Interactive Demos

Usage

Open any demo as a full page, or read the chapters where the relevant demos are embedded inline. The local HTML files use the enabled Embed HTML plugin and do not require a web server.

• Async Coordination: Barriers (barrier_intro.html)
• A 2-CTA Cluster (cta_cluster.html)
• MBarrier Arrive Timeline (mbarrier_arrive_timeline.html)
• mbarrier State and Operations (mbarrier_mechanism.html)
• MBarrier tcgen05.commit Timeline (mbarrier_tcgen05_timeline.html)
• mbarrier TMA Timeline (mbarrier_tma_timeline.html)
• Phase Tracking (phase_tracking.html)
• GEMM Data Pipeline on Blackwell (pipeline_arch.html)
• Scale Factors in TMEM (sf_tmem.html)
• Blackwell SM Architecture (sm_architecture.html)
• Swizzle Layout — SWIZZLE_128B (swizzle_128B.html)
• Swizzle Layout (swizzle_8x8.html)
• Swizzle Atoms — All Formats (swizzle_atom_general.html)
• Tensor Core (tcgen05_intro.html)
• Blackwell Thread Hierarchy (thread_hierarchy.html)
• Thread + Register Layout (thread_register.html)
• Distributed Layout (tile_distributed.html)
• Tiled Layout (tiled_layout.html)
• Swizzle Imposes Tiling Constraint (tiling_constraint.html)
• TIRx: Scope, Layout, Dispatch (tirx_dispatch.html)
• Tiling and Swizzling with 3D TMA (tma_3d.html)
• TMA: Tensor Memory Accelerator (tma_intro.html)
• TIRx Tensor Layout — Interactive Demo (tirx-layout-demo/index.html)