Richard Winterton

This is a workaround and not a solution but we could use an audio bridge for those that do not present and are not concerned about video? I think I...

True this is a problem since we don't have the AVX3 ISA but this is fixed in ICL. Some cpu's can't take advantage of this moving forward. Jan you may...

In order to take advantage of AVX3 we would have to detect the platforms capable of doing the AVX3 VL instructions. There are several that we should be able to...

I would agree with the capability to do shuffles. The pshufb is very useful in ARGB and YUV conversions as well as matrix manipulation. If I understand the proposal the...

Wonder if this instruction could be similar to what we talked about with the instrumentation instruction. Maybe specialty instructions. Not sure we want to group them together but effectively is...

I can let frank reply too. As you said one or two instruction move in the prefetch may not matter similar to the ITT proposal. We just wouldn't want it...

I think what Thomas is concerned with his when you put a prefetch instruction it may skid past where you may think it will be in the machine code? As...

Thomas is right that it is an idea of promoting existing SIMD instructions to a larger vector length. The idea is to use the same opcode for the larger vector...

There are other concerns to about validation of the wasm if cached. Perhaps this isn't relevant but a strict validation of the wasm code is going to require more than...