#Half rate fp64 gpu driver

Avoid just in time compilation of PSOs.Avoid compiling PSOs on threads with small stacks or in time sensitive code.Compiling PSOs is a great task for multi-threading.

#Half rate fp64 gpu driver

The driver will not spawn extra threads to compile.

This also allows easier state optimizations since there is no longer a need to track multiple small states. This removes any stutters that could happen in the previous API generations when the driver had to recompile shaders at the time of the draw call due to a change in state. This allows the driver to know everything up front so that it can compile the shaders into the correct assembly.

If they are used have at least 10 draws per bundle/secondary command buffer.Īll the smaller state structures where combined into a single state known as a Pipeline State Object.

Just fill the command buffers every frame.

They’re likely to hurt GPU performance (they only benefit CPU performance).

Avoid using bundles and secondary command buffers.

Ideally, submissions would only happen when syncing queues or at the end of a frame.

Try to batch command buffers together into a single submission to reduce overhead.

Each submit has a CPU and GPU cost associated with it.

Minimize the number of command buffer submissions to the GPU.

Try to reuse the same allocators for the same rendering workload to help minimize memory usage.

Command allocators grow in line with the largest command buffer allocated from them.

The number of command allocators should be greater than (or equal to) the number of threads recording commands multiplied by the number of frames in flight.

It is recommended to have a command allocator for each thread recording commands per frame.

Command allocators are not thread safe.

A good rule of thumb is to have at least 10 draws/dispatches.

Avoid having too many small command buffers.

Applications are responsible for multi-threading.

The driver does not spawn extra threads for low level APIs.

Using this can get better CPU utilization than higher-level APIs.

Low level APIs allow multi-threading of command buffer generation.

With the previous generation of APIs, the amount of multi-threading that could be done was severely limited to what the driver could muster up. One of the biggest optimizations is that an application can now multi-thread command buffer recording. Most of the CPU time spent in DirectX®12 (DX12) and Vulkan® will be spent recording draws into the command buffers.

Command buffers are the heart of the low-level graphics APIs.