Ben Ashbaugh

@FelixCLC_ @illuhad @ProjectPhysX PVC should "just work" with >4GB allocations, no special steps required. If it doesn't, please file a bug or otherwise let us know - thanks!

@bashbaug @illuhad @ProjectPhysX Following up, does this have ramifications for PVC? Colleague of mine ran into some bugs around different runtimes. Some on client parts (Skylake iGPU on xeon) having different behavior than ARC alchemist, having different behavior than CPU, which was different than PVC

Spec/language lawyers: Reading this line (#CL_DEVICE_MAX_MEM_ALLOC_SIZE" target="_blank" rel="nofollow noopener">registry.khronos.org/OpenCL/specs/3…), do you think: A: that the minimum allocation of the max mem allocation flag is 1:4? B: that the max allocation must be 1:4 of device memory in other words, given a 16GB card, is the maximum: #openCL #GPGPU #HPC

@FelixCLC_ @illuhad @ProjectPhysX Note, the maximum allocation size is the same for both our OpenCL and Level Zero drivers. The reason why >4GB allocations work through the DPC++ runtime is because it explicitly relaxes the allocation limits.

@illuhad @ProjectPhysX @bashbaug ICPX/DPCPP can and will accept single allocations >4GB, I've tested it myself. AFAIK DPCPP goes through level0 on it's way to dGPUs.

@ProjectPhysX @illuhad @FelixCLC_ Yeah, you would also need to tell the compiler allocations may be >4GB. Have a look at: #relaxallocationlimits-bool" target="_blank" rel="nofollow noopener">github.com/intel/opencl-i…

@illuhad @FelixCLC_ Our current client GPUs report a 4GB limit because it lets us use 32-bit addressing arithmetic more aggressively, which improves performance. It's not something that's required by the spec.

@illuhad @FelixCLC_ This matches my understanding as well. A 16GB card would need to support at least 1GB allocations, and could report up to 16GB (no upper limit). Only less-than-4GB cards could have a less-than-1GB allocation limit.

The call for submissions for the 11th International Workshop on #OpenCL and #SYCL closes on Friday Jan 20, 2023. Papers, tech talks and posters are all welcome. Join us in Cambridge, UK on April 18-20. iwocl.org @openclapi @thekhronosgroup @SYCLstd

@ProjectPhysX @FluidX3D @IntelGraphics I've filed an internal issue regarding the incorrect queries. Should be an easy fix. Feel free to file an issue on the compute-runtime GitHub also if you'd like. Thanks for raising these issues!

@ProjectPhysX @psychocoderHPC @FelixCLC_ @SquashBionic @TheMalcore @AMD Following up: Linux should be much better with the latest drivers. I see 15+GB reported on my 16GB GPU. Windows still needs some work though and I don't have an ETA at this time.

Ok, this hadn't come across my feed for some reason, but Intel Arch Alchemist White paper is OUT!! Find it here: intel.com/content/www/us… @SquashBionic @TheMalcore @ProjectPhysX #silicongang

I missed seeing everyone at the @thekhronosgroup face-to-face meeting this week but I guess attending remotely had some benefits? #opencl #sycl #kittiesofkhronos

Cool, this is my alma mater. Go Tritons! #oneAPI #USCD

If you need help debugging and tracing your command buffer applications, the OpenCL Intercept Layer has full support for both the base command buffer extension and the mutable dispatch extension. github.com/intel/opencl-i… #OpenCL

I also recently added command buffer emulation layer support for out-of-order command buffers for OpenCL implementations that support out-of-order queues. The same repo has sample applications that demonstrate how to use both new features. #OpenCL

Want to experiment with the new command buffer mutable dispatch extension right away? My command buffer emulation layer now supports this extension and should work with most OpenCL implementations. github.com/bashbaug/Simpl… #OpenCL

#OpenCL @openclapi 3.0.12 Published With Command Buffers Mutable Dispatch Extension phoronix.com/news/OpenCL-3.…

@stillpixel Haha, yeah. I was waiting for them to knock over my water and turn off my power strip just to prove me wrong!

I’ve been back at work for two weeks and I’m happy to report the new office assistants are working out great.

There's room for better GL/VK perf (benchmarks next week), but as a developer card, the A380 at $139 is a bargain... @openclapi 3.0 on open-source stack, @IntelSoftware oneAPI ecosystem for testing and porting work is great, SYCL & rest of software ecosystem around it.

Managed to squeeze in one last sabbatical hike before heading back to work next week. #desolationwilderness

@stillpixel We went down the coast so sadly we were too far west. Next time (if it’s still there)!

Epic road trip complete! 5000 miles, 34 nights (32 in a tent), 9 national parks, one flat tire 😥, countless memories. Pic from yesterday at Crater Lake.