Ben Ashbaugh

Hi @airlied! Thanks for the question. You're right that for **device** memory allocations, the pointer must be in the same virtual address space as the host application, but it does...

Thanks for the excellent report! I'm sorry this isn't working for you. Your device should support time based sampling. A few initial things to check: 1. Have you tried rebooting...

Hello, thank you again for your excellent description! I think I know what's happening, and in short you found a very reliable way to get into the "stuck" state I...

I've filed a public issue against the metrics library to track this issue: https://github.com/intel/metrics-discovery/issues/18

Yeah, although we try to maintain consistent output for API tracing / call logging, it's all very ad-hoc right now. The "real" way to fix this IMHO is to create...

Buffer hashes (option 1) should be fairly straightforward. I'll look into adding this, unless somebody else gets to it first. Buffer compression (option 2) sounds interesting but I have far...

Thinking about this a bit more. Hashing the memory objects is a good first step regardless, since this is effectively what we'll need to do for binary equality. Three interesting...

Just to be clear, the quick hack mentioned above is to change this: ```c #if defined( __MINGW64__ ) #include #else #include #endif ``` to this? ```c #include ``` If we...

A bit more info: the prefix and suffix defines are mostly used to ensure proper deprecation messages are emitted when a deprecated API gets used. The deprecation messages can be...

Yes, I agree that attaching prefix and suffix defines indicating when the API was **last supported** is most consistent with the current usage. Admittedly this is a little confusing, since...