AMD's ROCm HRX Loom v0.1 Released For Targeting AMD CPUs, GPUs & NPUs

At this year's AMD Advancing AI event Hyperloom was announced among the ROCm improvements to expect. Hyperloom aims for agentic workload optimizations and more with quote powerful abilities if everything works out right. As part of the latest round of ROCm advancements, the Loom SDK v0.1 release was released today. Loom is part of the ROCm HRX (HIP Runtime Extended) as a specialized compiler and IR stack. The Loom documentation describes itself as: "Loom compiles kernels and asynchronous device programs, links model programs, and specializes them for execution on CPUs, AMD GPUs, and AMD NPUs. It carries source facts, target facts, configuration, correctness cases, and compiler reports through the same authoring and execution path. The hrx-loom SDK packages the compiler, runtime, Python APIs, and command-line tools. loom-fetch installs a model's selected programs and assets; loom-serve loads the resulting card and serves requests over HTTP." There is the Loom programming guide for those wanting to learn more about making use of it. Loom supports targeting AMDGPU HSA emission for GPU execution, SPIR-V/Vulkan for portable execution, and x86 and WebAssembly support that is still maturing. As the first step, tagged overnight was the Loom v0.1 release as its very first tagged release after being in development as an open-source project as part of the ROCm HRX system. Those wishing to experiment with the Loom stack can find the Loom SDK v0.1 release via GitHub. It can also be installed via Python's pip as the hrx-loom package. This is certainly something to watch moving forward for AMD's software stack.