Multi-GPU programming traditionally requires developers to navigate complex trade-offs between performance and programmability. High-performance implementations typically rely on low-level HIP\/CUDA communication libraries that demand substantial engineering effort for even basic overlap patterns, while simpler abstractions often sacrifice performance. We present Iris, a multi-GPU communication library implemented entirely in Python and Triton that eliminates this trade-off. Iris provides tile-based symmetric memory abstractions that naturally align with Triton’s programming model, enabling developers to write single-source kernels that seamlessly interleave computation and communication. We demonstrate a taxonomy of compute-communication overlap patterns–from bulk-synchronous to fine-grained workgroup specialization–that can be implemented with minimal code changes in Iris, often requiring just a few additional lines within the same Triton kernel. Our evaluation shows that Iris achieves near-optimal bandwidth utilization in microbenchmarks and delivers up to 1.79x speedup over PyTorch and RCCL for GEMM+All-Scatter workloads, demonstrating that high-level implementations can match or exceed heavily-optimized libraries while dramatically simplifying multi-GPU programming.<\/p>\n","protected":false},"excerpt":{"rendered":"
Multi-GPU programming traditionally requires developers to navigate complex trade-offs between performance and programmability. High-performance implementations typically rely on low-level HIP\/CUDA communication libraries that demand substantial engineering effort for even basic overlap patterns, while simpler abstractions often sacrifice performance. We present Iris, a multi-GPU communication library implemented entirely in Python and Triton that eliminates this trade-off. […]<\/p>\n","protected":false},"author":351,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_jetpack_memberships_contains_paid_content":false,"footnotes":"","jetpack_publicize_message":"","jetpack_publicize_feature_enabled":true,"jetpack_social_post_already_shared":true,"jetpack_social_options":{"image_generator_settings":{"template":"highway","default_image_id":0,"font":"","enabled":false},"version":2}},"categories":[11,89,3],"tags":[2159,7,451,1782,14,2063,20,176,513,2182],"class_list":["post-30375","post","type-post","status-publish","format-standard","hentry","category-computer-science","category-nvidia-cuda","category-paper","tag-amd-radeon-instinct-mi300x","tag-ati","tag-benchmarking","tag-computer-science","tag-cuda","tag-hip","tag-nvidia","tag-package","tag-python","tag-triton"],"views":1028,"jetpack_publicize_connections":[],"jetpack_featured_media_url":"","jetpack_sharing_enabled":true,"_links":{"self":[{"href":"https:\/\/hgpu.org\/index.php?rest_route=\/wp\/v2\/posts\/30375","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/hgpu.org\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/hgpu.org\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/hgpu.org\/index.php?rest_route=\/wp\/v2\/users\/351"}],"replies":[{"embeddable":true,"href":"https:\/\/hgpu.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=30375"}],"version-history":[{"count":0,"href":"https:\/\/hgpu.org\/index.php?rest_route=\/wp\/v2\/posts\/30375\/revisions"}],"wp:attachment":[{"href":"https:\/\/hgpu.org\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=30375"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/hgpu.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=30375"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/hgpu.org\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=30375"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}