execute()

output_type MayaFlux::Yantra::TextureExecutionContext::execute ( const input_type &  input, const ExecutionContext &  ctx  )

inlineoverridevirtual

Override for CHAINED mode: skips collect_gpu_outputs().

Callers driving multi-pass image dispatch (e.g. ConnectedComponents merge loop) may flip an image slot to INPUT_OUTPUT for the cross-pass barrier in dispatch_batched, which leaves no OUTPUT-direction slot for collect_gpu_outputs' output_slot() lookup to find. Non-CHAINED modes are unaffected and fall through to the base implementation.

Parameters

input	Ignored for CHAINED mode (image already staged manually).
ctx	ExecutionContext; CHAINED mode requires pass_count and pc_updater.

Reimplemented from MayaFlux::Yantra::GpuExecutionContext< std::vector< Kakshya::DataVariant >, std::vector< Kakshya::DataVariant > >.

Definition at line 262 of file TextureExecutionContext.hpp.

    {
        if (ctx.mode != ExecutionMode::CHAINED && ctx.mode != ExecutionMode::CHAINED_INDIRECT)
            return Base::execute(input, ctx);
 
        if (!ensure_gpu_ready()) {
            error<std::runtime_error>(
                Journal::Component::Yantra,
                Journal::Context::BufferProcessing,
                std::source_location::current(),
                "TextureExecutionContext: GPU initialisation failed");
        }
 
        auto [ch_copies, structure_info] = extract_inputs(input);
 
        if (ctx.mode == ExecutionMode::CHAINED_INDIRECT) {
            dispatch_core_chained_indirect(ch_copies, structure_info, ctx);
        } else {
            dispatch_core_chained(ch_copies, structure_info, ctx);
        }
 
        return output_type {};
    }

References input, and MayaFlux::Yantra::ExecutionContext::mode.