Ben Vanik
Ben Vanik
A crash is very different than an assert - we shouldn't conflate the two. Crashes are bad memory accesses and such. Asserts are unimplemented functionality. On Mon, Aug 31, 2015...
May need to handle Windows weirdness with cc_binary/cc_test executables having .exe extensions: bazelbuild/bazel#3418
Pull #81 gets most of this done. Swiftshader has issues with multithreading (like, it has no thread safety at all >_>) so a mode that disables the present queue and...
Gonna fake it w/ MemoryPool for now.
Correct - the runtime still sees push constants as a way to build the argument buffers and doesn't know about the hidden runtime buffer. Incidentally this is how we'd add...
gather 1x1000xi64 is odd - does it actually need to be 64-bit? (that's very rare) of course it'd be good to support as-is but if we want to flush out...
Is this issue still valid? Sounds like they're the same now?
Make sure your backend adds the legacy_sync attr: https://github.com/google/iree/blob/main/compiler/src/iree/compiler/Dialect/HAL/Target/VulkanSPIRV/VulkanSPIRVTarget.cpp#L129-L131
there's no known issues with barriers within command buffers - vulkan and cpu both take advantage of them - so I'd check your usage of the API or try to...
That's for the inside of command buffers only and only execution barriers - the device queue operations are where fences/semaphores are - ensure your device.c file doesn't have "// TODO:...