Mike Apodaca

Results 9 comments of Mike Apodaca

Yes, you're correct that this area needed more work. We have a specification update coming soon that introduces the concept of “command queue groups”, which is similar to Vulkan’s [VkQueueFamilyProperties](https://www.khronos.org/registry/vulkan/specs/1.2/html/chap4.html#VkQueueFamilyProperties)....

Here is the latest documentation: + [Programming Guide: Command Queues](https://spec.oneapi.com/level-zero/latest/core/PROG.html#command-queues) + [zeDeviceGetCommandQueueGroupProperties](https://spec.oneapi.com/level-zero/latest/core/api.html#zedevicegetcommandqueuegroupproperties) + [ze_command_queue_group_properties_t](https://spec.oneapi.com/level-zero/latest/core/api.html#ze-command-queue-group-properties-t) + [ze_command_queue_desc_t](https://spec.oneapi.com/level-zero/latest/core/api.html#ze-command-queue-desc-t) + [ze_command_list_desc_t](https://spec.oneapi.com/level-zero/latest/core/api.html#ze-command-list-desc-t)

We had early drafts of the spec that did include the concept of timeline semaphores, but we have decided to pursue a different solution. We’re not quite ready to provide...

I think we forgot another negation in there. Thanks for pointing this out; we will fix it to clearly say we intend to maintain backward compatibility.

Unfortunately, this didn't make it into the most recent 0.95 spec. But this entire paragraph will be rewritten in the next release. The intent was to say: by not allowing...

Thanks for the feedback. Given the number of permutations of RGB formats, in particular, it was our intent to avoid the huge enumeration and instead define a programmable format interface,...

Good find. There are a couple of directions we could consider: 1. SetKernelArgs on the current active context - need implicit locks to be thread-safe 2. SetKernelArgs on the command...

> One thing I feel is quite implicit is the hidden mapping between input arguments and what ends up in essentially a constant buffer that is built inside the driver....

i have another use case for this. we are using google benchmark for GPU-based microbenchmarks. very often a specific GPU device does not support a feature we are writing a...