One man's trash is another man's... #OpenCL test setup! 🖖😋
Got new old dumpster hardware from #gpn24: #Intel i7-4771 Haswell #CPU (once retailed for $320!), Gigabyte Z87 mainboard, 2x 4GB DDR3 (🤑!), #AMD Radeon HD7790 2GB.

Added 2x 8GB RAM & fire-hazard Molex-PCIe adaptor for good measure.

Interesting: When was the last time you saw a budget #GPU with full PCIe x16, clamshell memory (chips on front+back of PCB), and muli-GPU connector? Today's GPUs seem much more cost-optimized in comparison.

Running the #OpenCL CTS on two machines right now for CL 3.1 conformance submissions...

I let you all guess which drivers I'm running against on rusticl here.

The presentations and video recordings from IWOCL 2026 are now available on the conference program page.

IWOCL 2026 brought together researchers and practitioners advancing OpenCL, SYCL, and heterogeneous computing. This year's program covered portability across GPU vendors, AI and LLM inference on OpenCL backends, safety-critical GPU programming, and much more.

Browse the full program, slides, and recordings: https://www.iwocl.org/iwocl-2026/conference-program/
#OpenCL #SYCL #HeterogeneousComputing #IWOCL #GPU

I published my notes on #OpenCL #OpenGL interop

https://mathr.co.uk/web/opencl-opengl-interop.html

so far focus is on writing to OpenGL textures from OpenCL kernels.

in my realworld test case this gives an overall improvement in total* runtime of about 10% vs downloading from OpenCL to CPU and reuploading to OpenGL

(* I did not subtract the cost of everything else the test case was doing (loading settings, running opencl kernels + opengl shaders, saving images), a microbenchmark of just the data transfer changes would show a much bigger improvement)

ICYMI: OCCT - 16.1.10 Update Notes

#AMD #Intel #Linux #LinuxGaming #NVIDIA #OCCT #OpenCL #PCGaming #Radeon #Vulkan

https://store.steampowered.com/news/app/3515100/view/528753960533622839

#FluidX3D #CFD v3.7 brings faster Q-criterion isosurface rendering with #OpenCL local memory optimization! 🖖🤠
https://github.com/ProjectPhysX/FluidX3D/releases/tag/v3.7

Instead of 32 velocities for each #GPU thread, now an 8x8x8 workgroup loads & reuses 11x11x11 velocities in L1$, a 12x VRAM BW reduction.

Fascinating insight: Which thread loads which cell from VRAM to L1$, and which thread renders which grid cell within the workgroup, can be very different!
https://github.com/ProjectPhysX/FluidX3D/blob/master/src/kernel.cpp#L2827-L2956

PS: plugged X-wing Gif in #GitHub preview 🖖😜

OpenCL 3.1 is here.

The Khronos Group has moved several capabilities into the core spec, including SPIR-V kernels, subgroups, and integer dot products.

Also includes improvements to the memory model and synchronization, plus better alignment with Vulkan via device UUID queries.

Implementations are already underway across major vendors and open source projects.

- Full Blog: https://www.khronos.org/blog/opencl-3.1-is-here?utm_medium=social&utm_source=mastodon&utm_campaign=OpenCL_3.1_is_here&utm_content=blog
- OpenCL specification GitHub
- Khronos Discord

#OpenCL #HPC #GPU #Compute #SPIRV