvidia.com/t/gtx-1660-super-tu116-not-exposing-fp16-on-driver-580-94-16/359199 "2026-06-11T11:48:21Z")
Good day, I recently did a clpeak test and it seems that Nvidia’s OpenCL driver on linux is not exposing cl\_khr\_fp16 even though Nvidia clearly advertises TU116 as having a separate FP16 pipe (and therefore double rate F… --- ## [\[BUG\] OpenCL #include directive causes CL\_OUT\_OF\_HOST\_MEMORY on driver 595.97](https://forums.developer.nvidia.com/t/bug-opencl-include-directive-causes-cl-out-of-host-memory-on-driver-595-97/366719)
**Author:** [@jean-michel.rouet](https://forums.developer.nvidia.com/u/jean-michel.rouet)\ **Replies:** 7\ **Last updated:** [May 7, 2026, 1:49pm UTC](https://forums.developer.nvidia.com/t/bug-opencl-include-directive-causes-cl-out-of-host-memory-on-driver-595-97/366719 "2026-05-07T13:49:38Z")
NVIDIA OpenCL Bug Report: include directive causes CL\_OUT\_OF\_HOST\_MEMORY Summary Using #include directives in OpenCL kernel source code causes clBuildProgram() to fail with CL\_OUT\_OF\_HOST\_MEMORY (-6) on recent NVIDIA dri… --- ## [Looking for clpeak results on Jetson Orin Nano Super](https://forums.developer.nvidia.com/t/looking-for-clpeak-results-on-jetson-orin-nano-super/367586)
**Author:** [@pauljurczak](https://forums.developer.nvidia.com/u/pauljurczak)\ **Replies:** 5\ **Last updated:** [May 7, 2026, 12:38am UTC](https://forums.developer.nvidia.com/t/looking-for-clpeak-results-on-jetson-orin-nano-super/367586 "2026-05-07T00:38:36Z")
The title says it all. I would appreciate the results if someone has them. --- ## [OpenCL & OpenGL Interop in Wayland](https://forums.developer.nvidia.com/t/opencl-opengl-interop-in-wayland/361973)
**Author:** [@martin.kodovsky](https://forums.developer.nvidia.com/u/martin.kodovsky)\ **Replies:** 1\ **Last updated:** [March 11, 2026, 10:49pm UTC](https://forums.developer.nvidia.com/t/opencl-opengl-interop-in-wayland/361973 "2026-03-11T22:49:41Z")
Hello, recently I switched to Wayland from X11 (using Fedora43 with either Gnome or KDE/Plasma as window manager). Unfortunately, my personal project written in C/glfw/glut/OpenCL stopped working in runtime. It is using… --- ## [Registering a numa-allocated host buffer with the GPU DMA engine for peak transfer performance in OpenCL](https://forums.developer.nvidia.com/t/registering-a-numa-allocated-host-buffer-with-the-gpu-dma-engine-for-peak-transfer-performance-in-opencl/361614)
**Author:** [@claudiu.balogh](https://forums.developer.nvidia.com/u/claudiu.balogh)\ **Replies:** 2\ **Last updated:** [February 25, 2026, 8:06am UTC](https://forums.developer.nvidia.com/t/registering-a-numa-allocated-host-buffer-with-the-gpu-dma-engine-for-peak-transfer-performance-in-opencl/361614 "2026-02-25T08:06:24Z")
OS: OpenSuse Leap 15.5 GPU: RTX Pro 5000 Blackwell driver version: 580.76.05 I am trying to register a large numa-allocated and pinned host buffer into an OpenCL context for peak transfer performance in smaller chunks… --- ## [clEnqueueNDRangeKernel call takes too much time on Nvidia GPUs](https://forums.developer.nvidia.com/t/clenqueuendrangekernel-call-takes-too-much-time-on-nvidia-gpus/337286)
**Author:** [@vdualb](https://forums.developer.nvidia.com/u/vdualb)\ **Replies:** 4\ **Last updated:** [January 10, 2026, 8:52am UTC](https://forums.developer.nvidia.com/t/clenqueuendrangekernel-call-takes-too-much-time-on-nvidia-gpus/337286 "2026-01-10T08:52:43Z")
Hello. I use OpenCL in my program to build finite element method SLAE. Algorithm is implemented using one OpenCL kernel. The problem is when running this code on Nvidia GPUs clEnqueueNDRangeKernel function call takes way… --- ## [Inconsistent OpenCL headers in cuda 12.9](https://forums.developer.nvidia.com/t/inconsistent-opencl-headers-in-cuda-12-9/333521)
**Author:** [@caleb.huck](https://forums.developer.nvidia.com/u/caleb.huck)\ **Replies:** 0\ **Last updated:** [May 16, 2025, 4:30pm UTC](https://forums.developer.nvidia.com/t/inconsistent-opencl-headers-in-cuda-12-9/333521 "2025-05-16T16:30:41Z")
In the newest version of the OpenCL headers, the symbol CL\_MUTABLE\_DISPATCH\_PROPERTIES\_ARRAY\_KHR was renamed to CL\_MUTABLE\_COMMAND\_PROPERTIES\_ARRAY\_KHR (defined in cl\_ext.h). But the cl.hpp included in the cuda toolkit i… --- ## [OpenCL Executable will not run in Runtime Container](https://forums.developer.nvidia.com/t/opencl-executable-will-not-run-in-runtime-container/311244)
**Author:** [@Antonio\_Ferreira](https://forums.developer.nvidia.com/u/Antonio_Ferreira)\ **Replies:** 0\ **Last updated:** [October 27, 2024, 7:57pm UTC](https://forums.developer.nvidia.com/t/opencl-executable-will-not-run-in-runtime-container/311244 "2024-10-27T19:57:33Z")
I am trying to build a container to run AutoDock-Vina-GPU-2.1. I can compile and run the code inside the HPC SDK container (docker run -it nvcr.io/nvidia/nvhpc:23.1-devel-cuda12.0-ubuntu22.04 or docker build -t test:vin… --- ## [OpenCL C for RTX 3060](https://forums.developer.nvidia.com/t/opencl-c-for-rtx-3060/303930)
**Author:** [@TrailingStop](https://forums.developer.nvidia.com/u/TrailingStop)\ **Replies:** 1\ **Last updated:** [August 20, 2024, 1:54pm UTC](https://forums.developer.nvidia.com/t/opencl-c-for-rtx-3060/303930 "2024-08-20T13:54:29Z")
Hi, not sure if this is the correct forum, but I have not found one that seems to fit better. I converting a cuda kernel into OpenCl. I use for developing a RTX 3060. When I run ‘clinfo’ I get these information (beside… --- ## [Backward Compatibility Issues with CUDA on Older Nvidia Architectures vs. OpenCL](https://forums.developer.nvidia.com/t/backward-compatibility-issues-with-cuda-on-older-nvidia-architectures-vs-opencl/295500)
**Author:** [@Montazeran8](https://forums.developer.nvidia.com/u/Montazeran8)\ **Replies:** 1\ **Last updated:** [June 7, 2024, 6:53pm UTC](https://forums.developer.nvidia.com/t/backward-compatibility-issues-with-cuda-on-older-nvidia-architectures-vs-opencl/295500 "2024-06-07T18:53:37Z")
Hello, I looked through the forum and could not find the answer to my question, so I created this topic. My question is about backward compatibility between CUDA and OpenCL. I am forced to use a very old Nvidia graphic… --- ## [Missing libnvidia-opencl.so in the docker](https://forums.developer.nvidia.com/t/missing-libnvidia-opencl-so-in-the-docker/280391)
**Author:** [@943359205](https://forums.developer.nvidia.com/u/943359205)\ **Replies:** 1\ **Last updated:** [May 20, 2024, 6:46pm UTC](https://forums.developer.nvidia.com/t/missing-libnvidia-opencl-so-in-the-docker/280391 "2024-05-20T18:46:27Z")
I pull the nvidia/cuda:12.3.1-devel-ubuntu22.04 image. But I found there isn’t libnvidia-opencl.so. I don’t know how to deal with it. I need it to use the opencl. my gpu is 4070ti and nvidia-smi info is below Sun J… --- ## [OpenCL segfault on call to clCreateFromGLRenderbuffer](https://forums.developer.nvidia.com/t/opencl-segfault-on-call-to-clcreatefromglrenderbuffer/290331)
**Author:** [@wellsilver](https://forums.developer.nvidia.com/u/wellsilver)\ **Replies:** 1\ **Last updated:** [April 21, 2024, 6:49pm UTC](https://forums.developer.nvidia.com/t/opencl-segfault-on-call-to-clcreatefromglrenderbuffer/290331 "2024-04-21T18:49:36Z")
Hi, I’m experiencing an issue with using cl\_khr\_gl\_sharing where it segfaults on only my gpu, and works completely fine on my integrated graphics I compiled it with msys64 (mingw64) on Windows 11, using the Khronos Open… --- ## [OpenCL support on Jetson Orin nano devKit](https://forums.developer.nvidia.com/t/opencl-support-on-jetson-orin-nano-devkit/283586)
**Author:** [@edu.app\_devnv](https://forums.developer.nvidia.com/u/edu.app_devnv)\ **Replies:** 1\ **Last updated:** [February 22, 2024, 1:10am UTC](https://forums.developer.nvidia.com/t/opencl-support-on-jetson-orin-nano-devkit/283586 "2024-02-22T01:10:20Z")
Does Jetson Orin nano devKit support OpenCL? If not, what are the options? --- ## [To be or not to be, Audio or OpenCL](https://forums.developer.nvidia.com/t/to-be-or-not-to-be-audio-or-opencl/273003)
**Author:** [@marcsitkin](https://forums.developer.nvidia.com/u/marcsitkin)\ **Replies:** 1\ **Last updated:** [November 16, 2023, 10:16am UTC](https://forums.developer.nvidia.com/t/to-be-or-not-to-be-audio-or-opencl/273003 "2023-11-16T10:16:42Z")
Using MX Linux as an OS, after recent updates audio is not enabled on cold boot, opencl is functional. On reboot, audio is enabled, opencl is not functional. Geforce 1060 card, driver 595 series. Have had to go back t… --- ## [Computing 'local-box' statistics for 3D images](https://forums.developer.nvidia.com/t/computing-local-box-statistics-for-3d-images/269003)
**Author:** [@Eyal.Rozenberg](https://forums.developer.nvidia.com/u/Eyal.Rozenberg)\ **Replies:** 2\ **Last updated:** [October 10, 2023, 3:39pm UTC](https://forums.developer.nvidia.com/t/computing-local-box-statistics-for-3d-images/269003 "2023-10-10T15:39:02Z")
Suppose I have raw 3D grayscale “image” (i.e. all voxels in memory consecutively), and I want to compute, for each voxel, some statistics regarding a fixed-size cube/box surrounding that voxel. min, max, average, median,… --- ## [Aarch64 centos8 and tesla t4 build opencl program failed](https://forums.developer.nvidia.com/t/aarch64-centos8-and-tesla-t4-build-opencl-program-failed/235007)
**Author:** [@dragon\_tygg](https://forums.developer.nvidia.com/u/dragon_tygg)\ **Replies:** 3\ **Last updated:** [September 7, 2023, 3:17am UTC](https://forums.developer.nvidia.com/t/aarch64-centos8-and-tesla-t4-build-opencl-program-failed/235007 "2023-09-07T03:17:06Z")
When I use the NVIDIA TESLA T4 series graphics card on the CENTOS8 system of the aarch64 system to accelerate OPENCL computing. When compiling OPENCL kernel program (clBuildProgram), it always returns - 11 (CL\_BUILD\_PROG… --- ## [OpenCL getInfo() not compiling with some options](https://forums.developer.nvidia.com/t/opencl-getinfo-not-compiling-with-some-options/264473)
**Author:** [@rainer.bitschi](https://forums.developer.nvidia.com/u/rainer.bitschi)\ **Replies:** 2\ **Last updated:** [August 27, 2023, 7:00am UTC](https://forums.developer.nvidia.com/t/opencl-getinfo-not-compiling-with-some-options/264473 "2023-08-27T07:00:51Z")
I am new to OpenCL, by now I am running the following code: int main() { std::vector\ platforms; std::vector\ devices; cl::Platform::get(&platforms); platforms\[0\].getDevices(CL\_DEVICE\_TYPE\_AL… --- ## [/dev/nvidia-uvm IO error on Ubuntu 22.04, 520 to 535 driver versions](https://forums.developer.nvidia.com/t/dev-nvidia-uvm-io-error-on-ubuntu-22-04-520-to-535-driver-versions/262153)
**Author:** [@FangQ](https://forums.developer.nvidia.com/u/FangQ)\ **Replies:** 2\ **Last updated:** [August 27, 2023, 2:28am UTC](https://forums.developer.nvidia.com/t/dev-nvidia-uvm-io-error-on-ubuntu-22-04-520-to-535-driver-versions/262153 "2023-08-27T02:28:02Z")
This is the same observation as I posted in another thread on disappearing OpenCL icd on Ubuntu 22.04 after some extended use on a Ubuntu 22.04 box, I noticed that CUDA programs also fails to find NVIDIA GPUs when this … --- ## [Kernel error -45 while running clinfo](https://forums.developer.nvidia.com/t/kernel-error-45-while-running-clinfo/262871)
**Author:** [@sanchit.grover](https://forums.developer.nvidia.com/u/sanchit.grover)\ **Replies:** 0\ **Last updated:** [August 11, 2023, 9:11am UTC](https://forums.developer.nvidia.com/t/kernel-error-45-while-running-clinfo/262871 "2023-08-11T09:11:55Z")
I recently upgraded the NVIDIA driver from 515 to 535 (tried 525 as well) and noticed the bug while running clinfo and one of our applications using the new CUDA 12. clinfo and the application are run inside a docker con… --- ## [CLBlast could not produce FP16 tuning results with NVIDIA GPUs like RTX4090 on Windows](https://forums.developer.nvidia.com/t/clblast-could-not-produce-fp16-tuning-results-with-nvidia-gpus-like-rtx4090-on-windows/255751)
**Author:** [@tjcgy](https://forums.developer.nvidia.com/u/tjcgy)\ **Replies:** 3\ **Last updated:** [June 8, 2023, 6:50am UTC](https://forums.developer.nvidia.com/t/clblast-could-not-produce-fp16-tuning-results-with-nvidia-gpus-like-rtx4090-on-windows/255751 "2023-06-08T06:50:00Z")
Dear Nvidia experts, I would like to report a potential problem on the OpenCL Windows driver on many NVIDIA GPU. Recently, my students and I are trying to tune an OpenCL lib called CLBlast to help the community to spe… --- ## [Trouble accessing OpenCL](https://forums.developer.nvidia.com/t/trouble-accessing-opencl/253931)
**Author:** [@adam.sys](https://forums.developer.nvidia.com/u/adam.sys)\ **Replies:** 2\ **Last updated:** [May 21, 2023, 2:50pm UTC](https://forums.developer.nvidia.com/t/trouble-accessing-opencl/253931 "2023-05-21T14:50:08Z")
I am exploring Dr. Moritz Lehmann’s open source FluidX3D code. It fails to compile on my machine.He provide a pre-compiled benchmark that fails to run for me. (Basic Tutorial - error when debug initiates · Issue #80 · Pr… --- ## [Regression? NVIDIA OpenCL ICD stops working in Ubuntu 22.04](https://forums.developer.nvidia.com/t/regression-nvidia-opencl-icd-stops-working-in-ubuntu-22-04/249076)
**Author:** [@FangQ](https://forums.developer.nvidia.com/u/FangQ)\ **Replies:** 2\ **Last updated:** [April 19, 2023, 7:57am UTC](https://forums.developer.nvidia.com/t/regression-nvidia-opencl-icd-stops-working-in-ubuntu-22-04/249076 "2023-04-19T07:57:25Z")
I have two Ubuntu boxes running 22.04 and neither of the systems can list the nvidia GPU in clinfo. These boxes have Box 1: RTX 2060 CUDA 11.3 driver 515.86.01 Box 2: RTX 3090 CUDA 11.3 driver 525.105.17 both bo… --- ## [CL\_INVALID\_OPERATION on clEnqueueAcquireGLObjects](https://forums.developer.nvidia.com/t/cl-invalid-operation-on-clenqueueacquireglobjects/250417)
**Author:** [@dr.elichan1](https://forums.developer.nvidia.com/u/dr.elichan1)\ **Replies:** 0\ **Last updated:** [April 19, 2023, 1:26am UTC](https://forums.developer.nvidia.com/t/cl-invalid-operation-on-clenqueueacquireglobjects/250417 "2023-04-19T01:26:49Z")
I downloaded the “OpenCL Marching Cubes Isosurfaces” example from “Open Computing Language OpenCL | NVIDIA Developer” and managed to compile it. However, it executes with the following error: !!! Error # -59 (CL\_INVALID… --- ## [Opencl build failed in aarch64 platform](https://forums.developer.nvidia.com/t/opencl-build-failed-in-aarch64-platform/249726)
**Author:** [@nukuws](https://forums.developer.nvidia.com/u/nukuws)\ **Replies:** 1\ **Last updated:** [April 13, 2023, 1:55am UTC](https://forums.developer.nvidia.com/t/opencl-build-failed-in-aarch64-platform/249726 "2023-04-13T01:55:49Z")
I want like to know if the nvidia implementation of opencl supports linxux/aarch64 platforms, I read here that linux only supports x86/64, is that right? When I run tests with pyopencl, I found that the same program wor… --- ## [Wrong return value for clCompileProgram](https://forums.developer.nvidia.com/t/wrong-return-value-for-clcompileprogram/247873)
**Author:** [@alvaro\_sch](https://forums.developer.nvidia.com/u/alvaro_sch)\ **Replies:** 1\ **Last updated:** [March 29, 2023, 2:18pm UTC](https://forums.developer.nvidia.com/t/wrong-return-value-for-clcompileprogram/247873 "2023-03-29T14:18:31Z")
I’m running on Arch Linux with opencl-nvidia 530.41.03-1 drivers, I got CL\_BUILD\_PROGRAM\_FAILURE instead of the documented CL\_COMPILE\_PROGRAM\_FAILURE in the latest version of the OpenCL spec. Here’s a minimal code snipp… --- ## [OpenCL can call tensor core?](https://forums.developer.nvidia.com/t/opencl-can-call-tensor-core/246596)
**Author:** [@user9725](https://forums.developer.nvidia.com/u/user9725)\ **Replies:** 1\ **Last updated:** [March 19, 2023, 8:44pm UTC](https://forums.developer.nvidia.com/t/opencl-can-call-tensor-core/246596 "2023-03-19T20:44:17Z")
As the title ,OpenCL can call tensor core ? If OpenCL can call tensor core ,need install extension library? --- ## [How to compile OpenCL code into binary for a GPU I do not physically have?](https://forums.developer.nvidia.com/t/how-to-compile-opencl-code-into-binary-for-a-gpu-i-do-not-physically-have/241711)
**Author:** [@AlexJoe](https://forums.developer.nvidia.com/u/AlexJoe)\ **Replies:** 9\ **Last updated:** [March 4, 2023, 6:55pm UTC](https://forums.developer.nvidia.com/t/how-to-compile-opencl-code-into-binary-for-a-gpu-i-do-not-physically-have/241711 "2023-03-04T18:55:35Z")
We are a small company developing image processing software. We use OpenCL for GPU acceleration (single code base for Nvidia, AMD and Intel GPUs), and we deploy our OpenCL kernels as compiled binary, one per each line of… [Next page](https://forums.developer.nvidia.com/tag/opencl/374.md?match_all_tags=true&page=1&tags%5B%5D=opencl)