orklev2-ci

What problem does it solve?

This Skill helps engineers understand and apply Orkid's ComputeInterface to correctly bind resources, dispatch compute shaders, synchronize passes, and integrate PyTorch tensors with GPU storage buffers so compute workloads run correctly and efficiently.

Core Features & Use Cases

• API Reference Mapping: Points to CI interface declarations and implementations (ci.h, shadman.h, vulkan_ctx.h, pyext_gfx.cpp) for precise method semantics.
• Dispatch & Synchronization Guidance: Explains beginDispatchPhase/endDispatchPhase, dispatchCompute/dispatchComputeIndirect, storageBarrier usage, and typical two-pass compute patterns.
• Resource Binding & PyTorch Support: Covers bindStorageBuffer, bindImage, bindSampler usage and ENABLE_PYTORCH helpers that wrap tensors as SSBOs for zero-copy or explicit copy workflows.
• Use Case: Diagnose a multi-pass compute blur that writes to SSBOs and image targets, ensure correct memory barriers and transitions, and optionally stream PyTorch tensor data into GPU buffers.

Quick Start

Ask how to bind storage buffers and images, issue compute dispatches, and use storageBarrier between passes in Orkid's ComputeInterface.